lighting and human performance ii

Lighting and Human Performance II

Beyond Visibility Models Toward a Unified HumanFactors Approach to Performance

Technical Report

EPRI Project ManagerW. Chow

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

Lighting and Human Performance IIBeyond Visibility Models Toward a Unified HumanFactors Approach to Performance

1006415

Final Report, October 2001

Cosponsors

National Electrical Manufacturers Association1300 North 17th Street, Suite 1847Rosslyn, VA 22209

Principal InvestigatorK. Pitsor

U. S. Environmental Protection AgencyOffice of Air and Radiation6202JWashington, DC 20460

Principal InvestigatorW. VonNeida

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES

THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS ANACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCHINSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THEORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM:

(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I)WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, ORSIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESSFOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON ORINTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUALPROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'SCIRCUMSTANCE; OR

(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER(INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVEHAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOURSELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.

ORGANIZATION(S) THAT PREPARED THIS DOCUMENT

Rensselaer Polytechnic Institute, Lighting Research Center

ORDERING INFORMATION

Requests for copies of this report should be directed to EPRI Customer Fulfillment, 1355 Willow Way,Suite 278, Concord, CA 94520, (800) 313-3774, press 2.

Electric Power Research Institute and EPRI are registered service marks of the Electric PowerResearch Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric PowerResearch Institute, Inc.

Copyright © 2001 Electric Power Research Institute, Inc. All rights reserved.

iii

CITATIONS

This report was prepared by

Rensselaer Polytechnic InstituteLighting Research Center21 Union StreetTroy, New York 12180-3362

Principal InvestigatorP. BoyceM. Rea

This report describes research sponsored by EPRI, National Electrical ManufacturersAssociation, and the Office of Air and Radiation, U. S. Environmental Protection Agency

The report is a corporate document that should be cited in the literature in the following manner:

Lighting and Human Performance II: Beyond Visibility Models Toward a Unified HumanFactors Approach to Performance, EPRI, Palo Alto, CA, National Electrical ManufacturersAssociation, VA, and U. S. Environmental Protection Agency Office of Air and Radiation, DC:2001. 1006415.

v

REPORT SUMMARY

To understand the relationship between lighting conditions and human performance, it is firstnecessary to identify the routes by which lighting conditions can affect human performance.There are three such routes: the visual system, the circadian photobiological system, and theperceptual system. This report updates and replaces an earlier work and explores the relationshipbetween lighting conditions and the ability to carry out tasks in interiors.

BackgroundIn 1989, the report Lighting and Human Performance: A Review was published by the NationalElectrical Manufacturers Association and the Lighting Research Institute. The current workperformed an extensive review of the literature published since 1989. This literature has beenorganized according to the three ways lighting can affect human performance: visibility,circadian photobiology, and psychological effects.

Objective• To summarize research to date concerning the relationship between lighting and humanperformance indoors in photopic conditions.

• To summarize progress in understanding the relationship between lighting and humanperformance since the previous publication in 1989.

• To develop a research agenda by which the impact of lighting conditions on humanperformance indoors in photopic conditions can be more clearly demonstrated and understood.

ApproachThe project team conducted an extensive review of the literature published since 1989. Theliterature was organized according to the three ways lighting can affect human performance:visibility, circadian photobiology, and the psychological “message” delivered by lighting. Inaddition, the role of factors such as age and fatigue, which can reasonably be expected to modifyhuman performance, was considered.

ResultsBased on the literature review, it was concluded that, since 1989, progress has been made inunderstanding lighting’s effect on human performance, at both conceptual and practical levels.The present report is based around a conceptual framework that maps out the routes along whichlighting conditions can be expected to influence human performance. Practical progress has beenmade in twelve different categories, six for direct effects and six for indirect effects. The sixcategories for direct effects are visual performance, task performance, color vision, visual search,age and individual differences, and fatigue. For indirect effects, the six categories are discomfort,

vi

light as an attention stimulus, light and arousal, light and mood, lighting’s influence on behavior,and lighting and hormone balance. From the literature review, a series of research agendas weredeveloped for advancing the study of the effects of lighting on human performance.

EPRI PerspectiveThe extensive and very thorough literature review presented here has resulted in a series ofresearch agenda recommendations that advance the study of lighting’s effects on humanperformance. Improvements in human performance produced by lighting can be expected tohave an impact on increasing productivity, reducing energy consumption, enhancing health, andimproving the quality of life. These four needs become the drivers for the research outlined here.

EPRI wishes to thank the U.S. Environmental Protection Agency, Office of Air and Radiation,and the Lighting Systems Division of the National Electrical Manufacturers Association(NEMA) for their support of this work.

KeywordsLightingHuman productivityVisibilityCircadian photobiology

vii

ABSTRACT

This report updates and replaces an earlier work entitled "Lighting and Human Performance: A Review" that was published by the National Electrical Manufacturers Association and the Lighting Research Institute in 1989. The objectives of this report are to summarize research to date concerning the relationship between lighting and human performance, indoors, in photopic conditions; to summarize progress in the understanding of the relationship between lighting and human performance since the previous publication in 1989; and to develop a research agenda through which the impact of lighting conditions on human performance indoors, in photopic conditions, can be more clearly demonstrated and understood.

To achieve these objectives, an extensive review of the literature published since 1989 has been conducted. The literature is organized according to the three routes whereby lighting can affect human performance; visibility, circadian photobiology and the psychological “message” delivered by the lighting. In addition, the role of factors such as age and fatigue, that can reasonably be expected to modify human performance, is considered. Based on the literature review, it is concluded that, since 1989, progress has been made in the understanding of the effect of lighting on human performance, at both conceptual and practical levels. From the literature review, a series of research agendas have been developed for advancing the study of the effects of lighting on human performance.

ix

ACKNOWLEDGMENTS

Acknowledgements are gratefully given to the following organizations for permission to reproduce the listed figures:

To the Illuminating Engineering Society of North America for Figures 4-4, 4-5, 4-7, 4-8, 4-17, 4-18, 4-19, 4-22 and 5-1.

To the Chartered Institution of Building Services Engineers for Figures 4-20 and 4-21.

To J. Wiley and Sons for Figures 4-6, 5-7 and 5-8.

To Elsevier Science for s 5-2 and 5-5.

To Sage Publications for Figure 5-3.

To Harvard University Press for Figure 5-4.

To Cambridge University Press for Figure 5-6.

To the American Physiological Society for Figure 5-9.

To Oxford University Press for Figures 7-1 and 7-2.

This report on a literature review of Lighting and Human Performance was prepared with the support of the U.S. Environmental Protection Agency, Office of Air and Radiation. Additional support was provided by the Lighting Systems Division of the National Electrical Manufacturers Association (NEMA) for printing the report. EPRI gratefully acknowledges these contributions.

xi

EXECUTIVE SUMMARY

This report updates and replaces an earlier work entitled "Lighting and Human Performance: A Review" that was published by the National Electrical Manufacturers Association and the Lighting Research Institute in 1989. Like the earlier publication, this report explores the relationship between lighting conditions and the ability to carry out tasks in interiors. The objectives of this report are to:

• Summarize research to date concerning the relationship between lighting and human performance, indoors, in photopic conditions

• Summarize progress in the understanding of the relationship between lighting and human performance since the previous publication in 1989

• Develop a research agenda through which the impact of lighting conditions on human performance indoors, in photopic conditions, can be more clearly demonstrated and understood.

To achieve the first objective, an extensive review of the literature published since 1989 was conducted. The literature was classified according to the three routes whereby lighting can affect human performance; visibility, circadian photobiology and the “message” delivered by the lighting. Also considered were modifying factors, i.e., other factors that can reasonably be expected to cause a decline in human performance.

Visibility issues cover the effects of the amount of light, the light spectrum and its distribution on task performance, as well as what is known about visual search and the validity of different approaches to modeling visual performance.

Circadian photobiology issues are concerned with the rapidly growing understanding of how light acts through the eye to determine the “platform” from which we operate for all types of task, both visual and non-visual.

“Message” issues involve the effects of lighting conditions on mood and behavior.

Modifying factors cover the effects of aging on the visual system and how light can be used to provide some compensation; the impact of prolonged work on human performance, and the effect of some medical conditions that can be treated with light on human performance.

As a result of the literature review, it is concluded that:

• Task visibility is the best understood of the routes whereby lighting conditions can affect human performance. A validated, quantitative model of visual performance exists and can be

xii

used to make predictions of the effect of changing lighting conditions on visual performance for tasks seen on-axis. What is required to make it more valuable is a system of task classification that identifies the importance of visual performance for task performance for any task. How lighting conditions influence the performance of tasks requiring off-axis vision is a subject deserving of study.

• Knowledge and understanding of how light exposure changes the human circadian system has increased dramatically over the last 20 years, although the photoreceptors involved have not yet been identified. There is no doubt that exposure to light at the right time can shift the phase of the human circadian rhythm over the succeeding twenty-four hours and can have an immediate effect on alertness. The right time is when the hormone melatonin is being produced. These circadian photobiological effects of exposure to light are important because they have the potential to affect the performance of all types of tasks, not just visual tasks. They are also important because the illuminances typical of current lighting practice in interiors are on the borderline of being effective for circadian photobiology. Research is needed in this area to determine what lighting conditions change circadian photobiology and how changes in circadian photobiology affect human performance.

• The literature in the area of mood and behavior is dominated by the variability found in response. The aspects of lighting that can cause discomfort are known, but the same factors can all be used to positive effect in the right context. Similarly, there is no doubt that different lighting conditions can produce different perceptions of objects and spaces but whether that has an effect on mood and behavior depends on the context and culture within which those perceptions occur. This uncertain response occurs because mood and behavior are determined by many factors; lighting conditions are only a small set of those factors. The problem for research in this area is not of determining whether lighting conditions can alter mood and behavior but rather in what situations mood and behavior are most sensitive to lighting conditions.

• There can be little doubt that the deterioration of visual capabilities with age or eyestrain and the deterioration of cognitive abilities through fatigue or ill-health will ultimately diminish task performance. The problems for any research agenda are, first, to determine whether the deterioration of visual capabilities with age or cognitive capabilities with fatigue are likely to occur under realistic working conditions; and, second, to identify the tasks that are most sensitive to these declines.

Based on this literature review, progress since 1989 can be considered at two levels, the conceptual and the practical. The previous document (Boyce et al., 1989) was organized around the photometric characteristics of lighting, namely, illuminance, light distribution and light spectra. The consequences of changing these dimensions for human performance were considered in terms of direct and indirect effects. Direct effects are those that operate either by changing the stimulus to the visual system or by changing the operating state of the visual system. Indirect effects are changes in human performance that occur because of changes in attention, arousal, mood or hormone balance. The conceptual framework used in the previous report was primitive, at best, and for the indirect effects was non-existent. The present report is based around a conceptual framework setting out the routes whereby lighting conditions can be expected to influence human performance. This is progress because conceptual frameworks are important for research. They form the unstated assumptions within which research is conceived.

xiii

As for practical progress, the previous publication summarized the state of knowledge in 1989 using a matrix in which research areas were divided up into twelve different categories, six for direct effects and six for indirect effects. The following two tables provide thumbnail sketches of the progress that has been made since 1989 in these twelve categories. Clearly considerable progress has been made in some areas and little or none in others.

Table ES-1 Lighting and Human Performance Matrix - Direct Effects

1989 Category Progress since 1989

Visual performance

Two quantitative models of visual performance have been developed. One has been independently validated. These models cover different sizes and luminance contrasts of the target and different illuminances. They do not consider color difference or blur.

Task performance

Quantitative models predicting the effect of lighting on the performance of specific tasks have been developed but no general model exists.

Color vision

For achromatic, near-threshold tasks, scotopically-enriched light spectra reduce pupil size and improve task performance. What happens for realistic suprathreshold tasks remains to be determined. No progress has been made in quantifying the effect of lighting on the performance of chromatic tasks.

Visual search No progress

Age and individual differences

Guidelines for lighting based on known changes in ocular physiology with age have been developed and shown to lead to better performance of tasks of everyday living.

Fatigue

Prolonged work in inappropriate lighting conditions can cause fatigue. The effect this has on task performance depends on the nature of the task and the freedom the worker has to modify how the task is done.

xiv

Table ES-2 Lighting and Human Performance Matrix - Indirect Effects

1989 Category Progress since 1989

Discomfort

Lighting conditions that cause discomfort and alter the stimulus the task presents to the visual system will change visual performance. Lighting conditions that simply cause discomfort without affecting the stimulus presented by the task, may or may not affect task performance.

Light as an attention stimulus

Lighting can be used to attract attention, but no progress has been made on quantifying the conditions necessary.

Light and arousal

Light exposure can increase arousal, particularly at night, when exposure to light suppresses the hormone melatonin.

Light and mood

Lighting can influence mood but so can many other factors. Changes in mood have been shown to affect task performance.

Lighting's influence on behavior

Lighting can influence behavior, either directly by attracting attention or providing necessary visual conditions, or by sending a message as to what is the appropriate behavior.

Lighting and (hormone) biology

Understanding of the circadian photobiology system has grown greatly but much remains to be determined, such as the spectral sensitivity of the system. Exposure to light at night can have a short-term arousing effect and a longer term phase-shifting effect. To ensure a phase-shifting effect, control of light exposure over 24 hours is necessary. Exposure to light at night can influence performance of some tasks, but why some tasks are sensitive and others are not remains to be determined.

Based on the literature review and the conclusions reached from it, a series of research agendas have been developed for the effects of lighting on human performance, operating through visibility, circadian photobiology, and "message", as well as for the modifying factors. The rationale behind these research agendas is that lighting is a means to an end, not an end in itself. For lighting research agendas to attract widespread support they should address the ends that are widely considered important. Four such ends, where improvements in human performance produced by lighting can be expected to have an impact, are identified. They are, increasing productivity, reducing energy consumption, enhancing health and improving the quality of life. These are the "drivers" of research.

These "drivers" operate on a three-dimensional matrix that identifies what aspects of a lighting installation can have an effect on human performance, in different application areas, through different routes. The attached figure shows an exploded version of the matrix. The heavily-shaded areas indicate high priority areas of study, i.e., conjunctions of lighting dimensions, application areas and lighting effects where little is known but there is reason to expect significant effects of lighting on human performance. The lightly-shaded areas indicate moderate priority areas of study, i.e., conjunctions where the effects of lighting on human performance are likely to be significant but only under a limited range of conditions. The hatched areas indicate conjunctions where the effects of lighting on human performance are well-established and there

xv

is little need for further research. The unshaded areas indicate the conjunctions where either there is believed to be little effect of lighting conditions on human performance or the particular conjunction is unlikely to occur in practice.

From considerations of this type, sixteen different topics have been identified as worthy of study. They range from the fundamental to the applied. All are important if lighting is to be used to enhance human performance most effectively. All can be related to one of more of the "drivers" of research. They are:

• Develop a task classification system that enables tasks with an important visual component to be identified.

• Investigate the effect of lighting conditions on the performance of tasks requiring different amounts of off-axis vision.

• Determine the spectral sensitivity of the human circadian system and then develop a suitable system of photometry.

• Test the models of the effects on the circadian system of combining the amount, spectrum, timing and duration of exposure to light.

• Investigate the effect of light exposure on alertness and task performance at the start and end of the normal working day.

• Examine the interaction between cognitive task complexity and the duration of performance at night before deterioration of performance occurs. Determine if this duration can be extended by the use of light exposure at night.

• Determine if an increase in alertness at night produced by exposure to light, is followed by greater fatigue when the light is removed.

• Monitor developments in the understanding of health impacts of light exposure at night.

• Establish a lexicon of lighting by determining what "messages" different lighting conditions deliver in a given context.

• Determine how sensitive mood is to lighting conditions, relative to architecture and decor, in realistic situations, and whether mood affects task performance consistently.

• Establish the lighting conditions necessary to draw people to a retail display and if that attraction diminishes over repeated exposures.

• Determine how to most easily ameliorate the effects of aging in the visual system and the consequences for task performance during the day.

• Determine if exposure to light can be used to ameliorate the effect of aging on the circadian system and the consequences for task performance during the day.

• Determine the effect of lighting conditions on the prolonged performance of skilled work.

• Determine whether light exposure improves the quality of work of people experiencing SAD and sub-SAD during winter.

• Using a sample of people who experience advanced and delayed phase sleep disorder, examine the effect of light exposure on absenteeism and quality of work done.

Finally, a brief review of the various research methods available to those studying the impact of lighting conditions on human performance is given.

xvii

CONTENTS

1 SCOPE ................................................................................................................................ 1-1

2 THE ROUTES TO HUMAN PERFORMANCE ..................................................................... 2-1

3 IMPORTANT ASPECTS OF LIGHTING .............................................................................. 3-1

4 VISIBILITY ISSUES: CURRENT KNOWLEDGE ................................................................. 4-1

4.1 Visibility in Context ....................................................................................................... 4-1

4.2 Illuminance and Task Performance .............................................................................. 4-2

4.3 Light Spectrum and Task Performance......................................................................... 4-5

4.3.1 Achromatic Tasks ................................................................................................. 4-5

4.3.2 Chromatic Tasks................................................................................................... 4-8

4.4 Light Distribution and Task Performance .....................................................................4-10

4.5 Polarization and Task Performance.............................................................................4-12

4.6 Visual Search ..............................................................................................................4-14

4.7 Lighting and Visual Performance.................................................................................4-21

4.7.1 The Analytical Approach ......................................................................................4-21

4.7.2 The Visibility Approach.........................................................................................4-24

4.7.3 The Stimulus-Response Approach.......................................................................4-25

4.7.3.1 The Relative Visual Performance (RVP) model ............................................4-25

4.7.3.2 The Visual Performance (VP) Model ............................................................4-36

4.7.4 Comparisons and Limitations...............................................................................4-36

5 CIRCADIAN PHOTOBIOLOGY: CURRENT KNOWLEDGE ............................................... 5-1

5.1 Circadian Photobiology in Context................................................................................ 5-1

5.2 The Structure of the Human Circadian System............................................................. 5-1

5.2.1 The Retina ............................................................................................................ 5-2

5.2.2 The Suprachiasmatic Nuclei ................................................................................. 5-3

5.2.3 The Pineal Gland .................................................................................................. 5-3

xviii

5.3 Characteristics of the Human Circadian System........................................................... 5-4

5.4 Effects of Light Exposure on the Human Circadian System.......................................... 5-6

5.5 Factors Determining the Effectiveness of Light Exposure............................................. 5-8

5.6 The Consequences of Trying to Work in Circadian Night.............................................5-10

6 “MESSAGE” ISSUES: CURRENT KNOWLEDGE.............................................................. 6-1

6.1 “Message” in Context ................................................................................................... 6-1

6.2 Visual Discomfort and Task Performance..................................................................... 6-1

6.3 Perception of Spaces and Task Performance............................................................... 6-5

6.4 Lighting and Mood........................................................................................................ 6-8

6.6 Lighting and Behavior................................................................................................... 6-9

7 MODIFYING FACTORS....................................................................................................... 7-1

7.1 Aging Vision ................................................................................................................. 7-1

7.2 Defective Color Vision .................................................................................................. 7-5

7.3 Prolonged Work............................................................................................................ 7-7

7.3.1 Eyestrain............................................................................................................... 7-7

7.3.2 Fatigue.................................................................................................................. 7-8

7.3.3 Mood Changes ....................................................................................................7-10

7.4 Lighting and Health......................................................................................................7-11

7.4.1 Sleep Disorders ...................................................................................................7-12

7.4.2 Seasonally Affective Disorder (SAD)....................................................................7-12

7.4.3 Migraines .............................................................................................................7-13

8 PROGRESS SINCE 1989.................................................................................................... 8-1

8.1 Background .................................................................................................................. 8-1

8.2 Progress in Concepts ................................................................................................... 8-1

8.3 Practical Progress ........................................................................................................ 8-1

9 RESEARCH AGENDA ........................................................................................................ 9-1

9.1 "Drivers" of Research ................................................................................................... 9-1

9.2 The Lighting Matrix ....................................................................................................... 9-2

9.3 Visibility: Research Agenda .......................................................................................... 9-6

9.3.1 Overview............................................................................................................... 9-6

9.3.2 Research Agenda ................................................................................................. 9-6

9.3.3 Justification ........................................................................................................... 9-6

xix

9.4 Circadian Photobiology: Research Agenda .................................................................. 9-6

9.4.1 Overview............................................................................................................... 9-6

9.4.2 Research Agenda: ................................................................................................ 9-7

9.4.3 Justification ........................................................................................................... 9-7

9.5 "Message": Research Agenda...................................................................................... 9-8

9.5.1 Overview............................................................................................................... 9-8

9.5.2 Research Agenda ................................................................................................. 9-8

9.5.3 Justification ........................................................................................................... 9-8

9.6 Modifying Factors: Research Agenda........................................................................... 9-9

9.6.1 Overview............................................................................................................... 9-9

9.6.2 Research Agenda ................................................................................................. 9-9

9.6.3 Justification ........................................................................................................... 9-9

10 RESEARCH METHODS ...................................................................................................10-1

11 REFERENCES .................................................................................................................11-1

xxi

LIST OF FIGURES

Figure 2-1 A conceptual framework setting out the three routes whereby lighting conditions can influence human performance. The arrows in the diagram indicate the direction of effect. ...................................................................................................... 2-2

Figure 4-1 Output of silk weavers over a fifteen week period compared with the amount of daylight typically available during those weeks (After Elton, 1920). ............................. 4-2

Figure 4-2 Mean monthly performance for cutting leather shapes for the years 1957/59 and 1959/61. Performance is normalized for the average performance over the years 1957/59 (After Stenzel, 1962). ............................................................................... 4-3

Figure 4-3 Mean performance score for proof reading good and poor quality text, plotted against illuminances for young (18 - 22 years) and older (49 - 62 years) subjects. The performance score was calculated from the accuracy of proof reading with a bonus for speed. The maximum possible performance score was 35 (After Smith and Rea, 1978)................................................................................................................ 4-5

Figure 4-4 Means and associated standard errors of the proportion of Landolt Ring orientations, presented for 200 ms on a spectrally neutral background, that were correctly identified, plotted against luminance contrast, for four different background luminances: a = 11.9, b = 27.7, C = 47.0, d = 73.4 cd/m2. In all four diagrams, the upper curve is for the scotopically-enriched illuminant (surround field scotopic luminance = 228 cd/m2) and the lower curve is for the scotopically diminished illuminant (surround field scotopic luminance = 13 cd/m2). Both illuminants produced a surround field photopic luminance of 53 cd/m2 (After Berman et al., 1993). .................. 4-7

Figure 4-5 Recognition performance, measured as the proportion of times the orientation of the word EXIT was correctly recognized, plotted against luminance contrast for color-normal observers. The legend is read as “letter color on background color”. The light sources used are LEDs. The “Green” LED has a peak emission at 530 nm. The “Red L” LED has a peak emission at 660 nm (After Eklund, 1999). ................... 4-8

Figure 4-6 The MacAdam ellipses plotted on the CIE 1931 chromaticity diagram. The boundary of each ellipse represents ten times the standard deviation of the color matches made for the indicated chromaticity (After Wyszecki and Stiles, 1982). ............. 4-9

Figure 4-7 The performance score on the numerical verification task plotted against luminance contrast. The different levels of luminance contrast were obtained by different ink pigment densities, and viewing from different directions, using light of different percentage polarization to modify veiling reflections (After Rea, 1981). ............4-11

Figure 4-8 The combined effects of percentage polarization, viewing orientation, ink pigment density and ink specularity on the performance score on the numerical verification task. The 0° viewing orientation allows veiling reflections to occur. The 90° viewing orientation does not. The percentage polarization at the angle from the luminaire to the task for each luminaire type is PM = 8%, MP = 19%, and LP = 99% (After Rea, 1981)............................................................................................................4-14

xxii

Figure 4-9 The pattern of eye fixations made by two inspectors examining men’s briefs held on a frame. S = start of scan path, C = end of scan of front and one side, rotation of frame and continuation of scan across back and sides, E= end of search. Inspector M examines only seams while inspector D examines fabric as well (After Megaw and Richardson, 1979).......................................................................................4-16

Figure 4-10 Mean of the median search time for detecting a single flaw in a glass sheet, plotted against flaw size (After Drury, 1975). ..................................................................4-17

Figure 4-11 The probability of detection of targets of (a) contrast = 0.058, size = 19 min arc; (b) contrast = 0.080, size = 10 min arc; (c) contrast = 0.044, size = 10 min arc; within a single fixation pause, plotted against deviation from the visual axis. Each curve can be used to form a visual detection lobe for each target by assuming radial symmetry about the visual axis.......................................................................................4-17

Figure 4-12 Mean search times for two observers searching for rectangular or square targets among an array of non-square targets, plotted against an index of discriminability. The discriminability index is given by (A0.5 - B0.5)2 where A and B are the areas of the targets and non-targets, respectively (After Bloomfield, 1975). Table 4-1 Mean time to find a target for different target specifications ............................4-18

Figure 4-13 Mean search times for locating a specified integer number from a random array of one hundred such numbers, plotted against illuminance (After Muck and Bodmann, 1961).............................................................................................................4-20

Figure 4-14 A Landolt ring chart. All the rings in a chart are of the same size and luminance contrast but different charts can have rings of various sizes and luminance contrasts. ......................................................................................................4-22

Figure 4-15 Mean performance scores for Landolt ring charts of different critical size and contrast plotted against illuminance (After Weston, 1945). .............................................4-23

Figure 4-16 The numerical verification task. The reference list is on the left and the response list is on the right. ............................................................................................4-26

Figure 4-17 The mean times taken, number of misses and number of false positives for the numerical verification task, plotted against luminance contrast (After Rea, 1981)..............................................................................................................................4-27

Figure 4-18 The mean reciprocal of time taken to perform the numerical verification task plotted against luminance contrast at four background luminances (After Rea, 1986)..............................................................................................................................4-28

Figure 4-19 The Relative Visual Performance (RVP) model of visual performance, based on the time taken to read the reference page of the numerical verification task (After Rea, 1986). ....................................................................................................................4-29

Figure 4-20 The difference in mean reaction time to targets of four different visual sizes (2, 14, 130 and 2,800 microsteradians) plotted against luminance contrast. For each figure the retinal illumination decreases from 801, 160, 31, 6.3, 1.6 to 0.63 trolands, from left to right (After Rea and Ouellette, 1988)..............................................4-30

Figure 4-21 The RVP model of visual performance based on reaction time data. Each element shows the RVP value plotted against luminance contrast and retinal illuminance, for a fixed target size (After Rea and Ouellette, 1991).................................4-31

Figure 4-22 Mean reading speed measured in words / second plotted against letter size in points, for different luminance contrasts (After Bailey et al., 1993)..............................4-33

xxiii

Figure 4-23 Mean work speeds for the data entry task for four print sizes. Each graph shows the mean work speed plotted against illuminance (lx) for four luminance contrasts. The error bars indicate 95% confidence intervals. Note that the plot for the 6 point print contains data for only three luminance contrasts. 6-point print of luminance contrast 0.10 could not be read by a significant number of subjects at any of the illuminances used (After Eklund et al., 2000). ................................................4-34

Figure 4-24 Normalized RVP values predicted by the RVP model and the VP model for the data-entry task plotted against the normalized measured mean work speeds. Data are also shown for the empirical data-entry task performance (ANN - DETP) model. This model was developed from the data-entry task performance data so it should provide a good fit (After Eklund et al., 2000). ......................................................4-35

Figure 4-25 Mean percentage correct identifications of three letter words presented for 100 ms in the fovea and at 1° eccentricity, at different levels of luminance contrast. The error bars are the standard error of the mean (After Timmers, 1978).......................4-38

Figure 4-26 Mean work speeds predicted by the data-entry task performance (ANN-DETP) model plotted against measured mean work speed. (After Eklund et al., 2000)..............................................................................................................................4-40

Figure 5-1 A simplified illustration of the retino - hypothalamic - pineal axis, i.e., from the retina to the suprachiasmatic nuclei (SCN) to the paraventricular nuclei (PVN) to the superior cervical ganglion and thence to the pineal gland................................................ 5-2

Figure 5-2 Melatonin concentration before, during, and after exposure to different levels of illuminance at night (After McIntyre et al., 1989a). ....................................................... 5-4

Figure 5-3 Mean levels of plasma melatonin in constant dim light in six individuals after exposure to defined fourteen-hour dark periods (hatched area) under controlled conditions (faint line) and after they were exposed to natural fourteen-hour winter nights (same hatched area, bold line). The individuals' use of electric lighting, after dark, appears to have shifted their melatonin rhythm several hours later than would be expected relative to solar night (After Wehr, 1997). .................................................... 5-5

Figure 5-4 A schematic illustration of the effect of light exposure on the phase of the circadian rhythm. Depending on the timing of the light pulse, the circadian rhythm can be advanced, delayed or left unchanged (After Moore-Ede et al., 1982)................... 5-7

Figure 5-5 Modulation of core body temperature by alternating blocks of bright and dim light from midnight to 9 a.m. Open circles or triangles represent the bright light conditions. Filled circles or triangles represent the dim light condition. Circles started with 90 minutes of bright light at midnight, Triangles started with 90 minutes of dim light at midnight (After Badia et al., 1991). ............................................................ 5-8

Figure 5-6 (A) The phase shift in the melatonin rhythm following exposure to 6.5 hours of light at different illuminances. The illuminances are measured at the cornea. (B) The suppression of melatonin during the light exposure. A four parameter logistic model has been fitted through the data. The model is represented by the continuous line and the 95% confidence intervals by the dotted lines (After Zeitzer et al., 2000). .................................................................................................................... 5-9

Figure 5-7 Speed and accuracy data collected from workers performing real world tasks at various times during the twenty-four hours (After Folkard and Monk, 1979). ..............5-11

Figure 5-8 Phase difference between the maximum core body temperature and the ideal time for that maximum to occur, for 21 continuous night shifts (After Monk and Folkard, 1983). ...............................................................................................................5-12

xxiv

Figure 5-9 The time courses of core body temperature and plasma melatonin concentration over 32 hours. The time courses of measures of sleepiness, i.e., mean eye blinks per 30 seconds, slow eye movements, stage 1 sleep and sleepiness rating, over 32 hours. Time courses of task performance for tasks requiring vigilance, mental arithmetic (cognitive throughput) and short term memory over 32 hours. The dotted line represents the subjects’ habitual bedtime. The error bars are standard errors of the means (After Cajochen et al. 1999)................................5-16

Figure 6-1 Mean detection speeds for locating a specified number among others at different illuminances and the percentage of subjects doing the task who considered the lighting "good" at each illuminance (After Muck and Bodmann, 1961)............................................................................................................................... 6-4

Figure 7-1 The contrast sensitivity functions for four observers of different ages. Open circles are for near viewing distance. Filled circles are for far viewing distance (After McGrath and Morrison, 1981).......................................................................................... 7-3

Figure 7-2 The average distribution of errors on the Farnsworth-Munsell 100 Hue test as a function of illuminance and age (After Knoblauch et al., 1987) ..................................... 7-4

Figure 9-1 The lighting matrix showing what aspects of a lighting installation can have an effect on human performance, in different application areas, through different routes. What parts of the matrix are chosen for study is influenced by the "drivers" of research. ..................................................................................................................... 9-3

Figure 9-2 An exploded version of the lighting matrix. ............................................................. 9-5

xxv

LIST OF TABLES

Table 4-1 Mean time to find a target for different target specifications....................................4-18

1-1

1 SCOPE

This report updates and replaces an earlier work entitled "Lighting and Human Performance: A Review" that was published by the National Electrical Manufacturers Association and the Lighting Research Institute in 1989 (Boyce et al., 1989). Like this earlier publication, this report explores the relationship between lighting conditions and the ability to carry out tasks in interiors. “Tasks” are activities that produce tangible outputs, such as the manufacture of objects; activities that lead to desirable outcomes, such as a better quality product; and activities commensurate with behaviors leading to desirable outcomes, such as increased sales of merchandise. “Interiors” implies that the lighting conditions are such that the visual system is always operating in the photopic state, rather than the mesopic and scotopic states commonly found outdoors, after dark. In the photopic state, the human visual system is capable of fine discrimination of detail and color.

The objectives of this report are to:

• Summarize research to date concerning the relationship between lighting and human performance, indoors, in photopic conditions

• Summarize progress in the understanding of the relationship between lighting and human performance since the previous publication in 1989

• Develop a research agenda through which the impact of lighting conditions on human performance indoors, in photopic conditions, can be more clearly demonstrated and understood.

2-1

2 THE ROUTES TO HUMAN PERFORMANCE

To understand the relationship between lighting conditions and human performance it is first necessary to identify the routes by which lighting conditions can affect human performance. There are three such routes: through the visual system, through the circadian photobiological system and through the perceptual system. Figure 2-1 shows a conceptual framework for considering the factors that influence progress down each route and the interactions between them.

The effect of lighting on vision is the most obvious impact of light on humans. With light we can see; without light we cannot. The visual system is an image-processing system. The optics of the eye form an image of the outside world on the retina. At the retina, some image processing occurs to overcome the limitations of the eye’s optics. Different aspects of the image are then processed through two different channels up to the visual cortex of the brain. The magnocellular channel processes information rapidly, but with little detail or color information, while the parvocellular channel provides detail of brightness, color, and texture, but at a slower rate (Sekular and Blake, 1994). Further, the visual system is organized spatially into two parts: the fovea of the retina, where fine detail is available, and the periphery, which is basically a detection system indicating where in the visual field the fovea should be directed. It has been estimated that about 80% of the cells in the visual cortex are devoted to the central 10 degrees of the visual field, which includes the fovea (Drasdo, 1977). The visual system is continuously adapting to the available amount of light. When a great deal of light is available, e.g., in daytime, the whole of the retina is active. In these conditions, called photopic vision, fine detail can be resolved and color can be seen. When there is very little light, e.g., outside on a moonless night, the fovea is blind and only the peripheral retina operates. In these conditions, called scotopic vision, neither fine detail nor color can be seen. There is an intermediate stage, called mesopic vision, which occurs outdoors around dawn or dusk, in which both detail and color can be seen but the level of discrimination is diminished relative to photopic vision.

The visual system is devoted to detecting differences in the visual environment. These differences can occur in both brightness and color, and can vary over both time and space. It is the interaction between the object to be seen, the background against which it is seen and the lighting of both object and background that determine the stimuli for the visual system and it’s operating state. The aspects of lighting likely to be important for visual performance are the amount of light, the spectrum of the light and the distribution of the light on and around the object.

The Routes to Human Performance

2-2

Figure 2-1 A conceptual framework setting out the three routes whereby lighting conditions can influence human performance. The arrows in the diagram indicate the direction of effect.


2-3

It should be noted that visual performance and task performance are not necessarily the same. Task performance is the performance of the complete task. Visual performance is the performance of the visual component of the task. Task performance is needed in order to measure productivity and to establish cost / benefit ratios for determining the effectiveness of providing a lighting installation. However, visual performance is the only component of task performance that can be directly affected by changing the lighting conditions.

Most visual tasks have three components; visual, cognitive and motor. The visual component refers to the process of extracting information relevant to the performance of the task using the sense of sight. The cognitive component is the process by which sensory stimuli are interpreted and the appropriate action determined. The motor component is the process by which the stimuli are manipulated to extract information and/or the appropriate action undertaken. Of course, these three components interact to produce a complex pattern between stimulus and response. Further, every task is unique in its balance between visual, cognitive and motor components and hence in the effect lighting conditions have on task performance. Task uniqueness makes it impossible to generalize from the effect of lighting on the performance of one task to the effect of lighting on the performance of another. The effect of lighting on the performance of a specific task depends on the structure of the task and specifically the importance of the visual component relative to the cognitive and motor components. Tasks in which the visual component is large will be more sensitive to changes in lighting conditions than tasks where the visual component is small. For example, proof reading is likely to be much more sensitive to lighting changes than reading for comprehension (Smith and Rea, 1978, 1982).

Another means whereby lighting conditions can affect work is through the circadian photobiological system. The most obvious evidence for a circadian system in humans is the existence of the sleep / wake cycle, but this cycle represents only a small part of the activity of the circadian system. Variations in many different hormonal rhythms occur over a 24-hour period. The organ that controls these cycles in humans is the suprachiasmatic nucleus (SCN). The SCN is linked directly to the retina (Roenneberg and Foster, 1997). When signals are transmitted from the retina to the SCN, no attempt is made to preserve their original retinal location. Rather, the parts of the retina supplying the SCN act like a simple photocell. This means that the aspects of lighting that influence the state of the SCN are the light spectrum and illumination reaching the retina, which in turn depend on the light spectrum of the light source used, the light distribution, the spectral reflectances of the surfaces in the space, the spectral transmittance of the optic media and where the observer is looking. Light operating through the circadian system can alter human performance by changing the “platform” from which the rest of the body functions. This means that light can influence human performance in all its aspects; cognitive as well as visual.

Our knowledge of how lighting conditions affect the human circadian system has grown rapidly in recent years, although much remains to be learned. Exposure to light has two distinct effects: a phase-shifting effect which is used in the entrainment of the endogenous circadian rhythm and in adaptation to sudden changes in the required sleep / wake cycle (Dijk et al.,1995); and an acute effect related to the suppression of the hormone melatonin at night (Campbell et al., 1995). Both these effects can be expected to affect human performance.


2-4

The third means whereby lighting conditions can affect work is through the perceptual system. The perceptual system takes over once the retinal image has been processed by the visual system. The simplest output of the perceptual system is a sense of visual discomfort, which may change the observer’s mood and motivation, particularly if the work is prolonged. The simplest response indicating visual discomfort is an aversion to light itself, commonly known as photophobia. Such a response is rare in interiors. When it does occur, it is usually associated with either a medical state in the observer, such as a migraine attack, or an extreme lighting condition. More common is discomfort associated with the ability to extract information from the visual environment. Lighting conditions in which it is difficult to achieve a high level of visual performance will be considered uncomfortable, as will conditions in which the lighting leads to distraction from the task, which can occur when glare and flicker are present. But perception is much more sophisticated than just producing a feeling of visual discomfort. In a sense, every lighting installation sends a “message” about the people who designed it, who bought it, who work under it, who maintain it, and about the place where it is located. Observers interpret the “message” according to the context in which it occurs and their own culture and expectations. The importance of this “message” is sometimes enough to override conditions that might be expected to cause discomfort. For example, lighting conditions that would be considered extremely uncomfortable in an office are positively desired in a dance club. According to what the intended or perceived “message” is, the observer’s mood and motivation can be changed.

While each of these routes will be discussed separately, it is important to appreciate that they can also interact. For example someone who is asked to work while sleep-deprived will be fatigued. This fatigue will affect task performance through both its cognitive and visual components. Conversely, people who are performing a task that is visually difficult for a long time will experience fatigue, even if they are not sleep-deprived. Another example would be a situation where the lighting provides poor task visibility, so that visual performance is poor. If the worker is aware of the poor level of performance and it fails to meet his or her expectations, then the worker’s mood may be altered. Multiple interactions of this type can occur. To further complicate the picture, it is necessary to appreciate that while visual performance for a given task is determined by the lighting conditions alone, a worker’s motivation can be influenced by many different factors apart from lighting conditions. As for the circadian system, this can be influenced by such factors as the timing of exercise and social cues as well as light exposure (Van Reeth et al. 1994). It is this complex pattern of interacting effects that has made the study of the relationship between lighting and human performance so prolonged and difficult.

3-1

3 IMPORTANT ASPECTS OF LIGHTING

When developing a lighting design, the designer has five aspects of lighting to manipulate: the amount, distribution, and spectrum of the light, the variation of these aspects over time, and the appearance of the luminaire. Taken together, the amount and distribution of light in a given interior at a given time determine the luminance pattern received at the retina when looking in different directions. The spectrum of the light determines the effectiveness of this retinal illumination in stimulating the retina. Knowing the amount, distribution, and spectrum of the illumination in the interior and how it varies over time is a necessary condition for predicting human performance, but it is not a sufficient condition. The other necessary information is the properties of the reflecting surfaces in the interior. These surfaces and the way they are illuminated define the visibility of a task, the stimulation provided to the circadian photobiological system, and, to a large extent, the perception of the interior. This report is concerned with the effects of lighting conditions on human performance, so emphasis will be given to the effects of the amount, distribution, and spectrum of light, but the spatial distribution and spectral reflectances of surfaces are also important.

4-1

4 VISIBILITY ISSUES: CURRENT KNOWLEDGE

4.1 Visibility in Context

The physiology, operation, and capabilities of the human visual system have been the subject of study for more than a century. One outcome of these studies is a well-developed understanding of how lighting conditions may alter the performance of tasks requiring input from the visual system.

The human visual system is an image-processing system that is designed to give a stable perception of the visual environment in changing conditions. Like all image processing systems, it works best with a well-defined image, i.e., an image in which the signal-to-noise ratio is high. To ensure a high signal-to-noise ratio and hence a high level of human performance, it is desirable that the stimuli should be large in size, of high contrast in both luminance and color, precisely focussed on the retina, and available for a long time in a known position. Steadily degrading any of these aspects of the stimuli, i.e., reducing the size, decreasing the contrast, blurring the retinal image, reducing the time for which the task can be seen, or putting the task in motion, can degrade performance until, eventually, conditions are reached where the task cannot be performed at all. Stimuli that make it almost impossible to perform the task are called threshold conditions. Stimuli where all the task details are easily seen are called suprathreshold conditions. In threshold conditions, all the different aspects of the stimulus presented by the task matter, small changes in any aspect of the stimulus are important, and many aspects will interact to determine the level of task performance possible. In suprathreshold conditions, task performance is relatively insensitive to changes in the stimuli provided by the task. The transition from threshold to suprathreshold conditions is not linear. Rather, it follows a compressive function, where small changes in stimuli make large differences to performance close to threshold, but the same changes make very little difference in suprathreshold conditions. In the most basic sense, the purpose of lighting designed to enhance the performance of visual tasks may be characterized as ensuring the task is in suprathreshold conditions, well away from threshold conditions.

Lighting can affect the performance of visual tasks in two ways. First, lighting can change the stimulus presented by the task by modifying the luminance and color contrasts of the task. Second, lighting can change the operating state of the visual system. This latter effect is achieved by increasing the retinal illuminance. As the retinal illuminance increases, the speed with which information is transmitted along neural fibers is increased, which therefore increases the speed with which the visual system handles information. Also, as the adaptation luminance increases the detail that can be resolved by the visual system becomes finer. These changes in the operating state of the visual system are the reason that task performance continues to improve somewhat with increasing illuminance even when the stimulus is in the suprathreshold condition.

Visibility Issues: Current Knowledge

4-2

4.2 Illuminance and Task Performance

The most direct approach to determining the effects of illuminance on the performance of a specific task is to carry out a field study. This approach has a long history. Among the earliest such studies were measurements of such difficult visual tasks as silk weaving (Elton, 1920), linen weaving (Weston, 1922) and typesetting by hand (Weston and Taylor, 1926). Figure 4-1 shows silk weavers’ output over fifteen weeks and the amount of daylight outdoors during those weeks, in an average year. There is a clear relationship between the output and the amount of daylight available. Such observations confirm what is common experience: performance declines when lighting conditions become inadequate for workers to clearly see the details necessary to perform a task. This may seem a statement of the obvious, but it has been the subject of a surprising amount of argument.

Figure 4-1 Output of silk weavers over a fifteen week period compared with the amount of daylight typically available during those weeks (After Elton, 1920).

At about the same time as these early studies were being done, another series of experiments was starting—the Hawthorne experiments (Snow, 1927; Roethlisberger and Dickson, 1939; Landsberger, 1958; Urwick and Brech, 1965; Parsons, 1974). The Hawthorne plant of the Western Electric Company in Chicago manufactured telephone apparatus. At the start of the studies, the company conducted three experiments on the effects of lighting on the output of a group of women who inspected parts, assembled relays, or wound coils. In the first experiment, the illuminances on the tasks were varied in a series of steps, both up and down. Output in all three departments changed, but showed no clear relationship to the illuminance. The second experiment used only the coil-winding department. The workers were split into two groups having the same level of experience. The control group was exposed to a relatively stable illuminance of 170 - 300 lx, while the test group was exposed to illuminances in the range 260 to 750 lx. The work output of both groups increased to a similar extent. The first and second experiments involved both electric lighting and daylight, the presence of daylight explaining some of the variability in illuminance. In the third experiment, daylight was eliminated. The


4-3

same control and test groups that were used in the second experiment were used in the third experiment, but this time the control group worked under a constant illuminance of 110 lx, while the test group experienced illuminances starting at 110 lx and decreasing in steps of 11 lx. After the initial illuminance was set, both groups showed a slow but steady improvement in output. However, when the illuminance experienced by the test group reached 33 lx, the members of the test group protested that they could hardly see what they were doing and their output dropped.

From these studies, the experimenters concluded that lighting was only one factor, and apparently a minor factor, among the many that affect worker output, and that there was a whole area of what they called “human relations” waiting to be explored. Sometimes this conclusion is distorted into a claim that these results imply that lighting has no effect on productivity. This claim, however, can easily be refuted. There is a continuum of task performance associated with lighting conditions ranging from no light to plenty of light. Without light we can see nothing; only when there is sufficient light to see the necessary details does performance become possible. It is self-evident that as the amount of light on the task increases, it becomes possible to see finer detail more quickly, so performance increases until it becomes limited by some factor other than the visibility of the necessary details.

Other studies in the literature support the role of illuminance in task performance. Stenzel (1962) measured the output from a leather factory over a four-year period, in the middle of which he introduced a change in lighting installation. The work involved punching out defect-free outer leathers from skins for handbags, purses and other leather goods, using iron shapes and mallets. From 1957 to 1959, the lighting was provided by daylight, supplemented by local fluorescent lighting giving an illuminance of 350 lx. From 1959 to 1961, daylight was virtually eliminated and a uniform 1000 lx was provided by general fluorescent lighting. The average monthly performance for 12 people who were present throughout the four years is shown in Figure 4-2. There is a statistically significant improvement in performance, with the higher illuminance giving the better performance.

Figure 4-2 Mean monthly performance for cutting leather shapes for the years 1957/59 and 1959/61. Performance is normalized for the average performance over the years 1957/59 (After Stenzel, 1962).


4-4

Buchanan et al. (1991) measured the effect of increasing the illuminance in the pharmacists’ working area on the dispensing error rate in a high-volume outpatient pharmacy. Based on an examination of 10,888 prescriptions, they found that increasing the illuminance from 485 lx to 1570 lx led to a statistically significant reduction in prescription error rate from 3.9% of prescriptions filled to 2.6%.

These studies illustrate three problems common in such field studies. First, only two lighting conditions were used, and those were widely different. Thus the results do little more than demonstrate that changes in lighting can influence the performance of the task studied. Second, there is considerable uncertainty about the most important aspect of lighting for the change in performance, because the change in lighting installations used to change illuminance altered several different aspects of the lighting simultaneously. In these circumstances, to ascribe the changes in output to the difference in illuminance alone may be misleading. Third, there is no guarantee that changes in aspects of the task other than the lighting conditions did not change during the measurement period. Often changes in lighting conditions are made at the same time as changes in decor, furnishings, equipment, working arrangements, and personnel. If such additional changes were part of the field study, then ascribing any changes in output to lighting alone is also misleading.

These limitations of field situations have led researchers to perform studies of illuminance and task performance in the laboratory, where tasks simulating real life tasks can be done under controlled conditions. Simulated work studies have been done by Stenzel and Sommer (1969), who examined sorting screws of different sizes and crocheting stoles; Smith (1976), who studied threading a needle; Bennett et al. (1977), who also studied needle probing as well as micrometer reading, map reading, pencil note reading, drafting, vernier caliper measurement, cutting, and thread counting over a range of illuminances from 10 lx to 5000 lx; and McGuiness and Boyce (1984), who examined kitchen work. Most of these tasks showed an improvement in performance with higher illuminances, although the amount of improvement varied for different tasks. This variation should not be surprising because these studies measured task performance; in addition to the differences in the stimuli the tasks present to the visual system, the tasks also differ in their cognitive and motor components. The variation is most evident in two studies by Smith and Rea, (1978, 1982). In the first study (Smith and Rea, 1978), the researchers measured the performance of a small number of subjects, in two different age groups, when proofreading texts of three different levels of print quality for misspelled words. Measurements of speed and accuracy were taken at four different illuminances ranging from 10 lx to 4890 lx. Figure 4-3 shows the effect of illuminance, which is to increase performance as illuminance increases, particularly for the older subjects reading the poor quality text. In the Smith and Rea (1982) study, the same apparatus and range of illuminances were used, but this time the subjects were asked to read a text and then answer questions about their comprehension of the text. This time, there was no difference in the speed and accuracy with which the task was done for the different illuminances and the different age groups although there was a small decremental effect of poor quality text. Reading for comprehension has a much larger cognitive component than proofreading.

The field studies and simulated work studies leave no doubt that lighting conditions can affect task performance. The problem is to identify the range of lighting conditions that allows an improvement in task performance to occur, for each specific task. However, two caveats must be kept in mind. The first is the concept of specificity. The results of any study of the effects of


4-5

lighting on task performance are inevitably specific to the task, because different tasks have different combinations of visual, cognitive, and motor components. This specificity makes it impossible to generalize from the performance of one task to another. The second caveat is the concept of “allowing” task performance. Lighting cannot produce task performance; only the worker can do that. What lighting can do is to make details easier to see and colors easier to discriminate without producing discomfort or distraction. The worker can then use this increased visibility to produce output if he/she is so motivated or is not limited by some other non-visual factor.

Figure 4-3 Mean performance score for proof reading good and poor quality text, plotted against illuminances for young (18 - 22 years) and older (49 - 62 years) subjects. The performance score was calculated from the accuracy of proof reading with a bonus for speed. The maximum possible performance score was 35 (After Smith and Rea, 1978).

4.3 Light Spectrum and Task Performance

4.3.1 Achromatic Tasks

The studies discussed above examine the effect of illuminance but do not consider any effects of light spectrum. Smith and Rea (1979) systematically investigated both light level and spectrum. Illuminances ranged from about 7 to 2000 lx; cool white fluorescent, metal halide and high pressure sodium lamps were used as illuminants. Moreover, they studied, in combination with the two lighting variables, the impact of the luminance contrast and the “quality” of the visual task as well as the subjects’ age. Subjects were in two age groups, younger than 30 years and between 50 and 60 years. Task materials were number lists printed in high and low contrast (0.8 and 0.3) and were both typed (8-point type; 12 characters per inch) and handwritten (numbers


4-6

were approximately the same size and spacing). All subjects were required to find mistakes in these task materials under the different light conditions. Light level, subject age, task contrast and task “quality” all had statistically significant effects, both in terms of task performance and subjective ratings of difficulty. Light spectrum showed no statistically significant effect for either measure.

More recently, Berman et al. (1993) showed that light spectrum can influence performance for small, briefly-flashed, low luminance contrast, achromatic tasks, specifically the accuracy with which the orientation of a Landolt ring can be identified. Figure 4-4 shows the percentage of correctly reported orientations for Landolt rings with a gap size subtending approximately 2 min arc at the subject's eye and presented at four levels of luminance contrast and four levels of background luminance, for two different light sources. One light source was scotopically-enriched, was greenish-blue in color and had a scotopic / photopic ratio of 4.31. The other light source was scotopically-diminished, was pink in color and had a scotopic / photopic ratio of 0.24. (For comparison, metal halide light sources have scotopic / photopic ratios of about 1.4 while high pressure sodium light sources have scotopic / photopic ratios of about 0.4). The marked non-white color appearance of the light sources used ensures that they are unlikely ever to be adopted for general interior lighting practice. However, they do produce differences in task performance, as shown in Figure 4-4. The scotopically-enriched light source consistently has better task performance than the scotopically-diminished light source, for different luminance contrasts and background luminances, until the performance reaches the maximum possible. A similar pattern of results has been found for elderly subjects exposed to the same light sources (Berman et al, 1994).

The proposed explanation of these findings rests on the role of pupil size. Specifically, pupil size in a large visual field is determined predominantly by the response of the rod photoreceptors, even in photopic conditions; the greater the response from the rods, the smaller the pupil area (Berman et al. 1992). For the light sources described above, the pupil area under the scotopically-enriched light source was 40% smaller than under the scotopically-diminished light source. A smaller pupil area has three effects on the retinal image; it reduces the retinal illumination, it increases the depth of field and it reduces aberrations. The first of these effects, the reduction in retinal illuminance, can be expected to degrade visual performance. The other two, increasing the depth of field and reducing aberrations, can be expected to improve the quality of the retinal image and hence to improve visual performance. All these effects are small, and the trade-offs they produce will depend on the inherent quality of the individual’s optical system. An individual who is perfectly refracted will gain little from increasing the depth of field, so might be expected to experience deterioration of visual performance under a light source that produces smaller pupil sizes. However, most people do not have perfect refraction. For these people, the evidence suggests that light sources that promote smaller pupil sizes can increase visual performance somewhat for achromatic resolution tasks, where the task conditions place it close to threshold; e.g., small size, low luminance contrast and limited exposure time. In the experiment described above, the gap in the Landolt rings subtended approximately 2 min arc at the subject's eye, the highest luminance contrast used was 0.4 and the Landolt ring was only displayed for 200 ms.


4-7

Figure 4-4 Means and associated standard errors of the proportion of Landolt Ring orientations, presented for 200 ms on a spectrally neutral background, that were correctly identified, plotted against luminance contrast, for four different background luminances: a = 11.9, b = 27.7, C = 47.0, d = 73.4 cd/m2. In all four diagrams, the upper curve is for the scotopically-enriched illuminant (surround field scotopic luminance = 228 cd/m2) and the lower curve is for the scotopically diminished illuminant (surround field scotopic luminance = 13 cd/m2). Both illuminants produced a surround field photopic luminance of 53 cd/m2 (After Berman et al., 1993).

These results are interesting but probably have limited utility for lighting practice. As noted above, the effect of light spectrum on task performance was not apparent in the Smith and Rea study (1979). Moreover, Rea et al. (1990) manipulated pupil size over a large range by changing the reflectance and size of the area surrounding the task. They found no statistically significant effect of surround size and reflectance and, thus, pupil size, on the performance of the numerical verification task (see Section 4.7.3.1), a task that was large in size, continuously viewed and presented at two levels of luminance contrast, one low and one high (0.15 and 0.86). There can be little doubt that pupil size can influence the visual performance of small, briefly-flashed, low-contrast achromatic tasks. The questions that still need to be answered are the extent to which changes in pupil size affect suprathreshold task performance and/or if people like the enhanced retinal image quality produced by smaller pupil sizes.


4-8

4.3.2 Chromatic Tasks

All the above discussion has been concerned with the effect of light spectrum on the performance of achromatic tasks, but it is also necessary to consider how light spectrum affects the performance of tasks involving color. There are two roles for color in tasks. The first role is to increase the visibility of the task by providing a color difference between the task and its background, in addition to any luminance contrast; e.g., printing in different colors. For this role the color appearance of the task is immaterial. The only thing that matters is how different it is from the color of the background. The second role of color is where the actual color of the task is meaningful; e.g., the color of a warning sign. The light spectrum can affect the performance of both types of task because light spectrum is one of the factors that determines the appearance of colors. Therefore, light spectrum can increase or decrease the color difference between the task and its background and can change the color name given to a surface of fixed spectral reflectance.

The effect of color difference on task performance is limited to low luminance contrast conditions. Eklund (1999) examined the ability of people to read an exit sign from a distance when the letters and the background could be varied in color and in luminance contrast. Figure 4-5 shows the accuracy with which the sign was read at different luminance contrast values, both positive and negative. When the letters and the background have the same color, the accuracy starts to decline to chance level (50%) as luminance contrast falls below about 0.3, but when there is a color difference there is no decline in the accuracy of reading the sign, even when the luminance contrast is zero. This result implies that light spectrum will only matter for the performance of chromatic tasks where luminance contrast is low. In these conditions, a light spectrum that enhances the color difference between the task and the background will also enhance task performance.

Figure 4-5 Recognition performance, measured as the proportion of times the orientation of the word EXIT was correctly recognized, plotted against luminance contrast for color-normal observers. The legend is read as “letter color on background color”. The light sources used are LEDs. The “Green” LED has a peak emission at 530 nm. The “Red L” LED has a peak emission at 660 nm (After Eklund, 1999).


4-9

The effect of light spectrum on tasks where the color of the target is itself meaningful depends on the degree of color discrimination required. Where only coarse discrimination is required, e.g., telling red from blue or green, as might occur for a warning sign, then only light sources with low CIE General Color Rendering Indices, such as the low- and high-pressure sodium and mercury vapor lamps, will cause confusion about colors (Jerome, 1977; Glass et al, 1983; Collins et al, 1986). Where fine color discriminations are required, e.g., when grading diamonds, great care is required in the choice of light source in order to maximize the change in the color used as the basis of judgment (yellow in the case of diamonds). This means that different light sources will be better for identifying colors in different objects. Recommendations of the light source to be used for fine color discrimination, in a number of different industries, have been made (IESNA, 2000).

If no specific recommendation is available, then a general rule is that the higher the CIE General Color Rendering Index, the better will be the ability to discriminate colors. The extent to which different light sources will make it possible to discriminate colors can be estimated by using the Macadam ellipses (Figure 4-6). Each Macadam ellipse sets the boundary at which a given percentage of people are able to determine that two colors, one with chromaticity coordinates at the center of the ellipse and one with chromaticity coordinates on the ellipse, are just noticeably different (MacAdam, 1942; Wyszecki and Stiles, 1982). MacAdam’s ellipses were determined in conditions that offer the maximum sensitivity to color differences: side-by-side comparison, unlimited observation time, foveal viewing, and photopic operation of the visual system. Changing any of these factors and adding distracting or confusing stimuli can be expected to increase the difference in color needed to reach discrimination threshold (Narendran et al., 2000).

Figure 4-6 The MacAdam ellipses plotted on the CIE 1931 chromaticity diagram. The boundary of each ellipse represents ten times the standard deviation of the color matches made for the indicated chromaticity (After Wyszecki and Stiles, 1982).


4-10

4.4 Light Distribution and Task Performance

The distribution of light on and around the task can affect task performance either by changing the contrast of the retinal image of the task or by altering the adaptation of the visual system. The contrast of the retinal image can be altered in two ways, by disability glare and by veiling reflections.

Disability glare is caused by scattering of light in the eye (Vos, 1984), the scattered light forming a luminous veil over the retinal image of the scene. The effect of the luminous veil is to reduce the luminance contrasts in the retinal image. As luminance contrast is a major determinant of visual performance, a change in luminance contrast can affect task performance. Whether it does or not depends on the luminance contrast in the absence of disability glare. Tasks that require extracting information from low contrast details, such as when the visual system is operating close to threshold, will be susceptible to disability glare. Tasks that are characterized by a high luminance contrast will not be sensitive to disability glare. Quantitative estimates of the effect of disability glare on on-axis visual performance can be obtained by applying the modified luminance contrast to the RVP model of visual performance (see Section 4.7.3.1). Disability glare is not common in interiors but can be produced by poorly aimed spotlights and by the view of a bright sky or the sun through a window.

Veiling reflections are luminous reflections from specular or semi-matte surfaces that physically change the contrast of the visual task and therefore change the stimulus presented to the visual system. The two factors that determine the nature and magnitude of veiling reflections are the specularity of the material being viewed and the geometry between the observer, the target and any sources of high luminance. If the object is a perfect diffuse reflector (i.e., a lambertian reflector), no veiling reflections can occur. If the object has a specular reflection component, veiling reflections can occur. The positions where they occur are those where the incident ray corresponding to the reflected ray that reaches the observer’s eye from the target comes from a source of high luminance. This means that the strength and magnitude of veiling reflections can vary dramatically within a single lighting installation (Boyce, 1978).

When the specularity is the same across the whole task, for example on the page of a glossy magazine, then the effect of veiling reflections is to decrease the luminance contrast, because the same veiling luminance is added to luminances in the print and the background. However, when the print and paper differ in specularity, veiling reflections may either increase or decrease the luminance contrast of the print depending on the relative reflectances and specularities of the print and background. Probably the most extreme case is where a combination of specularly reflecting, dark material is used for the print and a matte, light material is used for the background. In this case, strong veiling reflections can cause the polarity of the luminance contrast to reverse.

The effect of veiling reflections on luminance contrast may be quantified by adding the luminance of the veiling reflection to the appropriate components in the luminance contrast formula. For glossy ink writing on matte paper, the luminance of the veiling reflections should only be added to the luminance of the ink. For a glossy magazine page or a VDT screen, where there is a specularly reflecting transparent coating over the whole surface, veiling reflections occur over the whole surface. In this case the luminance of the veiling reflections should be added to all terms in the luminance contrast formula. The extent to which changes in luminance


4-11

contrast alter on-axis visual performance can be estimated using the RVP model of visual performance.

As would be expected for a phenomenon that changes luminance contrast, veiling reflections have been found to reduce task performance. Rea (1981) showed that performance of the numerical verification task (see Section 4.7.3.1) deteriorated with decreasing luminance contrast, no matter whether the change in luminance contrast was caused by changing the reflectance of the printing ink or by veiling reflections (Figure 4-7). Reitmaier (1979) and Sanders and Bernecker (1990) have demonstrated similar changes in task performance in the presence of uniform veiling reflections for printed material and computer displays, respectively.

Figure 4-7 The performance score on the numerical verification task plotted against luminance contrast. The different levels of luminance contrast were obtained by different ink pigment densities, and viewing from different directions, using light of different percentage polarization to modify veiling reflections (After Rea, 1981).

It is commonly assumed that a lighting installation that produces veiling reflections is poor quality lighting, but this is not always the case. For example, display lighting of specularly reflecting objects is commonly designed to produce highlights that reveal the specular nature of the surface. Physically, veiling reflections and highlights are the same thing. Highlights can also be used to enhance performance. For example, highlights attached to deflections in the paint surface have been used to enhance the detection of dirt particles in the paint finish of automobiles (Wiggle et al., 1997; Lloyd et al., 1999)

Both disability glare and veiling reflections affect task performance. Veiling reflections do so by changing the luminance contrast of the task while disability glare directly changes the luminance contrast of the retinal image of the task. Light distribution is a determinant of disability glare and veiling reflections, but these are not its only effects. Light distribution can also change the luminances of areas surrounding the task. There is plenty of evidence that threshold visual


4-12

performance is sensitive to the immediate surround of the target (Lythgoe, 1932; Adrian and Eberbach, 1969, McCann and Hall, 1980). For example, McCann and Hall (1980) measured the contrast sensitivity of sinusoidal gratings surrounded by areas of different sizes and brightnesses. Like many other researchers, they showed that the luminance of the surround should be the same as the average luminance of the gratings for optimum performance. However, they also showed that the area of the surround important to contrast sensitivity was dependent on the size of the grating. From such results it is reasonable to infer that the immediate surround is important to the performance of tasks, as indeed it must be because the immediate surround ultimately forms the background which is an element in calculating luminance contrast. But what about the area that is more remote from the task, especially for suprathreshold tasks? Rea et al. (1990) measured performance on the numerical verification task in a black room, on a desk where the angular subtense of the task background, which had the same reflectance as the paper containing the task, ranged from 12° to 100°. Outside the area of the desk defined by this angle, the surface was either black or grey. The results showed only a slight influence of the black or grey area on the performance of the numerical verification task, performance being determined by the contrast of the task itself rather than the relative luminance of the more remote surround. Similar results were found by Eklund et al. (1999), who measured performance on a data-entry task over a complete day in three small offices, lit by three different lighting installations that produced the same illuminance on the task but very different distributions of light on the walls and ceiling. There was no statistically significant difference in the performance of the data-entry task in the three offices, but there was a statistically significant difference in the performance of the data-entry task when the task was presented in different print sizes.

Such results imply that task performance is largely determined by task characteristics, and that the wider area outside the immediate surround of the task is unimportant, except when the visual task is near threshold. Thus, lighting of areas distant from the task will have little effect on task performance, which is probably why people who use a task lamp in an otherwise unlit room still find it satisfactory lighting. The other possibility to consider is that although the luminance distributions of remote surroundings have little effect on visual performance, these studies say nothing about how well people like such conditions when exposed to them every day. It is possible that the luminance distribution of the space will affect human performance more generally through changes in mood. Certainly, retailers go to considerable expense to provide a luminous environment that they hope creates the right mood and sends the desired “message” about their store.

4.5 Polarization and Task Performance

Physically, light is comprised of electromagnetic waves that vibrate in a direction perpendicular to the direction in which the light waves propagate. In unpolarized light, these waves vibrate equally in all possible directions. When the light is polarized the waves vibrate in some directions more frequently than in others. For 100% polarized light, all the waves vibrate in only one direction (Shurcliff, 1962; Lighting Research Center, 1993; Clear and Mistrick, 1996). The possibility of using polarized light to enhance task performance is worth considering because, given that the geometry between the luminaire, task and viewer is such that specular reflections can occur and that the plane of the task is appropriate for the plane of polarization of the light, polarized light can reduce veiling reflections. Reducing veiling reflections usually implies an increase in luminance contrast and, if the surface is colored, an increase in the saturation of the


4-13

color. Clear and Mistrick (1996) have shown that polarized light can increase, decrease, or leave unchanged the contrast of reflecting materials. The direction and magnitude of the contrast change is determined by the percentage polarization of the light, the reflection characteristics of the materials forming the contrast, and the geometry between the luminaire, the materials and the viewer.

Rea (1981) examined the effect of polarization on the performance of the numerical verification task. Three different luminaires were used, producing 99%, 19%, and 8% polarized light at 42° from the downward vertical, this being the mirror angle for the viewer. The numerical verification task was presented on a horizontal plane, while the polarizing luminaires emitted different percentages of vertically polarized light. Figure 4-8 shows the effect of the percentage of polarization, viewing geometry, ink pigment density and ink specularity on task performance. Examination of Figure 4-8 shows that the different percentages of polarization have no effect when the viewing angle is at 90° to the direction required for veiling reflections to occur. However, when the viewing angle is 0°, i.e., when the viewer is at the mirror angle from the luminaire, some effect of polarization is evident. The black matte ink material shows little change in performance with percentage polarization, but the gloss black ink material does, performance deteriorating as the percentage of polarization is reduced. As for the gray ink materials, both the gloss and matte ink materials show a reduction in performance as the percentage of polarization declines. These changes can be related to the measured luminance contrast of the print. Luminance contrast is virtually unchanged for the different percentages of polarization at 90°, but there are marked changes for 0°. Further, the matte gray materials show the lowest luminance contrasts and hence might be expected to show the largest changes in performance with percentage polarization.

There can be little doubt that, given the right conditions of viewing geometry and task specularity, polarized light can enhance task performance where it enhances luminance contrast from an existing low level. The change in visual performance that can be expected to follow from introducing polarized light can be predicted by applying the modified luminance contrast to the models of visual performance (see Section 4.7.3). The problem of determining the value of polarized light is essentially the practical one of deciding how frequently the necessary conditions are likely to occur, how easy it is to produce the polarized light, and what other means are available to enhance luminance contrast. The answers to these questions will vary from one situation to another.


4-14

Figure 4-8 The combined effects of percentage polarization, viewing orientation, ink pigment density and ink specularity on the performance score on the numerical verification task. The 0�� viewing orientation allows veiling reflections to occur. The 90�� viewing orientation does not. The percentage polarization at the angle from the luminaire to the task for each luminaire type is PM = 8%, MP = 19%, and LP = 99% (After Rea, 1981).

4.6 Visual Search

All the above discussion of the impact of lighting conditions on task performance has been concerned with tasks in known positions. If viewers know the location of the task they need to see, they will be able to bring the fovea of the retina to bear on that position and hence be able to examine the task with the most discriminating part of the visual system. However, there are a


4-15

number of tasks where the location of the relevant object or scene is unknown, e.g., visual inspection tasks where the objective is to find defects in the product that may occur anywhere on its surface. Such inspection tasks involve two separate but successive components. The first is the search for and identification of any defects. In this component, the observer scans the object looking for defects. The second component is making the decision on what to do about the defects. Lighting can only directly affect the first component.

Studies of eye movements made while searching for defects have revealed a common pattern of fixation and saccade. The observer searches through a series of fixation pauses with rapid saccadic eye movements between them. Figure 4-9 shows such an eye movement pattern made by an inspector examining mens’ briefs held on a frame. Observations of this sort illustrate that the search pattern is often systematic (Megaw and Richardson, 1979), based on the inspector’s expectations about where the defects are likely to occur. The fixation and saccade pattern of visual search implies that the defect, or something that may be a defect, is likely to be first detected off-axis, i.e., in the peripheral visual field, and subsequently confirmed by bringing the fovea to bear on it through a saccadic eye movement. Therefore, the essential act for rapid visual search is off-axis detection of a defect. For a uniform field, where any departure from uniformity is a defect, the probability of off-axis detection can be related to the visibility of the defect. Figure 4-10 shows search time plotted against defect size for the inspection of a sheet of glass for a single defect (Drury, 1975). It is clear that as the defect size increases, which will make it more visible, the search time decreases. The concept used to model search time is the visual detection lobe, i.e., a surface centered on the fovea that defines the probability of detecting the defect at different deviations from the fovea within a single fixation pause. Figure 4-11 shows some probability data for detecting targets of different sizes and contrasts. From such results it is possible to calculate a visual detection lobe for each target by assuming radial symmetry about the visual axis. As would be expected, such visual detection lobes have a maximum at the fovea; the probability of detecting the defect decreases as the defect is located further off-axis. Clearly, different defects will have different visual detection lobes. A large-area, high-contrast hole in some sheet material will have a large visual detection lobe while a small-size, low-contrast hole will have a small lobe. The size of the visual detection lobe matters because, provided the interfixation distance is related to it and the total search area is fixed, the total time taken to cover the search area is inversely proportional to the size of the visual detection lobe. Visual detection lobes can be measured directly by psychophysical procedures or estimated from threshold performance data available for peripheral vision (Boff and Lincoln, 1988). Howarth and Bloomfield (1969) have suggested a simple equation, based on a random search pattern, which can be used to predict search times. The basic form of the equation is

tm = tf x (A/a)

where tm = mean search time

tf = mean fixation time

A = total search area

a = the area around the line of sight within which the target can be detected in a single fixation, i.e., the visual detection lobe for the fixation time.


4-16

For the inspection of a given article, A is likely to be fixed and tf is likely to be reasonably constant so the mean search time becomes proportional to the reciprocal of the visual detection lobe.

Figure 4-9 The pattern of eye fixations made by two inspectors examining men’s briefs held on a frame. S = start of scan path, C = end of scan of front and one side, rotation of frame and continuation of scan across back and sides, E= end of search. Inspector M examines only seams while inspector D examines fabric as well (After Megaw and Richardson, 1979).


4-17

Figure 4-10 Mean of the median search time for detecting a single flaw in a glass sheet, plotted against flaw size (After Drury, 1975).

Figure 4-11 The probability of detection of targets of (a) contrast = 0.058, size = 19 min arc; (b) contrast = 0.080, size = 10 min arc; (c) contrast = 0.044, size = 10 min arc; within a single fixation pause, plotted against deviation from the visual axis. Each curve can be used to form a visual detection lobe for each target by assuming radial symmetry about the visual axis.

Inditsky et al. (1982) have proposed an alternative model of visual search performance using a measure of target visibility based on visibility lobe per glimpse and assuming a random search


4-18

pattern. It would be interesting to know if either of these models can accurately predict visual search times for targets of different sizes and contrasts.

For searching uniform, empty fields, the visibility of the defect off-axis is all that is necessary to predict the search time. However, for many inspection tasks, the defect appears not in a uniform, empty field but in a cluttered field, i.e., one in which many different items are present. In this situation, the visibility of the defect alone is not enough to predict the search time. The other factor that must be considered is the conspicuity of the defect, i.e., how easy it is to distinguish the defect from the other items in the search field. A high visibility is not enough to guarantee a high conspicuity. For high conspicuity, the defect should differ from the other items in the field on as many dimensions as possible. Figure 4-12 shows the mean search times for two observers searching for rectangular or square targets among an array of square non-targets, plotted against an index of discriminability. Discriminability is given by the square of the difference in the square roots of the areas of the targets and non-targets (Bloomfield, 1975). This result, and others like it, suggests that it should be possible to estimate an effective visual detection lobe where the lobe is determined not only by the target but also by the other items amongst which it is seen. Engel (1971, 1977) has shown how such an effective visual detection lobe can be measured and has demonstrated that it is related to the probability of finding a target within a fixed time.

It is important to appreciate that there are many different dimensions besides size on which the target can differ from the items around it. Williams (1966) studied search times for finding a specific item in a display of 100 items that could vary in size, shape, color, and the two-digit number contained. The inspector was asked to locate a particular number, where either the number alone was specified, or the number and various combinations of the size, color, and shape of the item that contained it were specified. The mean search times for the different target specifications are given in Table 4-1.

Figure 4-12 Mean search times for two observers searching for rectangular or square targets among an array of non-square targets, plotted against an index of discriminability. The discriminability index is given by (A0.5 - B0.5)2 where A and B are the areas of the targets and non-targets, respectively (After Bloomfield, 1975).


4-19

Table 4-1 Mean time to find a target for different target specifications

Target specification Mean time (s)

Number only 22.8

Number and shape 20.7

Number and size 16.4

Number, size and shape 15.8

Number and color 7.6

Number, color and shape 7.1

Number, color, size and shape 6.4

Number, color and size 6.1

It can be seen that the longest search times occurred when the number alone was specified and the shortest search times occurred when the color and size of the item in which the number was located were also specified. Table 4-1 also shows that some aspects of the specification can be more important than others. Specifically, whenever the color of the item in which the desired number lay was specified, short search times were achieved. Specifying the color reduces the items where the number might be to 20% of the total, but so does specifying the shape. Yet the latter has much less effect on search time. The explanation for this result is that the differences in color give a much larger effective visual detection lobe than do differences in shape. This is apparent from the measurements of eye movement patterns. Whenever the color was specified, fixations were made predominantly on items of that color. When the shape was specified, there was little change in the eye movement patterns from when the number alone was given. It is interesting to note that one of the evolutionary advantages claimed for color vision is the ability to detect targets against a dappled background where luminances are varying randomly (Mollon, 1989).

The finding in Williams (1966) that including color in the specification of the search item had a major impact on the speed of visual search should not be taken to mean that color coding is a guaranteed way to enhance visual search performance. If the differences between the colors used by Williams (1966) had been slight, it is doubtful if specifying the color would have been anywhere near as effective. A more general way to identify what dimension is important is to consider the signal-to-noise ratio between the target items and non-target items, on each dimension. The higher the signal-to-noise ratio on a given dimension, the greater is the importance of including that dimension in the specification of the search item.

Given that the efficiency of visual search is determined by the actual or effective visual detection lobe, the role of lighting conditions as regards visual search is to increase the size of the visual detection lobe. Many of the lighting techniques used for visual inspection are aimed at either increasing the visual size or luminance contrast of the defect, either by casting shadows or by using specular reflections (IESNA, 2000). Further, the light spectrum can be chosen to enhance a


4-20

difference in color between the defect and the rest of the area to be inspected. Probably the most widely applicable aspect of lighting which aids visual search is to increase the illuminance on the search area. Figure 4-13 shows the mean search times for finding a two-digit number located among 100 such numbers randomly arranged on a table, plotted against the illuminance on the table. Increasing the illuminance leads to shorter search times, particularly for the small-size, low-contrast target (Muck and Bodmann, 1961). While illuminance is generally a useful method of reducing search times, it should not be used without thought. If the effect of increasing illuminance is to decrease the luminance contrast, or effective visual size, visual search may take longer. Lloyd et al. (1999) demonstrated such a case for the detection of specks of dirt in painted automobile body shells. For this application, the defect is made more visible by attaching a highlight to each defect, the highlight arising from the local deflection in the paint surface caused by the dirt particle below the surface.

The problem with identifying the best lighting conditions for visual search is that they are likely to be specific for each situation. It is clear that lighting that increases the effective visual size or luminance contrast or color difference of the search item, or that makes the visual system more sensitive to differences in visual size, luminance contrast, or color differences, is likely to improve the performance of a visual search task. However, the specifics of such lighting depend critically on the area to be searched, what that area contains, and the luminous and color characteristics of the items in the area, including the defect.

Figure 4-13 Mean search times for locating a specified integer number from a random array of one hundred such numbers, plotted against illuminance (After Muck and Bodmann, 1961).


4-21

4.7 Lighting and Visual Performance

4.7.1 The Analytical Approach

The specificity of task performance measurements implies that the only way to understand how lighting might affect task performance in general is to measure its effect for every task of interest. To avoid this time-consuming and expensive option, an alternative approach has been adopted: the determination of visual performance rather than measurement of task performance. Visual performance is the performance of a task in which the impact of the cognitive and motor components has been eliminated, or at least minimized. As visual performance is solely determined by the capabilities of the visual system, and lighting influences both the capabilities of the visual system and the stimuli presented to the visual system, it is assumed that there should be a stable, generalizable relationship between the lighting conditions, the stimuli the task presents to the visual system, and visual performance.

The first attempt to produce a general model of the effects of lighting conditions on visual performance was made by Weston (1935, 1945), based on a suggestion by Beutell (1934). What Weston did was devise a very simple task in which the critical detail was easy to identify and measure. This task is usually known as the Landolt ring chart (Figure 4-14). It consists of a series of Landolt rings, with the gap in the ring being oriented in one of the cardinal directions of the compass. The critical detail of the Landolt ring is the gap. The size of the critical detail is the angular size of the gap and the critical contrast is the luminance contrast of the Landolt ring against its background. When doing the Landolt ring task, subjects are asked to read through the chart and mark in some way all the rings that have a gap orientated in a specified direction. The time taken to do this and the number of errors made under different lighting conditions are measured. These measures are then combined to form measures of speed and accuracy of work. Specifically, speed is measured as the total time taken divided by the number of rings correctly marked. Accuracy is measured as the total number of rings that could have been marked divided by the number of rings correctly marked. Speed and accuracy, as defined above, are then multiplied, the time taken to mark the rings with a gap in the specified direction when they were marked with red ink subtracted, and the reciprocal of the result taken. This combined form of speed and accuracy is known as the performance score. The subtraction of the time taken to mark all the rings made highly visible with red ink is an attempt to minimize the contribution of the cognitive and motor components of the task performance to the performance score, and hence to make performance score a measure of visual performance rather than task performance. The rationale behind this decision is that by marking the rings that need to be identified with red ink, the visual component is minimized so what is left are predominantly the cognitive and motor components.


4-22

Figure 4-14 A Landolt ring chart. All the rings in a chart are of the same size and luminance contrast but different charts can have rings of various sizes and luminance contrasts.

Figure 4-15 shows the results obtained by Weston in his second study (Weston, 1945). A number of conclusions can be drawn from these results. First, the effect of increasing illuminance follows a law of diminishing returns, i.e., that equal increments in illuminance lead to smaller and smaller changes in performance until saturation occurs. Second, the point where saturation occurs is different for different sizes and contrasts of critical detail, saturation occurring at lower illuminances for large-size, high-contrast tasks than for small-size, low-contrast tasks. Third, larger improvements in visual performance can be achieved by changing the task, i.e., changing either the size or contrast of the critical detail, than by increasing the illuminance, at least over any illuminance range of practical interest. Fourth, it is not possible to make a visually difficult task, such as searching for small-size, low-contrast items, reach the same level of performance as a visually easy task simply by increasing the illuminance. Although these conclusions have been derived from Weston (1945), they have been confirmed many times since, using the Landolt ring chart (Boyce, 1973) and using other abstract tasks (Khek and Krivohlavy, 1967) and simulated real tasks (Smith and Rea, 1978, 1982, 1987; Rea, 1981).


4-23

Figure 4-15 Mean performance scores for Landolt ring charts of different critical size and contrast plotted against illuminance (After Weston, 1945).

The analytical approach adopted by Weston has served to demonstrate the general form of the relationship between lighting conditions, task characteristics, and visual performance. It has also suggested how the visual difficulty of a task might be quantified and hence, how the effect of lighting on the performance of the visual component of the task might be determined. However, as used by Weston, the results do have some limitations. Rea (1987) reviewed Weston’s studies of 1935 and 1945 and observed that the performance scores achieved at different illuminances for similar rings were not consistent between the two studies, nor were the trends in performance score with illuminance. Rea (1987) also objected to the performance score metric. Specifically, he objected to the fact that the number of correct rejections, that is, the number of rings examined and correctly rejected as not having a gap in the specified direction, was ignored. Without considering the number of correct rejections, the measures of number of gaps inspected per unit time (speed) and the proportion of gap orientations correctly identified (accuracy) must be imprecise. There is also a more general objection to the performance score metric. While both speed and accuracy are important aspects of task performance, it is better to treat them as separate but related measures of performance rather than to multiply them together, as is done in the performance score metric . The ideal approach would be to consider the effect of lighting on speed at a constant level of accuracy or accuracy at a constant level of speed (Clear and Berman, 1990). Unfortunately, the multiplication of speed and accuracy to obtain a single-number measure of task performance is common (Muck and Bodmann, 1961; Waters and Loe, 1973; Smith and Rea, 1978, 1979) but it is nonetheless arbitrary and serves to hide the effect of the experimental conditions on two rather different aspects of task performance. For all these reasons, Weston’s results can be considered as indicative of general trends, but provide a dubious basis for a quantitative model of visual performance.


4-24

4.7.2 The Visibility Approach

While the analytical approach might be considered remote from the real world because it is based on one standard task, it is, at least, based on a measure of suprathreshold performance. The visibility approach was even more ambitious in that it attempted to develop a system for predicting the visual performance of real tasks at suprathreshold levels from threshold performance measurements. The concept behind the visibility approach is that the ease of seeing a task, such as a printed page, can be quantified by the separation of the stimuli the task presents to the visual system from their threshold values. The higher the task’s characteristics are above threshold, the greater the visibility of the task. It is then assumed that the visibility of the task is consistently related to task performance. The visibility approach was developed over many years by Blackwell (1959), (CIE 1972, 1981) based on his extensive measurements of threshold contrast at different luminances (Blackwell, 1946; Blackwell and Blackwell, 1980). The metric of task visibility was Visibility Level, this being defined as

Visibility level = Equivalent contrast / Threshold contrast

Threshold contrast was defined as the luminance contrast of the visibility reference task, namely detecting the presence of a 4-min arc luminous disc presented against a uniform luminance field for 0.2 s. Equivalent contrast was the luminance contrast of the visibility reference task that is matched in visibility to the task of interest. Blackwell developed an instrument, called a visibility meter, to measure equivalent contrast, as well as multiplying factors to correct for departures in the viewing conditions of the task of interest from those used in the visibility reference task. He then made visibility level measurements for the stimuli used in other visual performance studies. Initially, these measurements led to the belief that there was a universal relationship between Visibility Level and visual performance. However, as more data on more tasks were collected, it became obvious that this was not true. It was concluded that the missing factors were the extent of search and scan of the visual field and the need to gather information off-axis. This led to the development of another standard task, called the visual performance reference task. This consisted of five 4-min arc Landolt rings, one in a central position and the other four at the four cardinal points of the compass, equidistant from the center. By altering the luminance of the Landolt rings or the luminance of the background, the Visibility Level of the rings could be changed. By altering the presentation time or the separation of the central and peripheral rings, the difficulty of the task could be altered. Using the understanding gained from the visual performance reference task, a model was developed to predict the effect of lighting conditions on visual performance for a wide range of tasks (CIE, 1981). The model consisted of two sets of transfer functions operating in series. The first was concerned with how the lighting conditions affected the Visibility Level. The second was concerned with how the Visibility Level influenced visual performance. The model had three components linked to Visibility Level. These three components were related to extracting information from the details of the task, the stability of eye fixation and the precision of eye movements. The model could be made to fit independently obtained sets of experimental results, but only by varying the weighting of the four components. As more data sets were examined, the number of correction factors that had to be introduced to make the model fit increased to such an extent that the model lost all credibility.

Although the visibility approach is rarely mentioned today, it should be appreciated that the concept of Visibility Level does have one advantage; namely, it provides a single measurement of the way that several different aspects of a visual stimulus, e.g., size, luminance contrast, color


4-25

difference, and non-sharp edges, combine to make a task difficult or easy to see. Indeed it has been used for this purpose in other fields of lighting, such as road lighting (Lipinski and Shelby, 1993). The problem with the concept of Visibility Level is the tendency for people to assume that equal Visibility Levels correspond to equal levels of task performance. This is not the case. Different tasks show different relationships between task performance and Visibility Level depending on the nature of the task (Rea, 1984; Clear and Berman, 1990; Bailey et al., 1993). The error in the visibility approach developed by Blackwell was to assume that something as complex as the suprathreshold performance of tasks with different visual and non-visual components, occurring on- and off-axis, could be predicted from something as simple as an on-axis threshold measurement.

4.7.3 The Stimulus-Response Approach

The principle adopted in the stimulus-response approach is to establish a quantitative link between the physical characteristics of the task, the stimulus, and the performance of the task, the response. The Visibility Level of the visual characteristics of the task is considered to be an unnecessary intervening variable. In the stimulus-response approach the visual characteristics of the task are to be defined by quantities that can be directly measured.

Two quantitative models of the effect of lighting conditions on visual performance have been developed; the Relative Visual Performance (RVP) of Rea and Ouellette (1991) and the Visual Performance model of Adrian and Gibbons (1994, 1999). Each will be discussed in turn.

4.7.3.1 The Relative Visual Performance (RVP) model

The origin of the RVP model lies in a study of the effect of luminance contrast on performance of the numerical verification task (Rea, 1981). In this task, two printed pages, the reference page and the response page, each containing a column of 20 five-digit numbers, were used. The five-digit numbers on the reference page were random numbers. The corresponding numbers on the response page were the same except that some of the five-digit numbers differed by one digit (Figure 4-16). Data were collected on the time taken to compare the two columns, the number of discrepancies missed (misses), and the number of false discrepancies identified (false positives), for a constant illuminance on the printed pages of 278 lx and for a wide range of luminance contrasts for the reference page. Luminance contrast was varied by changing the reflectance of the ink used to print the numbers; by making the ink of either specular or matte reflectance; by changing the geometry between the luminaire providing the illuminance, the numerical verification task and the observer; and by varying the percentage of vertical polarization in the light incident on the numerical verification task. The response page was printed in matte ink on matte paper in a high luminance contrast. The visual size of the numbers was held constant by using the same font and point size (8 point type, 12 characters per inch) for the printing and by controlling the location of the subject’s head with a chin rest. Figure 4-17 shows the change in the mean time taken, misses and false positives against luminance contrast. It is evident that for a wide range of luminance contrasts, there is very little change in any of the performance measures. However, as luminance contrast drops below about 0.4, the time taken starts to increase, accelerating as luminance contrast decreases further. A similar pattern can be seen for the misses and false positive data, although they only start to increase from their very low level at lower luminance contrasts than those that cause an increase in time taken. These data illustrate


4-26

that both the speed and accuracy of performance deteriorate with reduced visibility in a non-linear manner. They also demonstrate that luminance contrast is a major determinant of task performance, no matter how that luminance contrast is achieved.

58313 58313

51424 51424

26538 26538

10508 10508

35148 35148

53427 53427

99147 99147

54483 54483

39154 39154

39417 39417

52807 52807

55394 55394

32393 32393

83118 83118

31510 31510

53009 53009

01632 01632

29394 29394

49619 49619

54101 54101

Figure 4-16 The numerical verification task. The reference list is on the left and the response list is on the right.


4-27

Figure 4-17 The mean times taken, number of misses and number of false positives for the numerical verification task, plotted against luminance contrast (After Rea, 1981).

The first complete version of the RVP model (Rea, 1986) was derived from data collected from people doing the numerical verification task using the same experimental materials, experimental room and procedures as in the earlier study (Rea, 1981). Data were collected for a range of illuminances from 50 lx to 700 lx (giving a range of background luminances from 12 cd/m2 to 169 cd/m2) and a range of luminance contrasts from 0.092 to 0.894.

The data on time taken to compare a set of 20 five-digit numbers, the number of misses, and the number of false positives were very similar to those obtained in the earlier study (Rea, 1981) and by others using the numerical verification task (Slater et al., 1983). In developing the RVP model, Rea decided to use only the time taken data. This decision was made for a number of reasons, the most important of which was that the numbers of misses and false positives were small and subject to random fluctuations. These factors made misses and false positives less reliable measures of performance than time taken. The second reason was that the variation of all three measures with luminance contrast was very similar. Figure 4-18 shows the change in the reciprocal of time taken plotted against luminance contrast at each of the background luminances. Figure 4-18 shows the compressive function with luminance contrast that would be expected from the results in Figure 4-17 but there are two effects of increasing luminance that should be noted. The first is that, at the same luminance contrast, performance is better at higher luminances. The second is that performance tends to saturate at lower luminance contrasts for higher luminances.


4-28

Figure 4-18 The mean reciprocal of time taken to perform the numerical verification task plotted against luminance contrast at four background luminances (After Rea, 1986).

The time taken to compare the reference and response pages used as the basis of Figure 4-18 is a measure of task performance, in that it includes both visual and non-visual components. To produce a time measure that could reasonably be called a measure of visual performance, Rea subtracted two elements of time from the total time taken. The first was an estimate of the time taken to make a mark against any response page number that contained a discrepancy. The second was the time taken to read the numbers in the response list. The remaining time was then the time taken to read the numbers in the reference list. It is the reciprocal of this time that was taken as a measure of visual performance from which the RVP model was developed. Figure 4-19 shows the form of the RVP model plotted against background luminance of the range 12 to 169 cd/m2 and luminance contrast over a range 0.08 to 1.0. The vertical axis is a relative measure (RVP) calculated from the reciprocal of the time taken to read the reference page, normalized to a value of 1.0 at a background luminance of 169 cd/m2 and a luminance contrast of 1.0. The shape of this model has been described as the “plateau and escarpment” of visual performance (Boyce and Rea, 1987). Over a wide range of task and lighting variables, the change in relative visual performance is slight, but at some point it will start to deteriorate rapidly. The evolutionary advantage of having a visual system that can cope with a wide range of visual conditions with little change in speed of response is obvious.


4-29

Figure 4-19 The Relative Visual Performance (RVP) model of visual performance, based on the time taken to read the reference page of the numerical verification task (After Rea, 1986).

The second form of the RVP model (Rea and Ouellette, 1988) was developed using a very different approach. In the previous approach, a task with some similarity to real-world tasks, the numerical verification task, was used and the performance measure later adjusted to make it a measure of visual performance. The second approach used a task that could be taken as a direct measure of the speed of visual performance. Specifically, the performance measure was simply the reaction time to the onset of a square stimulus. Detecting the presence of something as opposed to its absence is arguably the simplest visual task possible. It has very little cognitive component, and because the only motor component is to press (or release) a button when the stimulus is presented, the motor component is slight as well. Reaction time measurements were taken for stimuli with a wide range of luminance contrasts, both positive and negative; angular sizes; and adaptation luminances. An equation of the same form as that used to fit the reciprocal of time taken to do the numerical verification task was applied to the reciprocal of reaction time and fitted the data well. To convert the reaction times into a relative measure, the differences in reaction time between each stimulus condition and the shortest reaction time (obtained for the largest size, highest contrast and highest adaptation luminance) were calculated. This measure shows the increase in reaction time following reductions in visual size, luminance contrast, or retinal illuminance. Figure 4-20 shows the differences in reaction time that occurred for a range of luminance contrasts and visual sizes at retinal illuminances ranging from 0.63 to 801 Trolands.


4-30

Figure 4-20 The difference in mean reaction time to targets of four different visual sizes (2, 14, 130 and 2,800 microsteradians) plotted against luminance contrast. For each figure the retinal illumination decreases from 801, 160, 31, 6.3, 1.6 to 0.63 trolands, from left to right (After Rea and Ouellette, 1988).

At this point there were two alternative forms of the model, one based on the time taken to read the reference page of the numerical verification task (Rea, 1986) and one based on the difference in reaction time for detecting the presence of a target (Rea and Ouellette 1988). Fortunately for simplicity, Rea and Ouellette (1991) were able to develop a method for converting the reaction


4-31

time difference into units of RVP. They did this by identifying a common set of stimulus conditions and then developing a linear transformation between the two measures. The use of this method means that the differences in reaction times can be expressed in terms of RVP. Figure 4-21 shows the RVP scores for the reaction time measurements plotted against luminance contrast and retinal illumination, for each visual size of the detection target. The solid angle subtended at the eye of the average size of the digits used in the numerical verification task is 4.8 microsteradians, so this part of Figure 4-21 is comparable with Figure 4-19. The similarity between the two figures is obvious.

Figure 4-21 The RVP model of visual performance based on reaction time data. Each element shows the RVP value plotted against luminance contrast and retinal illuminance, for a fixed target size (After Rea and Ouellette, 1991).


4-32

Regardless of how internally consistent a model is, it cannot be said to be validated until it can be shown to predict the results of a study which is not part of the dataset used to construct the model. Therefore, what must be considered now is how well the RVP model predicts the results of independently collected data.

There are three answers to this question. The first comes from Rea (1987). In this paper, Rea examined a number of experiments in which the performance of a visual task had been measured, including those of Weston, and considered their suitability for acting as a validation of the RVP model based on the numerical verification task. For a variety of reasons, including insufficient documentation, confounding of variables, inappropriate performance measures, and the use of tasks requiring visual search with significant non-visual components, he concluded that only one of the previous studies was suitable as a check of the accuracy of prediction. In this study by McNelis (1973), people were asked to name two small letters presented briefly and separated by 10°. The letters were presented at different contrasts and at different background luminances. Predictions of RVP for the luminance contrast and illuminances used showed good agreement with the normalized accuracy scores for naming the letter viewed first, but the agreement was less successful for the lowest contrast used (0.125), possibly because of errors in the measurement of the luminance contrast.

The second answer comes from the study by Bailey et al. (1993). In this study, the speed of reading unrelated words was measured, when the words were presented in sizes ranging from 2- to 20-point print, at three different luminance contrasts (0.29, 0.78, and 0.98), and a range of background luminances from 11 to 5480 cd/m2. The reading speed was calculated from the time taken to read each row of words and the number of words in each row. The time taken to read each row was measured from a recording of the eye movements made during the reading. Figure 4-22 shows the measured reading speeds plotted against print size for print of three different contrasts. The plateau and escarpment shape of visual performance is again evident. Bailey et al. (1993) then applied the formula developed by Rea and Ouellette (1988) to the stimulus conditions in their reading speed experiment. The resulting fit of the predicted reading speed to the measured reading speeds was good for print sizes of five points and above, for all luminance contrasts and background luminances. However, as print sizes decreased below five points, extrapolations from the model were increasingly error-prone. This is not unreasonable because the data on which the RVP model is based did not cover stimulus sizes equivalent to 5-point print and smaller, so such smaller print sizes were outside the range of the model. It is likely too, that as print size decreases, the ability to read the word becomes limited by the ability to resolve the detail of the letters, regardless of luminance contrast. The fact that the basic formula used in the RVP model can be applied to independently collected data on a reading task and can give accurate predictions of reading speed over a range of print sizes of practical interest is encouraging as evidence that the underlying concept is correct.


4-33

Figure 4-22 Mean reading speed measured in words / second plotted against letter size in points, for different luminance contrasts (After Bailey et al., 1993).

The third answer comes from the work of Eklund et al. (2000). They measured the effect of different lighting and print conditions on the sustained performance of a repetitive, self-paced, data-entry task. Twenty-four subjects worked for almost four hours at a data-entry task in one of three identical, private, windowless offices. All three offices were lit by similar fluorescent, parabolic lighting systems. The installations were fitted with dimming systems to allow the illuminance on the work to be systematically changed to four levels (29, 103, 308, and 1035 lx). During the four hours of work, the subject entered sets of five 10-symbol alphanumeric codes, the sets being printed in a cyclical series of print sizes (6, 8 12 and 16 point) and luminance contrasts (0.10, 0.22, 0.47, and 0.93). In total, task performance measurements were taken for 60 combinations of illuminance, print size, and luminance contrast. The task performance measurements were the times taken to enter a block of fifty alphanumeric symbols correctly. Any data entry errors were detected by the software that controlled the experiment; subjects had to correct errors before proceeding, thereby increasing the time taken. The result of this procedure is to fix the accuracy of performance at 100%. Figure 4-23 shows the mean work speed, derived from the reciprocals of the work times, plotted against illuminance, for each luminance contrast and print size. These data were used to test the precision of the RVP model based on the reaction time data given in Rea and Ouellette (1991). The mean inked area of numbers printed in the different point sizes was measured, as were the luminance contrasts and the background luminances. These values were then inserted into the RVP model to predict the relative visual performance. The RVP values were then normalized at the value found for the largest size (16 point), highest contrast (0.92), and highest illuminance (1035 lx). The measured mean work speeds on the data entry task were also normalized for the same conditions. Figure 4-24 shows the normalized RVP values plotted against the normalized measured mean work speeds. It is clear that the RVP model fits the measured data well.


4-34

0.005

0.010

0.015

0.020

0.0250.005

0.010

0.015

0.020

0.025

Wor

k Sp

eed

(1/s

)

10 100 1000

Illuminance (lx)

0.005

0.010

0.015

0.020

0.025

Wor

k Sp

eed

(1/s

)

10 100 1000

Illuminance (lx)

0.005

0.010

0.015

0.020

0.025

C=0.93

C=0.47

C=0.22

C=0.10

6 Point

8 Point

12 Point

16 Point

Figure 4-23 Mean work speeds for the data entry task for four print sizes. Each graph shows the mean work speed plotted against illuminance (lx) for four luminance contrasts. The error bars indicate 95% confidence intervals. Note that the plot for the 6 point print contains data for only three luminance contrasts. 6-point print of luminance contrast 0.10 could not be read by a significant number of subjects at any of the illuminances used (After Eklund et al., 2000).


4-35

0.00

0.20

0.40

0.60

0.80

1.00

Pred

icte

d Pe

rform

ance

0.40 0.60 0.80 1.00

Performance

ANN Model

VP Model

RVP Model

Perfect Prediction

Figure 4-24 Normalized RVP values predicted by the RVP model and the VP model for the data-entry task plotted against the normalized measured mean work speeds. Data are also shown for the empirical data-entry task performance (ANN - DETP) model. This model was developed from the data-entry task performance data so it should provide a good fit (After Eklund et al., 2000).

These three comparisons serve to demonstrate the robustness and validity of the RVP model. The Eklund et al. (2000) study also removes some of the doubts about its utility. Some of these doubts have been derived from the conditions under which the reaction time data used to form the RVP model were collected. For example, the reaction time data were collected monocularly, using an artificial pupil, and with the subject’s head at a fixed distance from the stimulus. Further, the only task was to detect the presence of a square target of a fixed size, a task which does not require resolution of detail. In the Eklund et al. (2000) experiment, the subjects used natural pupils, could move closer or further from the data-entry material as they wished, and were asked to read alphanumeric characters. Other doubts have arisen because different letters and numbers of the same nominal print size vary in inked area. In the Eklund et al. (2000) experiment, the data-entry material was a random collection of letters and numbers of the same nominal size, and hence a collection of letters and numbers that varied in inked area. The fact that the RVP model is consistent with the mean work speed measured over a four-hour work period in real-world conditions suggests that these doubts are unjustified. Over time the variations in the stimuli presented to the visual system caused by different viewing distances and different inked areas average out and can be ignored.


4-36

4.7.3.2 The Visual Performance (VP) Model

The origins of the VP model lies in the work of Weston (1945), who performed a series of experiments using an array of Landolt rings and measured visual performance as the product of the speed and accuracy with which gaps of a given orientation were identified (see section 4.7.1). This can be considered a measure of visual performance because the motor component, in the form of the time taken to mark the identified rings was subtracted from the total time taken. Using Weston's data, Adrian and Gibbons (1994, 1999) undertook an extensive curve-fitting exercise to obtain a quantitative model to predict the product of speed and accuracy of performance on the Landolt ring task from the measured angular size, luminance contrast and background luminance. The result was a pair of equations predicting visual performance for different ranges of the luminance contrast, one being for a luminance contrast equal to or greater than 0.35 and one for less than 0.35, both equations applying to angular sizes equal to or greater than 1.5 min arc.

The value of such equations lies in how well they can predict the results of independently collected data. Adrian and Gibbons (1994, 1999) claim that the VP model produces relative visual performance predictions consistent with a number of early studies of visual performance (Simonson and Brozek, 1948; Muck and Bodmann, 1961; McNelis, 1973; Waters and Loe, 1973; and Smith and Rea, 1978). The key word here is consistent. There can be no doubt that the trends in the data and in the predictions of the VP model are similar but the predictions are rarely good fits to the data. This should not be too surprising, given the variety of performance tasks used, ranging from large area visual search to sustained, paced letter recognition, and the different measures of performance used by the different authors. One recent experiment that directly tested the ability of the VP model to predict visual performance is that of Eklund et al. (2000). They measured the effect of different lighting and print conditions on the sustained performance of a repetitive, self-paced, data-entry task. During four hours of work, the subject entered sets of five 10-symbol alphanumeric codes, the sets being printed in a cyclical series of print sizes (6, 8 12 and 16 point) and luminance contrasts (0.10, 0.22, 0.47, and 0.93). Each period of four hours was done at one of four fixed illuminances (29, 103, 308, and 1035 lx). The task performance measurements were the times taken to enter a block of fifty alphanumeric symbols correctly. Any data entry errors were detected by the software that controlled the experiment; subjects had to correct errors before proceeding, thereby increasing the time taken. The result of this procedure is to fix the accuracy of performance at 100%. To use the VP model, the size of the alphanumeric symbols, expressed as angular stroke width, the luminance contrast of the symbols and the background luminance based on the illuminance on the task and the reflectance of the page were measured. These values were then inserted into the VP model to predict the relative visual performance. The VP values were then normalized at the value found for the largest size (16 point), highest contrast (0.92), and highest illuminance (1035 lx). The measured mean work speeds on the data entry task were also normalized for the same conditions. Figure 4-24 shows the normalized VP values plotted against the normalized measured mean work speeds. It is clear that the VP model does not fit these measured data, predicting a much lower level of performance than actually occurred.

4.7.4 Comparisons and Limitations

Figure 4-24 poses a problem to anyone attempting to identify the best form of stimulus - response model. At first glance it appears that the RVP model is clearly better than the VP model


4-37

in predicting relative visual performance. However, it is undeniable that the VP model approximates the results of a number of the early studies of visual performance, although it should also be said that Rea (1987) examined many of these early studies and concluded, for a variety of reasons, that they were not suitable for checking the predictions of the RVP model. So how can this problem be resolved? One approach is to ask how much precision is required for useful predictions to be made. Unfortunately, the answer to this question will vary from task to task as the magnitude of the visual component varies. Another approach is to examine the various tasks that have been used in the development of the stimulus-response models to determine if there is some systematic difference in their visual requirements (Mistrick, 1996). One clue to a possible difference lies in the responses of the two models to changes in luminance contrast. Adrian and Gibbons claim that their VP model is much more sensitive to reductions in luminance contrast than the RVP model. One situation in which much greater sensitivity to luminance contrast is evident is when off-axis viewing occurs. Figure 4-25 shows the mean percentage correct identification of three-letter words presented for 100 ms, both in the fovea and one degree off-axis (Timmers, 1978). Even a slight deviation from foveal viewing makes visual performance much more sensitive to luminance contrast. Further, visual acuity can be shown to be dramatically reduced with only a slight deviation from the fovea (Aulhorn and Harms, 1972). A notable feature of the early studies that have been used to demonstrate the validity of the VP model is that they often involve either visual search, which means peripheral vision is required for the performance of the task, or they have limited time of exposure to the stimulus, which means there may not be enough time for the subject to bring the image of the target onto the fovea. Specifically, Simonson and Brozek (1948) presented letters on a moving belt, each letter being visible for 560 ms through a viewing slit ; Muck and Bodmann (1961) did a visual search task where the subject had to find a specified number from 50 others arranged at random; McNelis (1973) presented two letters separated by ten degrees, for 400 ms, the subject's task being to read and identify both letters; Waters and Loe (1973) used a modified version of the Landolt ring task in which there where three concentric Landolt rings so the subject had an element of visual search built into the task; and Smith and Rea (1978) used a proofreading task in which the subject had to find errors in a written text, again a visual search task. Even the Landolt ring task used by Weston (1945), who provided the data from which the VP model was constructed, involves a visual search element because the gap being sought can occur at a number of different positions. However, the reaction time task used in the second and final version of the RVP model does not have a visual search element.


4-38

Figure 4-25 Mean percentage correct identifications of three letter words presented for 100 ms in the fovea and at 1�� eccentricity, at different levels of luminance contrast. The error bars are the standard error of the mean (After Timmers, 1978).

It is important to appreciate that this proposed impact of off-axis performance is all speculation but at least it is speculation that can lead to testable hypotheses. It would certainly be interesting to test the extent to which the RVP and VP models could accurately predict the relative visual performance for a task that could be systematically varied in the degree of off-axis vision required. Until this is done, the decision as to which of the two stimulus-response models is better has to be in favor of the RVP model. It has been fully and rigorously developed, from an abstract model of visual performance to the performance of a realistic task done under realistic conditions. This suggests the possibility that the RVP model might be used to predict the percentage change in visual performance for other similar tasks. If one wants to know what the percentage change is for a particular task, where either visual size, luminance contrast, or illuminance is varied, one can calculate the RVP value for the two conditions and express the ratio of RVPs as a percentage. If there is no difference between the RVPs for the two conditions being compared, the percentage will be 100%. To scale what this percentage means for absolute task performance, one can measure the actual task performance at one combination of visual size, luminance contrast, and illuminance and calculate the RVP value for the same conditions. Then, any percentage change in RVP can be scaled to give the change in absolute amount of work done.

The existence of this possibility should not be taken to mean that the long story of the study of the relationship between light and task performance is at an end. A more accurate understanding would be that the RVP model represents a concept that can be applied to a limited range of tasks.


4-39

The tasks for which it is most suited are those that are dominated by the visual component, that do not require the use of off-axis vision, that present to the visual system stimuli that can be completely characterized by their visual size, luminance contrast, and background luminance only, and that have values for these variables that fall within the ranges used to develop the RVP model.

The limitation of a small non-visual component is derived from the fact that RVP measures visual performance, not task performance. These two types of performance are only likely to coincide when the non-visual components are relatively small or when parallel processing of information can occur, as seems likely to have happened in the data-entry task in Eklund et al. (2000). If the non-visual component is large and parallel processing is not possible, then RVP predictions will overestimate the consequences of a change in visual conditions on task performance. In order to extend the use of the RVP model to tasks with significant non-visual components, a task analysis procedure must be developed that could be used to determine the relative impacts of the visual and non-visual components on the performance of any task.

The limitation to tasks using only foveal vision occurs because the experiments used to derive and validate the RVP model all used tasks in which the subject knew where to look to gain the necessary information so there should be little use of peripheral vision.

The limitation to tasks with stimuli that can be completely characterized by the visual size, luminance contrast, and background luminance is necessary because these are the factors built into the RVP model. Other aspects of visual stimuli that can be important are color difference and retinal image quality. When applied to print, for example, Smith and Rea (1978) and Colombo et al. (1987) have both shown that the effect of illuminance was much greater for fragmented print than for entire print. As for color difference, Eklund (1999) has shown that the ability to read an exit sign at a distance is maintained even at zero luminance contrast, provided there is a color difference between the letters and the background.

The limitation to suprathreshold conditions is necessary because, in threshold conditions, virtually all aspects of the stimuli to the visual system matter and many of them can be expected to interact. The RVP model does not allow for such complexity.

Finally, it is important to realize that other models of task performance have been constructed. For example, Eklund et al. (2000) applied a sophisticated curve fitting procedure based on neural networks to their data to produce what they called the data-entry task performance (ANN-DETP) model. Figure 4-26 shows the predicted mean work times for the data-entry task plotted against the measured mean work times. Clearly the model is a good fit, but it is only applicable to the data-entry task and cannot be applied to other tasks. Clear and Berman (1990) had a more general aim in mind when they put forward a model of task performance in which performance is determined by the addition of two components: one visual component that can be related to a Visibility Level-type metric, and one non-visual component that is independent of Visibility Level. They show that a model of this form can be made to fit the Rea (1986) numerical verification task performance data and can handle both speed and accuracy measures of task performance. A similar approach was used by Bailey et al. (1993) to fit their reading speed data, but in this case, the visibility level metric was based on visual size rather than luminance contrast. It is important to appreciate that these models are not models in the same sense as the RVP model. The differences are that the RVP model is a validated model of visual, not task,


4-40

performance, and all the measurements needed to make a RVP prediction can be obtained from physical measurements of visual size, luminance contrast, and background luminance of a task. The task performance models are either applicable only to the specific task, or require additional information before they can be used.

0.010

0.015

0.020

0.025Pr

edic

ted

Wor

k Sp

eed

0.010 0.015 0.020 0.025

Actual Work Speed

Actual

Model

Figure 4-26 Mean work speeds predicted by the data-entry task performance (ANN-DETP) model plotted against measured mean work speed. (After Eklund et al., 2000).

5-1

5 CIRCADIAN PHOTOBIOLOGY: CURRENT KNOWLEDGE

5.1 Circadian Photobiology in Context

All plants and animals exhibit patterns of behavioral changes over an approximately 24-hour cycle that repeat themselves over successive days. These rhythms are called circarhythms. The physiology that controls such circarhythms is called the circadian system, from the Latin, circa, for “about”, and dies, for “day”, about a day. Circadian systems occur in a wide range of life, from unicellular organisms to insects, plants, fish, mammals, and humans. Further, cyclic patterns called circannual rhythms occur over the seasons, such as the seasonal breeding of mammals and seed germination in plants. These patterns are believed to be controlled by gradual change in the light / dark ratio signaled by the circadian system.

The human circadian system has been the subject of extensive study for only the last two decades. Knowledge of the circadian system is therefore much less developed than it is for the visual system. The outlines of the physiology of the human circadian system are established, but there are still many details to be determined. One thing that is known is that exposure to light can have two effects on the human circadian system. It can entrain the system by shifting the phase of the system, and it can have an immediate alerting affect. These effects are important for human performance because the ability to work is reduced during the sleep part of the circadian sleep / wake cycle. If the circadian system is not entrained, or the worker is asked to work during the sleep part of the cycle, as may be the case for people working night shift, task performance is likely to be degraded. Further, this degradation of task performance will occur for all types of tasks, not just visual tasks. What is missing from our understanding of how light exposure affects task performance through the circadian system is what light spectrum is most effective, how much light is needed to produce an effect, and how the exposure to light over the twenty-four hours is integrated.

5.2 The Structure of the Human Circadian System

Like the visual system, the human circadian system starts with the eye, but unlike the visual system, it does not transmit information directly to the visual cortex. Rather, after leaving the eye, the circadian system proceeds along the retino-hypothalamic tract (RHT) to the suprachiasmatic nuclei (SCN) and then by way of the paraventricular nucleus (PVN) and the superior cervical ganglion to the pineal gland (Figure 5-1). In dark conditions, the pineal gland synthesizes the hormone melatonin, which is then circulated throughout the body by the bloodstream. The anatomical linkage between the eye and the pineal gland is called the retino-hypothalamic-pineal axis (RHP). Each of the components of the RHP will be considered in turn.

Circadian Photobiology: Current Knowledge

5-2

Figure 5-1 A simplified illustration of the retino - hypothalamic - pineal axis, i.e., from the retina to the suprachiasmatic nuclei (SCN) to the paraventricular nuclei (PVN) to the superior cervical ganglion and thence to the pineal gland.

5.2.1 The Retina

Light reaching the retina provides signals to both the visual system and the circadian system, but through different neural connections. For the visual system, there are four kinds of photoreceptors: three cone types, each with a different photopigment, and one rod type, with a different photopigment from any of the cones. The photoreceptor, or photoreceptors, used to regulate the human circadian system have not yet been identified, but one thing is clear. It is that the three cone photoreceptor system that supports photopic vision in humans is not the primary input to the circadian system (Brainard et al, 2001a). This suggests that photoreceptors other than those involved in vision influence the human circadian system (Foster et al., 1991; Czeisler et al., 1995; Ruberg et al., 1996). The possibility of a unique photosensor for the human circadian system is supported by the discovery of photoreceptors based on vitamin B2 in mice (Miyamoto and Sancar, 1998). The photoreceptors used for vision in mice are based on vitamin A. Even more surprising is the finding that these photoreceptors are not located at the same level of the retina as the rod and cone photoreceptors used in vision, but rather at the collector and ganglion cell level. The search for the photoreceptors that provide input to the human circadian system is currently the subject of much activity, the two latest papers published on the subject suggesting


5-3

the photoreceptor has an opsin-like spectral sensitivity curve, with a peak sensitivity at about 460 nm (Brainard et al, 2001b; Thapan et al, 2001).

As for the characteristics of the channels that convey the electrical information from the retina to the SCN, these are more clearly understood. Measurements in mice and cats of the retinal ganglion cells that project to the SCN have shown that such ganglion cells are rare, spread evenly across the retina, and have extensive dendritic arbours (Moore et al, 1995; Provencio et al, 1998). Further, there is no attempt to preserve the location of each ganglion cell in its projection to the SCN. This structure emphasizes that as far as the circadian system is concerned, the retina is a “photocell” designed to collect irradiance information. This concept is reinforced by measurements of the electrical output of cat retinal ganglion cells that project to the SCN. These ganglion cells were found to have 2- to 5-degree receptive field sizes, to respond most strongly to still or very slowly moving stimuli, and to give sustained responses (Pu, 2000). All these characteristics suggest a system designed to be insensitive to sudden fluctuations in light pattern.

5.2.2 The Suprachiasmatic Nuclei

The suprachiasmatic nuclei (SCN) are paired structures in the hypothalmus of the brain. The SCN are recognized as the endogenous oscillator, i.e., the master clock, in mammals, including humans (Klein, et al., 1991). Measurements of the response of SCN neurons in rats to light indicate that they have very large receptive fields (20 to 40 degrees) with no surround antagonism (Groos and Mason, 1980). The lack of surround antagonism is markedly different from the receptive field properties of the rat visual system, but it must be remembered that the antagonistic surround is useful to identify edges in the visual scene, a property that is essential for an imaging system but unnecessary for a “photocell.” It has also been shown that SCN neurons in rats respond to continuous light stimulation with a sustained response that can be as long as 30 to 60 minutes (Meijer and Rietveld, 1989). Other measurements have shown that the output of the SCN in rats is characterized by a high threshold and a limited dynamic range, features that serve to convert the differences between night and day into a simple square wave response (Groos and Meijer, 1985).

5.2.3 The Pineal Gland

The pineal gland synthesizes and secretes the hormone melatonin. Melatonin is easily absorbed into the bloodstream and hence serves as a chemical messenger throughout the body (Menaker, 1997). Melatonin detectors, which act as receivers of the message, have been found in many parts of the body. The message carried by melatonin is that of time as determined by the SCN, the master clock. The essential role of melatonin is to synchronize the activation of many other physiological functions to the times in the 24-hour cycle when they should occur (Cagnacci et al., 1997). Normally, high levels of melatonin are secreted at night and low levels are secreted during the day (Klein et al., 1991; Wetterberg, 1993). However, the presence of light at night suppresses the synthesis of melatonin, the amount of suppression being determined by the retinal illuminance and the duration of exposure, for up to about one hour. Exposure durations longer than one hour have minimal further impact on melatonin concentration; recovery after exposure is complete within about one hour. Figure 5-2 shows the mean melatonin concentrations


5-4

measured for six subjects at half-hour intervals between 22:00 hours and 05:00 hours. Between midnight and 03:00 hours, they were exposed to an illuminance at the eye of 200, 400, and 600 lx from a “full-spectrum” fluorescent lamp. At other times the illuminance at the eye was 10 lx (McIntyre et al., 1989a). The pattern of suppression and recovery of melatonin concentration after exposure to light is clear.

Figure 5-2 Melatonin concentration before, during, and after exposure to different levels of illuminance at night (After McIntyre et al., 1989a).

5.3 Characteristics of the Human Circadian System

The retino-hypothalamic-pineal (RHP) axis has a number of important characteristics. Probably the most notable is the fact that it continues to oscillate even in the absence of any external cues to time. The period of this oscillation in humans is about 24.5 hours. This implies that when people live in conditions in which temporal cues have been removed, there will be a steady drift in sleep times over successive days. However, when time cues are available, the period of the oscillator is entrained to 24.0 hours so the timing of the sleep-wake cycle is stable. The fact that the RHP continues to oscillate even in the absence of light-dark variations demonstrates that the circadian system is not simply a passive response to the occurrence of external light-dark oscillations.

Given the presence of an endogenous oscillator that continues to operate even in the absence of light, what is the role of light exposure? The answer to this question is that light exposure entrains the endogenous oscillator to the actual light-dark cycle as it occurs over the day and over


5-5

the seasons. Depending on latitude, the length of the day and conversely the length of the night varies over the season. The longer the night, the longer the time melatonin is secreted. Animals that show distinct seasonal behavior have cells that measure the duration for which melatonin is present (Bartness and Goldman, 1989). These cells also regulate seasonal changes in behavior. Whether a similar process occurs in humans remains to be determined, but it is interesting to note that rates of conception in humans exhibit clear seasonal variations (Roenneberg and Aschoff, 1990 a and b); conception rates are higher in the spring or fall than in the summer or winter. It is also interesting to consider whether electric lighting has an impact on the seasonal adjustment of the circadian system. Conception rates suggest it does. Conception rates are highest in under-developed countries in the spring. In developed countries, where electric lighting is more widely available, conception rates are highest in the fall. Figure 5-3 offers more direct evidence that electric lighting can have an impact on the timing of the circadian rhythm. Figure 5-3 shows the mean timing of melatonin secretion for six individuals after exposure to a 14-hour dark period in controlled conditions and after exposure to 14-hour winter nights under natural conditions (Wehr, 1997). The shift in the timing may be due to the use of electric light in the evening, after the sun has set, although it is possible that other social or activity cues to time are present under the natural conditions and not under the controlled conditions.

Figure 5-3 Mean levels of plasma melatonin in constant dim light in six individuals after exposure to defined fourteen-hour dark periods (hatched area) under controlled conditions (faint line) and after they were exposed to natural fourteen-hour winter nights (same hatched area, bold line). The individuals' use of electric lighting, after dark, appears to have shifted their melatonin rhythm several hours later than would be expected relative to solar night (After Wehr, 1997).


5-6

5.4 Effects of Light Exposure on the Human Circadian System

The most commonly occurring impact of light exposure on the circadian rhythm is the entrainment to twenty-four hours produced by exposure to a regular light-dark cycle. But what happens if the light-dark cycle is disrupted, perhaps by light exposure occurring during what is normally a period of darkness? The answer is a shift in the phase of the circadian rhythm. Figure 5-4 shows schematically what happens to the sleep-wake cycle of people exposed to constant low light level conditions for 10 days, with a single pulse of bright light of about 2 hours duration occurring at some time on the fifth day. The upper part of Figure 5-4 shows that the subject's sleep period, indicated by the dark bar, free-runs throughout the ten days, and that, for some light pulse exposures, a step change in the phase of the sleep occurs on the sixth day. The magnitude and direction of this step change depends on when the light pulse occurs relative to the prevailing sleep-wake cycle. The pattern of magnitude and direction of phase shift is shown in the lower part of Figure 5-4, in a form that is called a phase response curve (PRC). What this shows is that exposure to a pulse of bright light during the circadian day has very little if any effect on the phase of the sleep-wake cycle in the next twenty-four hours. A pulse of bright light given early in the circadian night tends to delay the sleep-wake cycle in the next twenty-four hours, but a pulse of bright light given in the latter part of the circadian night tends to advance the phase of the sleep-wake cycle in the next twenty-four hours. The critical time at which the effect of a pulse of bright light changes from a phase delay to a phase advance is around the time that the body reaches minimum core temperature. For healthy young people, whose circadian system is entrained by a regular light-dark cycle, this minimum occurs about 1 to 2 hours before awakening. Given the rapid transition from phase delay to phase advance around the core body temperature minimum, what happens when the light exposure straddles the core body temperature minimum? The answer is that there is a reduction in amplitude of the circadian rhythm but no shift in phase, although the circadian system is then more sensitive to subsequent light pulses (Jewett et al., 1991). As for the magnitude of the phase shift, that depends on the magnitude of the light exposure: the higher the retinal irradiance, the greater the phase shift, until the limits of the system are reached. While the above understanding has been derived from strictly controlled light exposures, in everyday life people are exposed to light at many different times of the night and day. Fortunately, models of the effect of light exposure on the phase and amplitude of the circadian system are potentially able to predict the effect of complex patterns of light exposure (Kronauer, 1990, Kronauer et al., 1999, Forger et al., 1999).


5-7

Figure 5-4 A schematic illustration of the effect of light exposure on the phase of the circadian rhythm. Depending on the timing of the light pulse, the circadian rhythm can be advanced, delayed or left unchanged (After Moore-Ede et al., 1982).

The phase-shifting effect of light exposure occurs many hours after exposure. A more immediate effect is the suppression of melatonin synthesis and the consequent increase in alertness as measured by a change in the nature of electroencephalograph patterns, by an increasing core body temperature, and by reported feelings of alertness. Figure 5-5 shows the effect of alternate exposure to bright (>5000 lx) and dim (50 lx) light for 90-minute periods between midnight and 09:00 hours on core body temperature. The core body temperature trends towards a minimum at around 05:00 hours, but the alternate exposure to dim and bright light modulates that trend. Exposure to bright light tends to increase the core body temperature, while dim light tends to reduce it (Badia et al., 1991). Another aspect of these results is the speed of response of the core body temperature system to the onset and removal of light. Figure 5-2 suggests that the response occurs within about 30 minutes and recovery occurs in a similar time.


5-8

Figure 5-5 Modulation of core body temperature by alternating blocks of bright and dim light from midnight to 9 a.m. Open circles or triangles represent the bright light conditions. Filled circles or triangles represent the dim light condition. Circles started with 90 minutes of bright light at midnight, Triangles started with 90 minutes of dim light at midnight (After Badia et al., 1991).

5.5 Factors Determining the Effectiveness of Light Exposure

Given that the spectral sensitivity of the photoreceptors used to transmit the external light stimulus to the SCN is unknown, it is strictly inappropriate to discuss the role of light in this context at all. However, given the current state of knowledge, any other assumption is equally tentative. Further, most of the literature on the topic uses illuminance, measured in lumens / square meter (lux), as a metric for light exposure. This discussion will use the same terms.

Zeitzer et al. (2000) have established a sensitivity relationship for phase shifting and suppression of melatonin. They measured the melatonin concentration following exposure to 6.5 hours of light at a fixed illuminance centered 3.5 hours before the subject’s minimum core body temperature. The illuminances at the eye during the light exposure ranged from 3 to 9,100 lx. From the schematic shown in Figure 5-4, it would be expected that exposure to light at this time would lead to a delay in the circadian rhythm. Figure 5-6 shows the phase shift of melatonin concentration plotted against illuminance at the eye. The phase shift is in the expected direction but, more interestingly, the phase shift saturates, i.e., reaches 90% of the asymptotic maximum, at an illuminance of 550 lx while the half-saturation response is produced by an illuminance of about 100 lx.

Figure 5-6 also shows the percentage reduction in melatonin concentration during four hours of exposure to different illuminances at the eye. For melatonin suppression, saturation of suppression occurs at about 1000 lx and half-saturation occurs around 100 lx (Zeitzer et al., 2000). Similar results have been found by Brainard et al. (1988) and McIntyre et al. (1989b) for shorter exposure times.


5-9

Figure 5-6 (A) The phase shift in the melatonin rhythm following exposure to 6.5 hours of light at different illuminances. The illuminances are measured at the cornea. (B) The suppression of melatonin during the light exposure. A four parameter logistic model has been fitted through the data. The model is represented by the continuous line and the 95% confidence intervals by the dotted lines (After Zeitzer et al., 2000).

Illuminances at the eye of 1000 lx and 100 lx are approximately equivalent to horizontal work-plane illuminances of 5000 lx and 500 lx, respectively. This suggests that exposure to illuminances that occur in everyday lighting installations may not be enough to reliably entrain the human circadian system. If this is so, then daylight may be the main source of circadian entrainment. However, Wehr et al. (1995) failed to show any summer-to-winter seasonal variation in the duration of melatonin secretion when the male subjects were carrying out their normal activities, i.e., when exposed to artificial light in the early morning and evening, which suggests that conventional electric lighting can sometimes influence circadian entrainment.

While such findings are important for understanding the impact of lighting installations on the circadian system, a number of unanswered questions remain. For example, if exposure to everyday lighting installations is enough to ensure entrainment of the circadian system, why does the duration of melatonin secretion increase from summer to winter for 1 in 3 women and 1 in 8 men? Conversely, why does a seasonal change in melatonin duration not occur for 2 out of 3 women and 7 out of 8 men (Wehr, 1997)? Plausible answers lie in the possibility that modern lifestyles ensure that people are shielded from seasonal variations in natural light either because they spend most of their time indoors and/or because the amount of light provided by electric lighting installations is enough to determine the duration of melatonin secretion and/or because the provision of electric lighting allows for increased behavioral arousal in the evening, behavior that has been shown to influence the timing of the circadian system (Mrosovsky et al., 1989). Another factor to be considered is the difficulty in measuring the relevant light exposure in realistic situations. The illuminance that matters for entraining the circadian system is the


5-10

illuminance on the retina. This is evident from the work of Brainard et al. (1997), who have shown that melatonin suppression is greater for dilated pupils than naturally changing pupils and greater for two eyes exposed than for one eye exposed, for the same illuminance on the cornea. Further, Dawson and Campbell (1990) have pointed out that in most lit spaces, the illuminance reaching the retina can vary dramatically depending on the light distribution, the reflectances of the surfaces forming the space, and the direction of gaze. Thus, it is very difficult to know what the actual retinal illuminance is in realistic lighting situations, even at one moment in time, although a physical detector capable of measuring retinal illuminance has been developed (Van Derlofske et al., 2000).

Despite the acknowledged importance of the retinal illuminances, the photometric measure most commonly used in the study of the circadian system is the illuminance at the eye, although this is, at best, an approximation to the retinal illuminance because it ignores the transmittance of the ocular media of the eye. The illuminance on a horizontal plane, which is sometimes used as a measure of light exposure, is even further from the retinal illuminance. Also, it is important to remember that the spectral sensitivity of the photoreceptor(s) that provides a signal to the SCN in humans is presently unknown; therefore, illuminance, as conventionally measured, may not be the correct metric to quantify the effective stimulus to the circadian system. Of even greater significance is time, both the time of exposure during the cycle and the duration of exposure. The relationships between retinal illuminance, spectrum, timing and duration of light exposure need to be much more clearly understood before it will be possible to apply knowledge of the circadian system to lighting practice. Finally, there is the question of individual sensitivity. The maximum concentration of melatonin produced in darkness can vary widely between individuals (Waldhauser and Dietzel, 1985). This suggests that individuals will show wide differences in their sensitivity to light operating through the circadian system.

5.6 The Consequences of Trying to Work in Circadian Night

Humans are diurnal mammals that are active during the day and asleep at night. People will experience difficulty in performing any sort of task if they are asked to do it at a time when their circadian system is telling them to sleep, i.e., in its night period. Figure 5-7 shows speed and accuracy data collected from industrial workers performing different real-world tasks at various times during a twenty-four hour period (Folkard and Monk, 1979). There is wide variation in performance over the twenty-four hours, and performance during the night shift is usually worse than during the day.


5-11

Figure 5-7 Speed and accuracy data collected from workers performing real world tasks at various times during the twenty-four hours (After Folkard and Monk, 1979).

At first, the adverse results at night might seem inconsistent with the hypothesis that illuminances at levels typically used in interiors at night can entrain the circadian system. There are two reasons why these data do not directly support that hypothesis. The first is that the extent and speed of phase shifting is dependent on the pattern and intensity of light exposure, over the whole twenty-four hour period. Exposure to daylight, such as might occur on the journey to and from work, and which gives a much higher illuminance than most interior lighting, will probably dominate entrainment and, hence, may stop entrainment to the night shift. The second is that even with carefully controlled exposure to daylight, a 180-degree phase shift takes a number of days. Figure 5-8 shows the phase difference between the maximum core body temperature and the ideal time for the maximum to occur for 21 continuous nights of work in a laboratory (Monk et al., 1978). It is clear that adaptation begins from the first night, but it is only after about 12 nights that complete entrainment to the night shift occurs.


5-12

This pattern of slow adaptation suggests the possibility of using light exposure to systematically and rapidly phase shift the circadian system to the required condition. Czeisler et al. (1990) have demonstrated that a pattern of exposure to light can be developed that will produce a marked phase shift to an adapted state in four twenty-four hour days, even when the subjects are exposed to daylight on the journey home. This was achieved by long exposure to an illuminance in the range of 7000 to 12,000 lx during the night shift from 00:15 to 07:45 hours. Czeisler et al. (1990) also showed that, as a result of this adaptation, the subjects had a greater feeling of alertness and achieved better performance on mental arithmetic during the night shift than a control group who showed no adaptation.

Figure 5-8 Phase difference between the maximum core body temperature and the ideal time for that maximum to occur, for 21 continuous night shifts (After Monk and Folkard, 1983).

Eastman et al. (1994) performed a similar study on night-shift work, but in this case, light exposure and the wearing of dark welder’s goggles were experienced in all possible combinations. This meant that some subjects were exposed to 5000 lx illumination from light boxes at night, but were free to travel home without wearing the goggles. Others were exposed to less than 500 lx at night but wore the dark welder’s goggles during the journey home. Yet others were exposed to 5000 lx at night and wore the welder’s goggles on the journey home, while yet others were exposed to 500 lx at night but did not wear the goggles travelling home. All the subjects slept in darkened bedrooms receiving less than 500 lx in daytime. Either exposure to bright light at night or wearing the welder’s goggles during the day was effective in producing phase shifts in the circadian system. Both factors together gave the greatest phase shift; either factor alone gave some shift while having neither factor rarely produced any phase shift. This result emphasizes the point that to guarantee a phase shift, it is necessary to control light exposure throughout the twenty-four hour period.


5-13

Both the above studies were conducted in laboratories to determine the practicality of using controlled light exposure to accelerate adaptation to a sudden shift in work time. One application where controlled light exposure has been of practical value is in space flight. A space shuttle crew is required to start preparations for launch at about 2 a.m. After launch, the crew is split into two teams, both of which work twelve-hour shifts. Crews on early shuttle flights complained of fragmented and disturbed sleep. In 1990, the conference room in the crew quarantine quarters was fitted with a luminous ceiling capable of producing about 10,000 lx. The crew entered the quarantine quarters one week before launch and went through a light exposure pattern of bright light and darkness designed to adapt them to their anticipated work schedule. The returning crew reported much better sleep patterns in flight. Melatonin samples indicated that adaptation had occurred (Czeisler et al., 1991). Reports of the application of light exposure to more mundane industrial activities are rare, but one that is available suggests that bright light exposure has only limited value (Bjorvatn et al., 1999). Even though schedules for adaptation to night shift work have been published (Eastman, 1990), controlled light exposure as a means to adapt to night-shift work has rarely been used in practice. Possible reasons for this lack of impact on lighting practice are the difficulty of ensuring compliance with the light exposure pattern throughout the twenty-four hours in a conventional industrial context, the design problem of providing the illuminance at the eye needed to phase shift rapidly without causing visual discomfort, and the failure to identify a problem in night-shift work, probably caused by the low expectations of night-shift workers.

Another situation in which there is a need to rapidly shift circadian rhythm is after air travel across several time zones. The outcomes of such travels are similar to those of night-shift workers and include difficulty in sleeping at a time consistent with the destination, gastrointestinal illness, and decrements in alertness and performance. Collectively these symptoms are known as jet lag and occur because of the misalignment between the endogenous circadian clock and the exogenous light-dark cycle. Jet lag is usually associated with travel in an east-west direction. Travel in a north-south direction may cause similar symptoms due to fatigue and sleep loss, but it does not involve any phase misalignment between the endogenous and exogenous components of the circadian system. The time for circadian rhythms to resynchronize depends on the number of time zones crossed and the direction of travel. For flights across six to eleven time zones, re-entrainment occurs at a mean rate of about 90 minutes per day for westward travel and about 55 minutes per day for eastward travel (Aschoff et al, 1975; Klein and Wegmann, 1980). This difference between eastward and westward flight occurs regardless of whether the flights are outward or return, and whether they are day or night flights. The explanation of this difference is that re-entrainment after westward flight occurs by gradual phase delays, i.e., by temporarily having periods longer than 24 hours, while re-entrainment after eastward flight occurs by gradual phase advances, i.e., by temporarily taking periods shorter than 24 hours. As the free-running endogenous period is greater than 24 hours, phase advances represent a bigger change than phase delays, so entrainment will be faster for westward than eastward flights. Of course, these rates of re-entrainment are averages and cover wide individual differences. Re-entrainment rate is proportional to the difference in phase between the endogenous clock and the light-dark cycle, so re-entrainment is greatest immediately after the flight and progressively decreases. Further, different circadian rhythms re-entrain at different rates, and older people take longer to re-entrain than younger people (Boulos et al., 1995).

Given the proven ability of light exposure to shift the phase of the circadian rhythms, it is to be expected that exposure to light at the right time should be able to speed up the process of re-


5-14

entrainment after a rapid transit across time zones. Measurements of re-entrainment rates made before the role of light was understood offer some support for this view. Klein and Wegmann (1974) found that re-entrainment after travel across six time zones happened 50% faster for subjects who were allowed outdoor activities every other day than for subjects confined to their hotel rooms. Rapid adjustment has also been shown by military units following eastward airlifts, presumably because of their outdoor activity upon arrival (Graeber et al., 1981). More recent field studies have been less successful, although there is some evidence that exposure to bright light at the destination tends to lead to a consolidation of sleep in one sustained period (Czeisler and Allen, 1987; Cole and Kripke, 1989; Sasaki et al., 1989).

The most plausible reason for the lack of success of these field studies is the lack of control over the subject’s light exposure over the whole 24-hour period. There can be little doubt that exposure to light at the right times can accelerate re-entrainment, but whether this is a practical proposition for most travelers seems unlikely. Probably the best that can be expected is to modulate light exposure when outdoors by dark sunglasses. Software had been developed for scheduling exposure to light in order to ensure rapid resynchronization of the circadian system following transmeridional travel (Houpt et al., 1996). Of course, modifying behavior in order to match a schedule of light exposure is only worthwhile if the traveler is likely to stay at his or her destination for a number of days. For short stays of one or two days, it is probably better to try to maintain entrainment on the return destination. If this is the aim, the reverse pattern of outdoor light exposure is desirable.

Even if it were possible to accelerate re-entrainment at the beginning and end of a period of night-shift work or after a long flight across time zones, that would still leave a number of days when people would have to work during the circadian night. It is interesting to consider what the effect of this situation would be on task performance. First, the effect of trying to work in the circadian night can affect all types of tasks, not just visual tasks. This is because the circadian system affects the “platform” from which we operate and consequently affects all parts of the brain and body. Tilley et al. (1982) studied the sleep patterns and performance of shift workers operating a weekly, alternating, three-shift system. Workers doing night shift, who had to sleep during the day, had shorter duration sleep which was of degraded quality. As for performance, both simple reaction time and four-choice reaction time were longer during night shift relative to day and afternoon shift and tended to show a deterioration over the number of days on night-shift, probably because of the accumulation of sleep debt caused by the inferior sleep duration and sleep quality during the day. Cajochen et al. (1999a) also studied the sleep and performance patterns of people kept awake for 32 hours, i.e., a period covering a conventional day, starting at 08:00 hours and extending to 16:00 hours the next day. Figure 5-9 shows the patterns in core body temperature and plasma melatonin, both well-established circadian rhythms; eye blink rate, slow eye movements, stage 1 EEG patterns, and ratings on the Karolinska sleepiness scale, all of which are related to sleep; and performance on a reaction time task, a mental arithmetic task, and a short-term memory task. It is important to note that during the 32 hours of wakefulness, the subjects were exposed to a constant illuminance of 15 lx, i.e., there was no bright light treatment in this study. Figure 5-9 shows the expected pattern of decreased core body temperature and increased plasma melatonin at night. The sleep measures all show an increased propensity to sleep during the night with some recovery the next day. The performance measures all show a decrement in performance over the night with some recovery the following day although not enough to recover to the level of performance achieved at the beginning of the trial. Of particular interest are the results on the reaction time task. What is evident in these data is the increase in


5-15

range in reaction times at night. The fastest 10% of the reaction times at night are almost as fast as they are during the day, but the slowest 10% of the reaction times are ten times slower. This spread in reaction times is consistent with one of the most commonly observed effects of continuous work without sleep, the presence of periods of no response or lapses (Wilkinson, 1969). These periods are correlated with periods of lower arousal and even microsleeps, measured by EEG signals (Cajochen et al., 1999a). A number of task characteristics determine the likelihood of lapses occurring. Tasks that are of long duration (i.e., more than 30 minutes), monotonous, and externally paced seem to be more likely to show lapses during sleep deprivation. Conversely, tasks which are considered of short duration, interesting or rewarding, and which are self-paced are less likely to show lapses, although the self-paced task may be done more slowly to maintain the same level of accuracy (Froberg, 1985). It is important to realize that the change in the number of lapses is relative. All tasks show some decline with increasing sleep deprivation but the decline is less for the short duration, interesting, rewarding, self-paced tasks. One aspect of task structure of particular interest is the extent to which short-term memory is required. Tasks requiring the use of short-term memory seem particularly sensitive to sleep deprivation, an observation consistent with the finding of Cajochen et al. (1999b) that the frontal areas of the brain, which are associated with short-term memory, are more susceptible to sleep loss than occipital areas, the site of visual processing.

Even though the effects on performance of working during the circadian night are likely to be limited when some sleep is possible during the day, there are clear advantages in accelerating re-entrainment. In principle, this can be done by correctly timing the exposure to light over the whole twenty-four hours. However, this requires a degree of compliance that many people are unwilling to give. Even if they are willing, it still takes a period of several days before re-entrainment approaches completion. During this time, performance of sensitive tasks may be reduced. But there may be another possibility. The results shown in Figure 5-9 suggest that the onset of sleepiness and the deterioration of the performance measures are closely related to the increase in plasma melatonin. It is well established that exposure to light can rapidly suppress melatonin. The interesting possibility is that exposure to bright light at night could lead to an improvement in performance of some tasks. There is some evidence to support this possibility. French et al. (1990) measured performance on a battery of cognitive tests while working between 18:00 and 06:00 hours and exposed to either 3000 lx or 100 lx illumination. They showed that oral temperatures were elevated by the bright light between 21:30 and 05:30 and that performance on six of the ten cognitive tasks was improved by exposure to bright light, the biggest effect being on serial subtraction and addition. As core body temperature and melatonin secretion are correlated (see Figure 5-9), it can be hypothesized that these changes in performance are due to the suppression of melatonin by the bright light. A similar pattern of results was obtained by Badia et al. (1991), with subjects working at night being exposed to alternating 90-minute periods of bright (>5000 lx) and dim (50 lx) light. Of the six cognitive performance tasks, three showed significantly increased levels of performance during the bright light periods, although all tasks showed deterioration over the night. This fluctuation in performance with light exposure was paralleled by a fluctuation in core body temperature. Boyce et al. (1997) showed a similar pattern for people working three successive nights under the same lighting installation. Out of seven tasks performed, two showed statistically significant improvements when the work was carried out under 2,800 lx from midnight to 08:00 hours or from midnight to 02:30 with a steadily declining illuminance until 08:00 hours. Work over the same time periods at a fixed 200 lx illumination or with an increasing illuminance that only reached 2,800 lx at 05:30 produced lower levels of performance.


5-16

Figure 5-9 The time courses of core body temperature and plasma melatonin concentration over 32 hours. The time courses of measures of sleepiness, i.e., mean eye blinks per 30 seconds, slow eye movements, stage 1 sleep and sleepiness rating, over 32 hours. Time courses of task performance for tasks requiring vigilance, mental arithmetic (cognitive throughput) and short term memory over 32 hours. The dotted line represents the subjects’ habitual bedtime. The error bars are standard errors of the means (After Cajochen et al. 1999).


5-17

Unfortunately, none of these studies measured melatonin concentrations over the working period. However, they all measured core body temperature. All three studies show the expected drop in core body temperature to a minimum around 04:00 followed by an increase at later hours under the low illuminance condition. Further, the negative correlation between core body temperature and melatonin concentration is well established, so the link between performance and the suppression of melatonin by bright light at night and the consequences for task performance are at least plausible if not demonstrated. The other common feature of these results is that some tasks are more sensitive to the effect of sleepiness than others. This may explain why others have failed to show any effect of bright light exposure on performance of a limited range of tests, despite evidence that the phase of the core body temperature had moved further for the subjects exposed to bright light (Campbell, 1995).

To summarize this discussion, the effects of light exposure at night can be divided into those on physiology and those on performance. Bright light can undoubtedly be used to shift the phase of circadian rhythms and to suppress melatonin rapidly. Such exposure can be used to correct the sudden misalignment between the circadian clock and the light-dark cycle caused either by starting or finishing night-shift work or by rapid travel across time zones. It can also be used to increase alertness at night without necessarily shifting the phase of the circadian clock simply by suppressing melatonin. What the consequences are for task performance during the circadian night is much less clear. Some tasks appear to be less sensitive than others to working at night. Specifically, tasks that are interesting or rewarding to the worker, that are self-paced, and that are complex but solvable appear to be more resistant to decline during prolonged work than do tasks which are repetitive, boring, externally paced, and of little interest to the worker. Whether suppression of melatonin and/or the shift in circadian phase produced by bright light exposure are sufficient to overcome the negative effects of prolonged work at night depends on the structure of the task and the context in which it is being performed. This implies that there are no easy answers and no guarantees when it comes to predicting the effect of exposure to light at night and its effects on task performance.

6-1

6 “MESSAGE” ISSUES: CURRENT KNOWLEDGE

6.1 “Message” in Context

Unlike visibility and circadian photobiology, our understanding of how the “message” sent by lighting affects human performance is not built upon a foundation of physiology. Rather, the foundations of the “message” issues are in psychology. Much of the current knowledge about the effect of lighting conditions through “message” is empirical. There is a shortage of over-arching concepts that would enable the various empirical findings to be understood and predictions to be made about others.

The other characteristic of this field is its variability with context and culture. Given that the lighting conditions deliver a “message,” it seems likely that the same lighting conditions can deliver different “messages” in different contexts and to different cultures, in the same way that the same words carry different meanings depending on where and how they are spoken. This implies that any applications of lighting to deliver a “message” will have to be carefully matched to the context in which they are delivered and the culture to which they are addressed.

6.2 Visual Discomfort and Task Performance

One of the more common situations where a lighting installation can be said to be delivering a “message” is where people complain about visual discomfort. The over-arching concept here is that people need to extract information from the visual environment, and lighting conditions can help or hinder this process. Visual discomfort occurs when there is difficulty in extracting information from the visual environment. The features of the visual environment that can cause visual discomfort are:

Visual task difficulty: Any visual task that is close to threshold contains information that is difficult to extract. The usual reaction to a high level of visual difficulty is to bring the task closer to increase its visual size. As the task is brought closer, the accommodation and convergence mechanisms of the eye adjust to keep the retinal image sharp. This adjustment can lead to muscle fatigue and hence symptoms of visual discomfort.

Under- and over-stimulation: Discomfort can also occur either when there is no information to be extracted or when there is an excessive amount of repetitive information. Examples of no information occur when driving in fog or in a “whiteout” snowstorm. In both cases, the visual system is searching for information which is hidden but which may appear suddenly and require a rapid response. The stress experienced while driving in these conditions is a common experience. As for overstimulation, the important point is not the total amount of visual

“Message” Issues: Current Knowledge

6-2

information, but rather the presence of large areas of the same spatial or temporal frequency. Wilkins (1991) has associated the presence of large areas of specific spatial frequencies in printed text with the occurrence of headaches, migraines, and reading difficulties.

Distraction: The visual system employs a large peripheral field that detects the presence of objects that are then examined using the small, high-resolution fovea. For this system to work, objects in the peripheral field that are bright, moving, or flickering automatically attract attention. If, upon examination, these bright, moving, or flickering objects are repetitive and prove to be of no interest, they become sources of distraction because their attention-gathering power is not diminished after one examination. Ignoring objects that automatically attract attention is stressful and can lead to symptoms of visual discomfort.

Perceptual confusion: The visual environment consists of a pattern of contrasts, developed from the pattern of luminances produced by the reflectances of the surfaces in the field of view and the distribution of illuminance on those surfaces. Perceptual confusion occurs when a pattern of luminances that is solely related to the illuminance distribution conflicts with the pattern of luminances associated with the reflectances of the surfaces.

Visual task difficulty, under- and over-stimulation, distraction, and perceptual confusion can all be influenced by the lighting conditions. The aspects of lighting that cause such effects, and hence can cause visual discomfort, are many and various. Insufficient light for the performance of a task is an obvious factor, but so are flicker, glare, shadows, and veiling reflections.

Flicker: A lighting installation that produces visible flicker will be almost universally disliked, unless it is being used for entertainment. Individual differences, and the fact that electrical signals associated with flicker can be measured by electrical potentials in the retina, even when there is no subjective report of visible flicker (Berman et al., 1991), imply that the frequency of light output fluctuations from discharge lamps may be too low to ensure comfort. High-frequency control gear for discharge lamps and/or the mixing of light from lamps powered from different phases of the electricity supply can minimize the probability of discomfort. (The same approach can be used to diminish any stroboscopic illusions as well.) The use of high-frequency control gear has been associated with a reduction in the prevalence of headaches (Wilkins et al., 1989).

Glare: Glare occurs in two ways. First, it is possible to have too much light. Too much light produces a simple photophobic response, in which the observer narrows his eyes, blinks, or looks away. Too much light is rare indoors but is common in full sunlight. Second, glare occurs when the range of luminances in a visual environment is too large. Glare of this sort can have two effects: an elevation of threshold visual performance and a feeling of discomfort. Glare that increases threshold visual performance is called disability glare and is discussed in Section 4.4.

As for discomfort glare, this, by definition, does not cause any increase in threshold visual performance but does cause discomfort. There are many different national systems for predicting the magnitude of discomfort glare produced by interior lighting installations (CIBSE, 1994; IESNA, 2000). The precision with which they predict an individual’s sense of discomfort is low (Stone and Harker 1973).


6-3

Shadows: Shadows are cast when light coming from a particular direction is intercepted by an opaque object. If the object is big enough, the effect is to reduce the illuminance over a large area. This is typically the problem in industrial lighting, where large pieces of machinery cast shadows in adjacent areas. If the object is small, the shadow can be cast over a meaningful area that in turn can cause perceptual confusion, particularly if the shadow moves. An example of this is the shadow of a hand cast on a blueprint. This problem can be reduced by increasing the interreflected light in the space or by providing local lighting whose position can be adjusted.

Although shadows can cause visual discomfort, it should be noted that they are also an essential element in revealing the form of three-dimensional objects. Techniques of display lighting are based around the idea of creating highlights and shadows to change the perceived form of the object being displayed.

Veiling reflections: Veiling reflections occur when a source of high luminance, usually a luminaire or a window, is reflected from a specularly reflecting surface, such as a glossy printed page or a VDT screen (see Section 4.4). The luminance of the reflected image changes the luminance contrast of the printed text or the VDT display. The extent to which this changes visual performance can be estimated using the RVP model, but the extent to which it causes discomfort is different. Bjorset and Fredericksen (1979) have shown that a 20% reduction in luminance contrast is the limit of what is acceptable, regardless of the luminance contrast without veiling reflections. Like shadows, veiling reflections can also be used positively, but they are then called highlights. Display lighting of specularly reflecting objects is necessarily concerned with producing highlights to reveal the specular nature of the surface.

Many lighting guides and codes give advice on how to avoid flicker, glare, shadows, and veiling reflections and how to determine which of these aspects of the visual environment are likely to be important in different applications (CIBSE, 1994; IESNA, 2000).

There can be little doubt that aspects of lighting that cause visual discomfort because they make the task more difficult to perform by changing the visibility of the task (e.g., disability glare) can affect task performance. The interesting question is whether task performance can be affected by lighting conditions that cause visual discomfort but do not change the visibility of the task. Figure 6-1 shows the mean detection speed for finding a number from many laid out at random on a table, and the percentage of people who considered the lighting good (Muck and Bodmann, 1961). As might be expected, increasing the illuminance on the table from 50 lx increases mean detection speed and the percentage who consider the lighting good. However, as the illuminance exceeds 2000 lx, the percentage considering the lighting good declines even though the mean detection speed continues to increase. This result indicates that it is possible to increase visual discomfort without producing a decline in task performance. Stone and Groves (1968) show a similar result. They measured performance on a Landolt ring chart while the subjects were exposed to various levels of discomfort glare. No statistically significant changes in task performance were found, even for what was predicted to be just intolerable glare.


6-4

Figure 6-1 Mean detection speeds for locating a specified number among others at different illuminances and the percentage of subjects doing the task who considered the lighting "good" at each illuminance (After Muck and Bodmann, 1961).

It can be concluded that there is no necessary link between the occurrence of visual discomfort that does not affect task visibility and a decline in visual performance. The key word here is “necessary.” Such a link may occur but whether it does or not will depend on the motivation of the worker and the duration of the task. Figure 2-1 suggests that lighting conditions that are considered uncomfortable may influence task performance by changing mood and motivation even when they have no effect on the visibility of the task. In a laboratory experiment, people are usually highly motivated to succeed and so will persevere with the task even in the presence of visual discomfort. It is doubtful that the same degree of motivation occurs everywhere outside the laboratory. As for task duration, lighting conditions that are marginally uncomfortable may be tolerated if the task only has to be done for a short time. However, if prolonged work leads to ever-stronger symptoms of visual discomfort, eventually task performance is likely to be impacted.


6-5

6.3 Perception of Spaces and Task Performance

Fortunately, abhorrence of occupant complaints ensures that most lighting installations are designed so that they do not produce visual discomfort and that they ensure adequate visual performance. Lighting installations can, however, produce very different lighting conditions in a space. The question to be considered here is whether lighting installations that produce different perceptions of a space, can produce a change in task performance.

The first point to establish is whether different lighting installations in the same setting can produce different perceptions of the setting. Flynn et al. (1973, 1979) had observers experience a conference room lit in six different ways. The observers’ responses were collected on thirty-four rating scales. Factor analysis showed that evaluations were made on five independent dimensions (evaluative, perceptual clarity, spatial complexity, spaciousness, and formality), although only three of these dimensions showed much separation between the lighting conditions (the evaluative, perceptual clarity, and spaciousness dimensions). An examination of the different rating scales loaded on each dimension showed that the installations that illuminated the walls of the room as well as the area around the conference table were preferred. The perceptual clarity dimension was related to the illuminances on the table, regardless of how it was produced. The spaciousness dimension showed a more variable pattern of responses, but it seemed to be based on patterns of light that provided more or less illumination on walls. Those installations that illuminated the walls were perceived as larger and more spacious. Such results support Flynn’s concept that lighting provides a number of cues that people use to interpret or “make sense” of a space and that these cues are at least partly independent of the room that is being experienced (Flynn, 1977).

Hawkes et al. (1979) carried out an evaluation of a small rectangular office lit in eighteen different ways. The illuminance on the desk was always 500 lx, but the distribution of light in the rest of the office varied greatly. Factor analysis of the data collected revealed two independent dimensions, brightness and interest. The brightness dimension was clearly related to the amount of light in the room, while the interest dimension appeared to be primarily related to how that light was distributed.

One criticism of the studies of Flynn et al. and Hawkes et al. is that the number of subjects was too small for a robust factor analysis, given the number of scales used. Veitch and Newsham (1998a) overcame this problem by having 292 observers rate the appearance of an open-plan office lit by one of nine different lighting installations. The evaluation was made with the subject sitting in a cubicle in the office after having worked on a computer and on paper for about 5 hours. Factor analysis of the ratings revealed three dimensions named brightness, visual attraction, and complexity. However, these three factors in total explained only 46% of the variance, probably because the differences between the lighting installations were smaller than in the earlier studies. Examination of how the different lighting installations related to each of these dimensions showed that the direct lighting system produced a greater impression of brightness than the indirect lighting. Also, lighting installations using parabolic luminaires were considered less complex and less bright than installations using lensed luminaires, and task-ambient lighting installations were considered less bright than direct lighting installations alone.

These three sets of studies, separated by almost thirty years and all using functional interiors where real work was performed, show some interesting similarities and differences in their factor


6-6

structures. All show that brightness is one dimension on which people evaluate a visual environment. Another appears to be concerned with variety, meaning non-uniformity in light distribution away from the work surface. Beyond these dimensions, there is no agreement. Flynn et al. (1973) have a dimension of spaciousness. Veitch and Newsham (1998a) have a dimension of complexity. Hawkes et al. (1979) do not have a third dimension. The fact that brightness is a consistent dimension on which the visual environment is evaluated should not be too surprising, given that brightness is related to the amount of light in the space and the amount of light in the space is the primary characteristic of the lighting installation. The importance of brightness is also consistent with Kaplan’s information-processing model of environmental appraisal (Kaplan, 1987). This model rests on the idea that the information available in a scene is central to our appraisal of it, and brightness is a marker for how much information can be revealed. As for variety, there is no doubt that humans need some variety, but not too much. Monotonous environmental conditions, if taken to extremes, lead to cognitive breakdown (Corso, 1987). On the other hand, too much variation in environmental conditions can lead to confusion and distress. It is also true that variety, unlike brightness, can be introduced through aspects of the visual environment other than lighting. For example, variety can be introduced into a room by changing the decor, by changing the furnishings, and by changing the people in it. Using a non-uniform light distribution is but one way of introducing variety and may not be the most potent. This applies even more strongly to higher-level perceptions, such as whether the interior is friendly or formal or patriotic. What this suggests is that any attempt to consistently link specific lighting conditions to higher-level perceptions across many different contexts will meet with little success. There are too many other factors beyond the photometric conditions that influence higher-order perceptions, and those factors will vary with different contexts for different cultures. It may still be possible to make a link between lighting conditions and higher-order perceptions for specific contexts in specific cultures, but how valuable such a finding would be would depend on how specific the conclusion was. There is little merit in determining what type of lighting makes a meeting room appear formal if that perception can be altered dramatically by changing the furniture. What seems to be essential is that lighting should “make sense”, and for lighting to make sense it must be orchestrated with the furnishing and the architecture.

Whatever their limitations, the above studies do demonstrate that different lighting installations can produce different perceptions of the same space. So the question now becomes, to what extent do these different perceptions produce a change in task performance? Harvey et al. (1984) had people perform a task involving checking lists of numbers printed on paper against a similar list displayed on a VDT, under direct and indirect lighting. After doing this for one hour, the subjects were asked to give ratings for various aspects of the quality of the direct and the two indirect lighting systems. There was no difference in performance of the task, but there was a clear difference in preference.

Veitch and Newsham (1998a) measured the performance of an array of tasks done by a large number of individuals over a full working day, under nine different lighting installations classified according to power density and perceived lighting quality. A complex pattern of interactions was obtained, where some tasks were performed better under one lighting installation and others under another installation. Most of the changes in task performance were small (1 - 3% of variance explained). Interestingly, the strongest effect any visual condition had on the performance of any task was for the change of the VDT screen polarity on a typing and a proofreading task. Changing the screen type from negative to positive polarity explained 7 - 9% of the variance in task performance, probably because the higher background luminance of the


6-7

positive polarity screen made the display less subject to interference from reflected images of the rest of the room. This is a direct effect on task visibility.

One limitation of these two studies is that the different lighting installations produced different illuminances and different light distributions on the paper tasks and on the VDT monitor which may have changed the visibility of the tasks. These differences make it impossible to determine the extent to which changes in visibility or changes in the perception of the space are responsible for any changes in task performance. A study by Eklund et al. (1999) avoided this problem. They had temporary office workers work for eight hours on a data-entry task in three private, windowless offices, all with the same decor and furniture. Three different lighting installations were used, one in each office. All three lighting installations provided a similar illuminance on the task without veiling reflections or disability glare, so for the same task they provided similar task visibility. However, the three lighting installations were very different in light distribution over the room, ranging from very uniform indirect lighting to very concentrated overhead lighting. There was no difference in performance of the task under the three lighting installations. Somewhat more surprisingly, there was no difference in the workers’ perceptions of the three lighting installations, despite the opinions of lighting experts that the lighting installations clearly differed in quality. However, changes in point size of the material to be entered, and hence in task visibility, did produce statistically significant changes in task performance.

Eklund et al. (2000) had people perform the same data entry task for four hours, but this time they always had the same lighting installation with the same light distribution, but with two different levels of decor. In one condition, the office was bare and achromatic. In the other, some colorful decor was added to the room. Again, there was no difference in task performance between the two decors, although the subjects did perceive the chromatic decor as more colorful, attractive, and interesting. Again, changes in the task visibility created by changing print size, luminance contrast, or illuminance on the task did produce statistically significant changes in task performance.

These studies clearly demonstrate that changes in task visibility can be reliably expected to change the performance of visual tasks but the effects of differences in perception of lighting are much less certain. There are several reasons why this might be so. One is that people naive in lighting may be much less sensitive to lighting conditions that do not affect task visibility or visual comfort than lighting experts, in the same way that wine connoisseurs are more sensitive to wine than people who do not drink wine frequently. If people are not sensitive to changes in lighting conditions that do not affect task visibility or visual comfort, then it is unlikely that changes in such conditions will change their mood and hence their motivation to perform the task. Another is that, given a long enough exposure to lighting conditions that do not affect task visibility or visual comfort, people habituate to them. Therefore, the conditions become less and less significant to them and hence less and less likely to change their mood. Another is that the range of lighting conditions may not have been extreme enough, although the three lighting installations used in Eklund et al. (1999) covered what might be considered the extremes of current lighting practice. Finally, the nature of the tasks may have been too strongly visual. It might be that tasks that involve less visual and more cognitive and judgmental elements would be more sensitive to changes in mood than highly visual tasks, the performance of which would be dominated by the task visibility. It might also be important that the data-entry task focussed attention on a small area, where all the information needed to do the task was available in that small area. The rest of the room contained no information relevant to the task, so the lighting of


6-8

the rest of the room was irrelevant to the worker. There seems little doubt that for tasks in which information is spread over many different parts of the room, the light distribution over the whole room will matter. In such a case, light distribution affects the ability to extract information to do the task, i.e. it affects task visibility.

Despite all this speculation, one conclusion is clear: Only lighting conditions that affect task visibility are certain to affect task performance. Lighting conditions that do not affect task visibility may or may not affect performance, but that will depend on the specific situation, because many other factors besides lighting can affect mood and motivation. It is the balance between the impact of these factors on mood and motivation and the impact of the lighting conditions on mood and motivation that determine whether lighting conditions that do not affect task visibility will have a significant effect on human performance.

6.4 Lighting and Mood

Given the apparent difficulty in demonstrating an effect of lighting conditions that do not affect task visibility or cause visual discomfort on task performance, it is appropriate to ask whether lighting can have an effect on mood. Mood is defined as “the core feelings of a person’s subjective state at any given moment and is not necessarily about anything” (Russell and Snodgrass, 1987).

A number of attempts have been made to determine the effects of illuminance and lamp spectrum on positive or negative mood. Belcher and Kluczny (1987a) measured the shift in mood following exposure to two different illuminances (2150 lx and 215 lx). It was found that women’s positive mood tended to deteriorate over time at the higher illuminance but not at the lower illuminance, while men did not show any statistically significant difference over time at either illuminance. Baron et al. (1992) found no sex difference in the effect of illuminance and correlated color temperature (CCT) on mood. Both sexes were found to be calmer and more awake under 150 lx than 1500 lx provided by lamps with a warm CCT. Knez (1995) found that for lamps with high CRI (CRI = 95), men’s negative mood decreased over time under a cool CCT lamp but women’s mood changed most over time under a warm CCT lamp. When exposed to lamps with a low CRI (CRI = 51 to 58), an interaction between illuminances and CCT was found. A good mood was best maintained over time with a cool CCT at 300 lx or with a warmer CCT at 1,500 lx. McCloughan et al. (2000) examined mood for two illuminances (260 lx and 810 lx) provided by lamps with the same CRI but different CCTs (3,000 K and 4,000 K). In this study mood was broken down into components of anxiety, depression, hostility, positive affect, and sensation-seeking. After five minutes exposure, it was found that the lower illuminances were associated with higher levels of sensation-seeking, while hostility was higher under the warmer CCT than under the cooler CCT. After a further 30 minutes, a complex series of interactions between illuminance, CCT, and gender on the negative components of mood were found.

Given the widely varying nature of these findings, one might be tempted to decide that the interaction between lighting conditions and mood is meaningless. However, before reaching such a conclusion it should be noted that lighting conditions that are inadequate for task performance or that cause other forms of visual discomfort can be expected to increase negative mood. Eklund et al. (2000) showed that when asked to do a data-entry task with a wide range of visual


6-9

difficulty, lower illuminances led to less positive affect and more negative affect. This is to be expected, so the question of interest is whether lighting conditions that are adequate for any task to be performed and that do not cause visual discomfort can affect mood. Is it realistic or naive to suppose that simple photometric conditions, such as different illuminances, CCTs and CRIs, can have consistent effects on mood? An answer to this question can be obtained by considering the underlying factors that drive mood. Purcell and Nasar (1992) suggest that mood effects are related to discrepancies between expectations about a space and the actual space. When the discrepancies are minor, there is a positive hedonic response, but as the discrepancy increases, there is an increasingly negative response. The expectations of people in the above studies are unclear. The study of the effects of lighting on expectations and mood deserves further attention. However, assuming expectations matter, such studies must be done in settings about which people have known expectations, hopefully using a wide range of lighting conditions.

6.6 Lighting and Behavior

Of course, using lighting to influence mood is basically a means to an end, that of influencing behavior towards a desired outcome. It is undeniable that lighting conditions can influence behavior. At the simplest level, people will change their behavior to compensate for poor lighting conditions. Smith and Rea (1982) observed that while reading materials with poor visibility, some people moved closer to the task to increase its visual size. Rea (1983) found that subjects given the freedom to change their lighting conditions by moving a desk lamp did so in a way that reduced veiling reflections. Rea et al. (1985) examined the influence of lighting geometry, task background luminance, task luminance contrast, and task specularity on performance of the numerical verification task. Again, they found that when these variables combined to produce poor task visibility, people altered their posture so as to increase the visibility of the task. While low visibility can cause people to modify their behavior, whether such modifications affect task performance depends on whether people can positively change the visibility of the task or the capabilities of the visual system (e.g., through reading glasses).

The effects of lighting on behavior are more extensive than simply overcoming problems of poor visibility. This is apparent when considering how lighting can be used to guide movement and direct attention. Taylor and Socov (1974) explored the effect of lighting on peoples’ choice of passageway at an exhibition. They found that more people chose the passageway lit to a higher illuminance. Similar findings have been reported in an investigation of the effects of spatial distribution of light on seat choice and orientation in a cafeteria (Flynn et al., 1973). The measurements showed that people selected seats facing bright areas. When the lighting was changed to highlight a different surface, patterns of seat selection and orientation changed to face the new bright area. The effects of wall lighting on desk selection have also been observed (Yorks and Ginthner, 1987). Subjects entered a room, and sat at one of three desks to complete a series of questionnaires. Desks were located next to the door, in the middle of the room and at the far side of the room opposite the door. When the wall opposite the door was illuminated, most subjects crossed the room and sat at the desk located next to that wall. When that wall was not illuminated, most subjects sat at the desk located next to the door.

Why people behave in this way in relation to light distribution is a matter of speculation. Whether it is because they expect to see better where there is more light, or because they


6-10

interpret the low light level as meaning that that part of the space is out of use, or because of some deeper conditioned response remains to be determined.

There is also the possibility that the message delivered by the lighting may influence behavior unrelated to vision. For example, a significant reduction in sound level was found in a school hallway when the illuminance was low (Sanders et al., 1974). Conversely, Veitch and Kaye (1988) found that the conversation among female college students engaged in a mock job applicant evaluation task was louder under low illuminances. Further, Carr and Dabbs (1974) examined the effects of lighting, distance, and topic intimacy on verbal and visual behavior. They found that as the intimacy of the topic increased, subjects tended to look less at the questioner and to say less in response to questions. When dim lighting was used this response was enhanced. The subjects waited much longer before answering the questions and made shorter eye contact with the questioner. The subjects considered the dim lighting very intimate but inappropriate for the discussion of intimate subjects with a stranger. This can be contrasted with the results of Butler and Biner (1987), who established preferences for illuminances for different common activities. They found that the presence of a romantic partner consistently lowered subjects’ preferred illuminances. Gifford (1988) studied the effects of lighting and decor on interpersonal communication between friends. He found that brighter lighting stimulated more general conversation and more intimate conversation and that over time lower illuminances dampened both general and intimate communication.

If they do nothing else, these conflicting results demonstrate that behavior is not a matter of an autonomic response to light stimulation alone. The chosen behavior depends on the context in which the exposure occurs. This is evident in the study of Page and Moss (1976). They had subjects teach another person a list of pairs of nonsense syllables. For every error made while learning, the “teacher” could administer an electric shock of varying degree to the “learner.” The results showed that the “teacher” tended to administer more intense shocks of longer duration to the “learner” in dimly lit settings. The explanation for this behavior is a matter of speculation. It could be a matter of physiology (Hartung, 1978) or a matter of psychology, the dim lighting conditions providing a cloak of anonymity to the “teacher.”

Probably the most interesting aspect of the effect of lighting conditions on behavior from the commercial viewpoint is whether lighting conditions can be used to increase sales of a product. Certainly, display lighting is designed with this aim in mind. Providing a highlight on an object can be used to draw attention to it. LaGiusa and Perney (1974) used lists of words displayed at the front of classrooms. Supplementary lighting was used to highlight the word lists in one condition of the experiment, but not in the control condition, and the students’ behavior was observed. Significantly more inattentive behaviors were coded in the control condition than when the word lists were highlighted.

Further, depending on how the object is lit, the observer’s perception of the object can be changed. Mangum (1998) examined the perception of three museum objects under six different lighting installations, all subject to the constraint that no illuminance on the object could exceed 50 lx. The perception of the object was identified by the use of a balanced (positive and negative) word list from which observers had to select all the words that applied to their perception of the object. When viewing a doll, clothed in materials that varied in texture, color, and reflection properties, under diffuse illumination, the words most frequently chosen to describe the observers’ perceptions were “unattractive,” “unpleasant,” “obscured,” “veiled,” “bland,”


6-11

“boring,” “mundane,” and “ordinary.” When the same object was lit with directional lighting, using key-light, side-light, and back-light, the words most frequently used were “interesting,” “attractive,” “eye-catching,” “clear,” “pleasant,” “revealing,” “dramatic,” and “spectacular.” Similar results were found for a plain vase, although other forms of directional lighting produced more dramatic perceptions for this object. Clearly, changing the lighting can change the higher-order perceptions of an object. Whether it does so in any particular case will depend on the nature of the object and the extent to which the form, texture, color, and reflection properties of the object are modified by the change in lighting.

Both claiming attention and changing perceptions of objects are desirable in order to sell, but these effects do not guarantee a sale. Areni and Kim (1994) examined the influence of what they called bright and soft lighting on the behavior of people in a wine store. They found that, under bright lighting, people examined and handled the merchandise on the shelves at eye level more, although there was no overall change in sales. Cuttle and Brandston (1995) measured the impact of relighting on sales in two furniture galleries. The new lighting provided higher illuminances and a more even light distribution. Over five months, sales in one gallery increased by 35%, but there was no consistent change in sales in the other gallery. Bear in mind, however, that a furniture store is characterized by small numbers of people spending large sums of money, infrequently, to buy goods with which they will live for many years. In another study, Boyce et al. (1996) tracked sales at a supermarket, a business characterized by large numbers of people spending modest sums of money, regularly, to buy goods which will be rapidly consumed. In this case, the refurbishment of the in-store bakery, a refurbishment that involved changes in the layout of the sales area and in the display methods as well as changes in the electric lighting and the introduction of daylight through skylights, produced statistically significant, sustained increases in the dollar value of sales. Changes in other parts of the store, which were lighting changes alone, did not produce any statistically significant changes in number or dollar value of sales. On a larger scale, Heschong-Mahone (1999a) showed that sales were significantly higher in supermarkets with skylights relative to a set of matched supermarkets in the same chain that did not have skylights. The presence of skylights can be expected to increase the illuminances in the store, to enhance color rendering, and to provide some meaningful variation in light distribution over the day, as well as requiring a generally higher ceiling.

There can be little doubt that lighting conditions can have an effect on overt behavior. However, the problem in predicting behavior is that it is influenced by many factors besides lighting. Lighting conditions can clearly change some of the factors that influence the behavior we choose, such as the level of arousal and the perception of a product but, for many situations, the other factors that determine behavior will overwhelm any effect of lighting conditions. How much influence any lighting installation has on behavior is probably a matter of how extreme the lighting conditions are and how closely the desired behavior is linked to visibility. Where visibility is important, lighting conditions are likely to have a clear effect on behavior. Where visibility is adequate and many other factors are influencing behavior, then the effect of lighting conditions will be difficult to predict because the chosen behavior will be dependent on the specific situation.

7-1

7 MODIFYING FACTORS

The effects of lighting conditions on task performance discussed above have typically been studied using young people, with normal color vision, working for short periods, while in good health. If any of these factors is changed, human performance is likely to deteriorate.

7.1 Aging Vision

The factors that determine the operating state of the visual system are the amount of light that reaches the retina and the wavelengths that make up the light. The factors that determine the clarity of the retinal image are the ability to focus the image of the external object on the retina, the extent to which light is forward scattered as it passes through the eye, and the presence of stray light. Virtually all these characteristics change with age (Werner et al., 1990).

The optical power of the eye is determined by the curvature of the cornea and the thickness of the lens. If there is a mismatch between the distance of the retina from the lens and the total optical power of the eye, the image of the outside world will not be in focus on the retina, so the resulting retinal image will be blurred. Blur has been shown to be a potent cause of reduced visual performance (Johnson and Casson, 1995). The inability to bring an image of a nearby object to focus on the retina with the unaided eye increases with age and is usually overcome by altering the optical power of the eye, using spectacles or contact lenses.

The optical factors affecting the amount of light reaching the retina are the pupil size and the spectral absorption of the components of the eye. The ratio of maximum to minimum pupil area decreases with age, the maximum decreasing more than the minimum (Weale, 1992). This average reduction in pupil area has two opposing effects. It reduces the amount of light reaching the retina, due to smaller pupil area and because of greater absorption of light passing through the thickest part of the lens. A small pupil also provides a greater depth of focus and reduced optical aberrations. As for the spectral absorption of the eye, as light passes through the optical components of the eye the transmittance of light is reduced for all visible wavelengths, but particularly for the short wavelengths (Boettner and Wolter, 1962). As the eye ages, the amount of light absorbed increases, particularly at short wavelengths (Lerman, 1980).

In addition to absorbing light, transmission through the optical components of the eye scatters light. This is important because scattered light degrades the retinal image, and providing more light does not help. The scattering is primarily large-particle scattering, so is independent of wavelength. Measurements have shown that about 30% of scattering occurs at the cornea (Vos and Boogaard, 1963), with most of the rest occurring at the lens, vitreous humor, and fundus (Boettner and Wolter, 1962). The amount of scatter increases with age, due mainly to changes in the lens (Wolf and Gardiner, 1965). Scattered light degrades the retinal image by reducing the

Modifying Factors

7-2

difference in luminance on either side of an edge, thereby reducing the contrast of the edge. Lower edge contrast reduces acuity, and, more generally, the visibility of high spatial frequencies.

Straylight is characterized by a homogenous distribution of luminance over the whole retinal image. Straylight within the eye is produced by back-reflection from the components of the eye, transmittance through the eye wall, and fluorescence in the lens of the eye (Boynton and Clarke, 1964; van den Berg et al., 1991; van den Berg, 1993). Straylight matters because it falls uniformly across the retinal image, thereby reducing the contrast of all parts of the image. The amount of straylight generated by these causes increases with age. This is particularly noticeable for lens fluorescence. The effect of lens fluorescence is negligible in young eyes. But as aging continues and the luminance of straylight due to fluorescence increases and fluorphores with emission wavelengths in the most sensitive part of the visual spectrum emerge, the effect of lens fluorescence will be noticeable as haze over the visual scene (Jacobs and Krohn, 1976; Weale 1985).

It is important to appreciate that the neural characteristics of the visual system also change in later years. Specifically, the number of ganglion cells in the retina (Balazsi et al., 1984), the number of neurons in the visual cortex (Devaney and Johnson, 1980), and the sensitivity of foveal color mechanisms also decline with age (Werner and Steele, 1988). These changes are important because they imply that the magnitude of improvement in visual capabilities that can be achieved by changing the lighting is inevitably limited. Returning the optical characteristics of the eye to what they were before aging will not restore vision to its pristine state.

The functional consequences of these age-related changes in the optical and neural characteristics are reduced visual field size (Weale, 1992), reduced sensitivity to movement, (Weale, 1963), reduced visual acuity (Adrian, 1995), increased threshold contrast (Blackwell and Blackwell, 1971), reduced contrast sensitivity (Owsley et al., 1983), increased sensitivity to glare (Vos, 1995), and poorer color discrimination (Knoblauch et al., 1987). Figure 7-1 shows the contrast sensitivity function for people of different ages (McGrath and Morrison, 1981). Examination shows that the effect of age is to decrease the maximum contrast sensitivity and decrease the range of spatial frequencies over which resolution can occur. Figure 7-2 shows the average distribution of errors made on the Farnsworth-Munsell 100 hue test as a function of illuminance and age (Knoblauch et al, 1987). Zero error is indicated by a smooth circle. As the number of errors increases, the circle becomes larger and more ragged. The distance from the center point in a given direction is a measure of the number of errors in color discrimination made at a particular hue. Examination of Figure 7-2 suggests that, in old age, errors are most common on the vertical axis, which is consistent with a reduced response from the short-wavelength cones. A similar response is obtained from people of any age with tritanopia, i.e., defective color vision associated with the absence of short-wavelength cones.

Modifying Factors

7-3

Figure 7-1 The contrast sensitivity functions for four observers of different ages. Open circles are for near viewing distance. Filled circles are for far viewing distance (After McGrath and Morrison, 1981)

Modifying Factors

7-4

Figure 7-2 The average distribution of errors on the Farnsworth-Munsell 100 Hue test as a function of illuminance and age (After Knoblauch et al., 1987)

These changes in visual function are measures of threshold performance. Others have measured the effect of age on suprathreshold task performance. Weston (1949) measured the change over five years in performance of the Landolt ring task for a range of age groups. Some deterioration was found to take place for all age groups. However, the deterioration was virtually independent of age for a high contrast Landolt ring of gap size 4.5-min arc, but was greater for older age groups for a smaller 1.5-min arc ring. The finding that the effects of age are greater as the task gets closer to threshold is to be expected. It is consistent with the work of Akutsu et al. (1991) who measured reading speed for different print sizes, all at high contrast. Older people showed a reduction in reading speed relative to that of young subjects when reading either very small or very large print. Figure 4-3 shows performance on a proof-reading task with two different levels of print quality, by a group of young and old subjects, plotted against illuminance (Smith and Rea, 1978). As would be expected, increasing illuminance improves performance for both young and old, but the improvement is greater for the older group. However, it should be noted that even with the highest illuminance used, about 5000 lx, the performance of the old group remains below that of the young group. This failure to fully compensate for the ravages of age by increasing the illuminance is due to the fact that increasing the illuminance can only compensate for the increased absorption of light. It cannot compensate for the increased scattering of light nor for the generally slower cognitive responses of the elderly. Nonetheless, the fact that older people benefit from more light is worth noting. Certainly, it is consistent with the observation of

Modifying Factors

7-5

Werner et al. (1990) that “many of the characteristics of the senescent visual system are similar to those of the younger visual system operating at a lower light level.”

But simply providing more light to compensate for the additional light absorption may not be enough. Knowledge of the characteristics of the aging eye suggest that light has to be provided in such a way that both disability and discomfort glare are minimized and reflections that reduce luminance contrast are avoided. There is also the question of light spectrum. Berman et al. (1994) have shown that, at the same luminance, a light source that stimulates the rod photoreceptors to a greater degree produces better visual performance on a difficult achromatic task by a group of people in their sixties, than a light source that does not stimulate the rod photoreceptors so much, presumably because of the change in pupil size. For making discriminations between colors, there can be little doubt that using a lamp with a higher CIE General Color Rendering Index would make discriminations easier because, generally, such lamps give surface colors greater saturation and therefore position them further apart in color space. However, as regards color appearance, Werner and Kraft (1995) argue that there is little to be gained by adjusting the light spectrum to compensate for the additional absorption of short-wavelength light in the lens at older ages. This is because the visual system makes adaptive neural changes over the years to maintain color constancy over the life span. From the point of view of compensating for the deterioration of vision over time, Werner and Kraft recommend increasing the illuminance rather than adjusting the light spectrum. The value of such an increase in everyday life can be seen from the work of Kosnik et al. (1988), who asked healthy adults in the age range 18 to 100 years about everyday visual problems. People rated the frequency with which they experienced difficulty in such routine activities as reading, recognizing objects, picking out a face in a crowd, seeing in dimly lit environments, seeing moving objects, and so on. Compared to their younger counterparts, the older adults reported they took longer carrying out visual tasks and had more trouble with glare, dim illumination, and near visual tasks.

Finally, it is important to mention that aging effects are also apparent in the circadian system. The most obvious change with increasing age is an attenuation of amplitude of the change in melatonin concentration from day to night, although an advance of phase, a shortening of period, and a desynchronization of rhythms have also been found. Why these changes occur is not clear, but age-related changes in the optics and retina of the eye, and in the suprachiasmatic neuclei and the pineal gland are probably involved, as well as behavioral changes such as less physical activity and less exposure to light. Changes in circadian rhythms are frequently associated with a worsening of sleep quality and a consequent decrease in daytime alertness and decline in cognitive performance (Myers and Badia, 1995). Using light to improve sleep quality might be a useful approach to improving the performance of older workers.

7.2 Defective Color Vision

Defective color vision is characterized by abnormal color matching or color confusions. Colors that look different to people with normal color vision may look the same to people with defective color vision. People with defective color vision are classified into three broad classes: monochromats, dichromats, and anomalous trichromats, according to the number of photoreceptors present and the nature of the photopigments present in the photoreceptors. Monochromats, although very rare, occur in two forms: rod monochromats, where there are no cone photopigments, and cone monochromats, where there is only one cone photoreceptor.

Modifying Factors

7-6

Neither type of monochromat has any color vision. They are truly color-blind and see only differences in brightness. Both dichromats and anomalous trichromats have some perception of color, although not the same perception as people with normal color vision. Dichromats have two cone photopigments. They see a more limited range of colors than people with normal color vision and may have a different spectral sensitivity to light, depending on which cone photoreceptor is missing. Dichromats with the long-wavelength cone missing are called protanopes. Dichromats with the medium-wavelength cone missing are called deuteranopes, while dichromats with short-wavelength cones missing are called tritanopes. Anomalous trichromats have all three cone photopigments present, but one of the cones contains a photopigment that does not have the usual spectral sensitivity. Anomalous trichromats who have a defective long-wavelength photopigment are called protonamalous. Anomalous trichromats who have a defective medium-wavelength photopigment are called deuteranomalous, while anomalous trichromats who have a defective short-wavelength photopigment are called tritanomalous. The color vision of anomalous trichromats can vary widely from almost as bad as a dichromat to little different from someone with normal color vision.

Overall, about 8% of males and 0.5% of females have some form of defective color vision; about half of these are deuteranomalous. Steward and Cole (1989) surveyed people with defective color vision and found that 75% of such people have some trouble with everyday tasks, such as selecting colored merchandise and judging the ripeness of fruit. Mistakes in such activities may be embarrassing and cause inconvenience, but they are hardly life-threatening. Somewhat more serious is work where color-coding is used to draw attention to important information. Cole and McDonald (1988) examined the performance of color-defective observers on a task that required information acquisition from the video display of an electronic flight instrument system. Usually color-coding of information decreases response times and reduces errors. However, people with defective color vision showed slower response times and higher error rates than color-normal observers when color-coding was used, but produced the same level of performance as color-normal observers for monochromatic displays. The seriousness of slower response times and more errors depends on their consequences. Unfortunately, there are some occupations where mistakes in identifying a color can cause a dangerous situation, such as pilots, railway engineers, and electricians. In these occupations, color is used to convey information relevant to the safety of the activity. People with defective color vision are necessarily excluded from such occupations (Vingrys and Cole, 1988). Another class of occupations that is difficult for people with defective color vision is that where color discrimination is an inherent part of the task and a failure to make fine color discriminations will lead to a poor quality product, for example, the paint and printing industries. Color-defective people will find it difficult to gain employment in some parts of such industries.

There is little that can be done to overcome the effects of defective color vision in general, although filters can sometimes be used to enhance specific color differences (Birch, 1993). The most general advice to make life easier for people with color-defective vision is to avoid low reflectance colors and color combinations that such people find difficult to discriminate (Birch, 1993). As for lighting, the use of light sources with a high CIE General Color Rendering Index is recommended, as these sources will tend to make colors more saturated. Certainly, care should be taken with narrow-band light sources, such as light emitting diodes (LEDs), as these may appear very dim to some color-defective people. Finally, it is important to remember that for some color-coding applications, there are recommended limits for the colors that should be used. For example, the CIE recommends areas on the CIE 1931 Chromaticity Diagram within which

Modifying Factors

7-7

red, green, yellow, blue and white signal lights should lie. These areas are designed so that the red signal will be named as red, the green as green, and the yellow as yellow by people with the most common forms of defective color vision (CIE 1994).

7.3 Prolonged Work

The effects of lighting conditions on task performance, discussed above, will be evident regardless of task duration. When work is prolonged, however, other phenomena can occur that will interact with the lighting conditions. These phenomena are eyestrain, fatigue, and mood changes. These phenomena can occur over a wide range of task durations, depending on the suitability of the lighting for the task. However, when work is continuous for periods longer than 24 hours, sleep deprivation sets in. Long duration tasks that are monotonous and complex will be very sensitive to sleep deprivation, while tasks that are of short duration, simple and interesting, and that the worker has a high incentive to perform well will be resistant to the effects of sleep deprivation (Froberg, 1985). Performance in conditions of sleep deprivation is outside the scope of this report, so it will not be considered further. What will be considered are the phenomena of eyestrain, fatigue, and mood changes. All these phenomena can occur during work of normal duration, given the wrong lighting conditions.

7.3.1 Eyestrain

While eyestrain is a term widely used in everyday speech, it has proved difficult to define in scientific terms. The problem for eyestrain arises from the belief that the eye itself cannot be strained (Cogan, 1974), although the muscles controlling the movement of the eyes and the curvature of the lens in the eye can be strained (Fry 1974). This is really an argument over semantics. There is no doubt about the symptoms that are associated with eyestrain: irritation of the eyes, e.g., inflammation of the cornea and eyelids; breakdown of vision, e.g., blurring of vision; and the more remote effects of headaches, indigestion, and giddiness. No matter what these symptoms are called, they are undesirable. If they occur frequently, they are likely to affect motivation and therefore human performance.

Lighting and visual task conditions have been shown to cause eyestrain. Whenever lighting and task conditions combine to require the visual system to operate at the limits of its capabilities, either optically or perceptually, the symptoms associated with eyestrain are likely to occur. Optically, this means that either the muscles controlling the position and accommodation of the eye have to hold a fixed position for a long time, or they have to make more frequent and/or more powerful changes than usual. Perceptually, this means that it is difficult to extract the information required from the visual world. The difficulty arises either because the stimuli to the visual system are barely visible, or because the stimulus being sought differs little from other surrounding stimuli. Lighting conditions that have been shown to lead to eyestrain are inadequate illuminance for the task (Simonson and Brozek, 1948), excessive luminance ratios between different elements of a task (Wibom and Carlsson, 1987), and lamp flicker, even when it is not visible (Wilkins et al., 1989). Despite this list, it should not be thought that eyestrain is inevitable during a normal eight-hour working day. Carmichael and Dearborn (1947) measured the eye movement patterns of people continuously reading books in high-contrast, 10-point print for six

Modifying Factors

7-8

hours, at an illuminance of 160 lx, but found no changes. Apparently the visual system is perfectly capable of prolonged activity without strain, under the right conditions.

7.3.2 Fatigue

The definition of fatigue, too, has been the subject of much argument (Bartley and Chute, 1947; Floyd and Welford, 1953; Cameron, 1974; Hockey, 1983; Craig and Cooper, 1992; Brown, 1994). Partly this is because of the difference between physiological (muscular) fatigue and psychological (cognitive) fatigue. Physiological fatigue is easily measurable, has a clear relationship to energy expenditure, is largely specific to the muscles used, and can invariably be demonstrated. According to Grandjean (1969) it occurs in two forms: static, in which the muscles remain in a state of increased tension in order to sustain a particular posture; and dynamic, in which the muscles tense and relax rhythmically. Both types of muscular fatigue can impair perceptual-motor tasks. Static muscular fatigue, in particular, may distract the worker from the cognitive demands of a task. In comparison, psychological fatigue is not easily measured, does not have a clear relationship to energy expenditure, is a general rather than specific response to stress, and cannot easily be demonstrated within a conventional working day (Craig and Cooper, 1992). Regardless of how fatigue is defined, prolonged work at mainly cognitive tasks produces a number of different symptoms. It is these symptoms and their potential for changing task performance that are of interest.

Early research on sustained cognitive work was dominated by the analogy to muscular fatigue, the assumption being that prolonged mental work would lead to a decrement in performance in the same way that prolonged work with the same muscle always led to a decrement in the strength of muscle contractions. Unfortunately, this did not occur. Poffenberger (1928) shows the results of five and one-half hours of continuous work on four different cognitive tasks: digit addition, sentence completion, composition judgment, and an intelligence test. The first two hours showed no change in any of the tasks, but after that the addition task declined, sentence completion and composition judgment were unchanged, and performance on the intelligence test improved. These results raise the important point that the impact of prolonged work on a cognitive task depends on the nature of the task. The task type which seems most likely to produce a performance decrement is one in which it is necessary to stay alert in readiness to respond to rare events which may occur at unknown times, e.g., a sentry at a military post (Mackworth, 1948). The boredom of such activity is obvious. This suggests that any task which is repetitive, unvarying, and uninteresting, such as the digit addition used by Poffenberger (1928), is likely to produce a performance decrement, if prolonged sufficiently, and particularly if the task is externally paced.

The perceived failure of the simple muscular fatigue analogy and the demands of war changed the emphasis of fatigue studies to the question of how skill deteriorated after prolonged work. Drew (1940) and Bartlett (1953) reported studies of the performance of air crews making long-duration flights in a flight simulator using only instruments. The picture that emerged was of disintegration of skilled performance and deterioration in mood. As the flight time increased, alertness decreased and four changes occurred in air crew behavior. They were:

• deteriorating coordination of responses leading to roughness in handling the controls

• tendency to require larger-than-normal changes in stimuli before responding

Modifying Factors

7-9

• narrowing of the field of attention, shown by a reduction in the number of instruments checked

• increased irritability leading to more aggressive responses to other people and the aircraft

Interestingly, a similar picture of breakdown in coordination has been obtained for a physical activity. Bates et al. (1977) filmed the movements of runners at the start and at the end of a relay race. The coordination between the movements that make for smooth running was much worse at the end of the race.

Since this pioneering work, the same pattern of skill deterioration has been shown for a variety of tasks and environments. Dureman and Boden (1972) showed a similar disintegration of skill for four hours’ driving in a simulator. Bursill (1958) and Hockey (1970) showed narrowing of attention in hot, humid, and noise conditions respectively. It would be interesting to explore what happens to skilled work, as opposed to simple repetitive work, done in poor lighting conditions.

The variability in response timing implicit in worsening coordination has also been shown in prolonged work of a repetitive nature. Bills (1931) measured the fluctuations of response time to a task requiring fast responses, e.g. color naming. He found occasional much longer response times, which he called blocks. Bertelson and Joffe (1963) measured reaction time to a choice task of 30 minutes’ duration, which provided no respite in that as soon as a response was made, another response was called for. The distribution of reaction times changed over the 30-minute period, with the number of long reaction times markedly increasing. Of more relevance to the effect of lighting is a study by Kehk and Krivohlavy (1967). The subject had to report the orientation of the gaps in a series of Landolt rings presented as fast as the subject could handle, for a period of thirty minutes. Two different lighting conditions were examined, one with and one without a glare source at 45 degrees to the line of sight. The average work speed did not change over the thirty minutes and was no different for the two lighting conditions. The variability in responses, however, increased dramatically over time for the glare source conditions but showed no change over time without the glare source. Kogi and Saito (1973) have shown similar oscillation of performance on a continuous tracking task. It is important to note that this variability in response times does not necessarily lead to a performance decrement. Warren and Clarke (1936) report that over sixty-five hours of continuous performance, the frequency of long response times steadily increased but the average response time did not change. This suggests that the long response times allow accumulated fatigue to dissipate, and that the longer times are made up by shorter response times later in the task. This in turn implies that the increased variability of response is only likely to lead to a worsening of performance in tasks that require a fast response and that are externally paced.

This discussion of psychological fatigue serves to emphasize two facts. The first is that a variety of changes in the nature of performance are possible when work is prolonged. The second is that the way in which performance may change depends on the nature of the task. Tasks that are varied but complex seem likely to deteriorate by a worsening of coordination and a narrowing of attention, i.e., by deterioration in the quality of work. Tasks that are simple, repetitive, and uninteresting seem likely to deteriorate by a performance decrement, i.e., by a simple decline in output.

Modifying Factors

7-10

While this understanding is useful, it ignores the potential people have for changing their work strategies to maintain performance in the presence of increased feelings of tiredness. Schonpflug (1983) reported a study in which information had to be gathered and either held in mind to aid in later decisions, or referenced and consulted at a cost of longer time taken to make the decision. When the subjects felt fatigued, either because of working under time pressure or because of distracting noise, they placed less reliance on memory, so that they either took longer to complete the task by looking up the information, or made more errors by relying on incomplete or poorly remembered information. Welford et al. (1950) have shown a similar memory deficit among air crews after completing a long flight.

More generally, the concept behind the choices people make when feeling fatigued is that people who are fatigued would rather not make further effort. This implies that, given a choice between a number of actions which differ in the probability of success and the effort required, as work is prolonged people will tend to choose actions which require less effort even if they entail greater risk of errors (Shingledecker and Holding, 1974). Of course, this is implicit in the flying simulator studies, because as flight duration increased, pilots gradually chose to ignore many of the instruments, including the fuel gauge. Brown et al. (1970) have shown an increase in risk-taking behavior for prolonged driving. Over four 3-hour driving sessions, subjects showed little difference in speed, but the number of risky overtaking maneuvers increased by 50% between the first and fourth session. Even simple repetitive mental arithmetic has shown the same trade-off between decreased effort and increased risk, with increasing time spent in 95 dBA noise (Holding et al., 1983). These results serve to emphasize that the likely effects of prolonged work will depend on the nature of the work and the freedom workers have to modify their strategies.

The idea that lighting conditions that are inadequate for the task can cause fatigue is suggested by the work of Simonson and Brozek (1948). They had people copy small letters subtending 10 min arc at the eye presented in an aperture for 0.56 s, for two hours, at a fixed illuminance. Six different illuminances were used: 20, 50, 150, 500, 1000, and 3000 lx. Performance was sampled five minutes after starting work, after 60 minutes, and after 110 minutes. After completing the work, subjects answered a questionnaire concerned with their feelings of eyestrain and tiredness. At 20 lx, the performance decrement from the beginning to the end of the work period was greatest, as was the variability in performance and the subjective feelings of eyestrain and tiredness. By comparison, at 1000 lx, the decrement in performance and the variability of performance were minimal and the feelings of eyestrain and discomfort were least. An illuminance of 20 lx is almost certainly inadequate for this task, so the performance decrement, increased variability, and increased feelings of eyestrain and fatigue should be expected, as should the improvement when the illuminance was increased to 1000 lx.

7.3.3 Mood Changes

All the above discussion has been concerned with effects on output. The flying studies also showed an increase in irritability after prolonged performance. A subjective feeling in itself may affect some aspects of task performance through mood changes known as positive and negative affect. Positive affect, defined as pleasant feelings produced by commonplace events or circumstances, has been shown to influence cognition and social behavior. Specifically, positive affect has been shown to increase efficiency in making some types of decisions, and to promote innovation and creative problem-solving. It also changes the choices people make and the

Modifying Factors

7-11

judgments they deliver. For example, it has been shown to alter people’s preferences for resolving conflict by collaboration rather than avoidance and also to change their opinions of the tasks they perform (Isen and Baron, 1991). Negative affect has the opposite effects. Boyce and Eklund (1996) and Eklund et al. (2000) have shown that prolonged work on a monotonous data-entry task produced a reduction in positive affect but little change in negative affect.

The factors that determine positive affect are both small and wide. The factors are small, because the stimuli that have been shown to generate positive affect are low-level stimuli, ranging from receiving a small but unexpected gift from a manufacturer’s representative to being given positive feedback about task performance. They are wide, because the physical environment, the organizational structure, and the organizational culture can influence positive affect. Lighting is clearly part of the physical environment, and lighting conditions such as the illuminance and the correlated color temperature of the lighting have been shown to change mood (McCloughan et al., 1999) and to change behavior in a way consistent with positive affect (Baron et al., 1992). This suggests that lighting can produce mood changes (see Section 6.4). Poor lighting conditions will likely generate fatigue and negative affect. Conversely, lighting which is perceived to be much more attractive than is usual might produce positive affect.

Overall, there can be little doubt that the combination of inappropriate lighting for a task and the need for prolonged performance of the task is likely to produce eyestrain, fatigue, and mood changes. These, in turn, are likely to be associated with changes in task performance. The difficulty is in predicting what that change in task performance will be. The difficulty arises because people will, where the task allows, manipulate the way they do the task, vary their performance over time, and change the level of errors they are willing to accept. Thus, the consequences of prolonged work in inappropriate lighting conditions depend greatly on the specific nature of the task and the consequences of errors. Fortunately, there is no reason why eyestrain, fatigue, and mood changes should occur, at least as long as task duration is limited to a normal working day. By ensuring that the lighting conditions allow a high level of visual performance of the task, combined with visual comfort, the likelihood that the adverse effects of prolonged work will occur can be minimized. A high level of visual performance can be guaranteed by making sure the stimuli presented to the visual system by the task, i.e., visual size, luminance contrast, and color difference, are well above threshold; by ensuring that the observer has the correct optical refraction; and by lighting the task to the level required to produce the necessary retinal illumination.

7.4 Lighting and Health

The use of light for purely medical purposes is outside the scope of this review, particularly where the condition being treated is sufficiently severe as to exclude sufferers from the workforce. However, a number of chronic medical problems can be alleviated with light and, if untreated, can be expected to impact the performance of the individual when at work and of the organization when the individual is absent from work through sickness.

Modifying Factors

7-12

7.4.1 Sleep Disorders

Most employment requires people to be present at their place of work at a given time and, while there, to be awake and working. People who are chronically sleep- deprived, as many night-shift workers are, or whose body is telling them to go to sleep while at work, are not likely to be the most effective or cooperative employees. There are several different forms of sleep disorder. The forms that are amenable to treatment by light exposure are those that involve the timing of sleep. Specifically, the problems are sleep that starts too late, ends too early, is too fragile, or is irregular. People with delayed sleep phase syndrome report difficulty in getting to sleep before about 1 a.m. to 3 a.m. and problems regaining alertness or energy before about 11 a.m. Advanced sleep phase syndrome is associated with having trouble staying awake in the early evening and having fragile sleep in the second half of the night. Delayed sleep phase syndrome is more common in the young, while advanced sleep phase syndrome is predominantly found in the elderly, but either can occur throughout life.

Exposure to bright light is known to shift the phase of the human circadian rhythm, but the direction and magnitude of the phase shift depend on the timing of the exposure (see Section 5.4). For delayed sleep phase syndrome, it is believed that exposure to bright light in the morning will help people with this condition to fall asleep earlier. For advanced sleep phase syndrome, exposure to bright light in the evening has been shown to make sleep longer and more regular (Lack and Wright, 1993; Terman et al., 1995). About 16 million people in the United States experience advanced or delayed sleep phase syndromes to some degree.

A light box can deliver bright light exposure at the right time. Light boxes are available that will produce 2,500 – 10,000 lx on the face. Products with a large lens or white-painted reflector, which ensure that the brightness of the lamp is spread over a large area, are most likely to avoid visual discomfort. The most effective light spectrum is not yet known.

7.4.2 Seasonally Affective Disorder (SAD)

SAD is a form of depression characterized by a seasonal cycle. Typically, a person suffering from winter SAD will experience increased lethargy, hypersomnia, lack of libido, and carbohydrate cravings during winter. The symptoms disappear in summer and return the following winter. Summer SAD is characterized by loss of appetite and insomnia. About ten times more people experience winter SAD than experience summer SAD. Winter SAD has been shown to be more prevalent in higher latitudes and to be correlated with the amount of exposure to light. Exposure to bright light during the day has been shown to alleviate these symptoms (Rosenthal et al., 1988; Terman et al., 1995). About 6 million people in the United States experience winter SAD and many more experience minor symptoms, sometimes called sub-SAD.

One approach to providing bright light exposure for people with SAD is simply to spend time outdoors or to sit in a window looking out during the day. If these activities are not practical for any reason, a light box can be used (Lam and Levitt, 1999).

Modifying Factors

7-13

7.4.3 Migraines

Migraine has been described as a neurovascular reaction to changes in an individual’s internal or external environment (Wilkins, 1991). A migraine attack is much more than a severe headache. Nausea, vomiting, intolerance of smells, and photophobia are all part of migraine attack. The exact cause of a migraine attack is unknown. What is known is that migraineurs are more sensitive to light than people who do not experience migraine, even when they are headache-free (Main et al., 1997). This means migraineurs are much more likely to experience glare from luminaires and to complain about high light levels. In addition, migraineurs are likely to be hypersensitive to flicker and high-contrast, repetitive patterns. It is estimated that about 26 million people in the United States suffer from migraine.

8-1

8 PROGRESS SINCE 1989

8.1 Background

This report updates and replaces an earlier work entitled "Lighting and Human Performance: A Review" that was published by the National Electrical Manufacturers Association and the Lighting Research Institute in 1989 (Boyce et al., 1989). Like this report, the earlier publication explored the relationship between lighting conditions and the ability to carry out tasks in interiors. The purpose of this section is to summarize the progress that has been made in understanding the relationship between lighting conditions and human performance since 1989, under the subject categories used in the earlier publication..

8.2 Progress in Concepts

Based on the literature review, progress since 1989 can be considered at two levels, the conceptual and the practical. The previous document (Boyce et al., 1989) was organized around the photometric characteristics of lighting, namely, illuminance, light distribution and light spectra. The consequences for human performance of changing lighting conditions along these dimensions were considered in terms of direct and indirect effects. Direct effects are those that operate either by changing the stimulus to the visual system or by changing the operating state of the visual system, i.e. changes that operate through changes in visibility. Indirect effects are changes in human performance that occur because of changes in attention, arousal, mood or hormone balance. The conceptual framework used in the previous report was primitive, at best, and for the indirect effects was non-existent. The present report is based around a conceptual framework setting out the routes whereby lighting conditions can be expected to influence human performance. This is progress because conceptual frameworks are important for research. They form the unstated assumptions within which research is conceived.

8.3 Practical Progress

As for practical progress, the previous publication summarized the state of knowledge in 1989 using a matrix in which research areas were divided up into twelve different categories, six for direct effects and six for indirect effects. Progress in each of these categories is summarized below. The first six categories are direct effects. The last six categories are indirect effects.

Visual performance: Two quantitative models of visual performance, the RVP model (Rea and Ouellette, 1991) and the VP model (Adrian and Gibbons, 1994, 1999), have been developed. The RVP model is applicable to on-axis tasks and has been independently validated for a number of such tasks. The VP model is believed to be applicable to tasks that entail some off-axis vision but

Progress Since 1989

8-2

has not yet been independently validated. Both these models cover a range of luminance contrasts, sizes of task detail and adaptation luminances. Neither considers color differences between the task and its background, nor the effect of retinal image blur.

Task performance: Quantitative models for some specific tasks have been constructed, e.g., reading words (Bailey et al., 1993), data entry of alphanumeric codes (Eklund et al., 2000), but no general model exists. Given the unique combination of visual, cognitive and motor components that exists for every task, it is difficult to see how a general model could exist.

Color vision: For achromatic, near-threshold tasks, it has been shown that scotopically-enriched light spectra reduce pupil size and improve task performance (Berman et al., 1993). What happens for realistic suprathreshold tasks is open to question, a number of studies using suprathreshold tasks having failed to show any effect of scotopically-enriched light spectra (Rea et al., 1990; Vrabel et al., 1995). No progress has been made in quantifying the effect of lighting on the performance of chromatic tasks.

Visual search: No progress has been made in this area for performance in interiors.

Age and individual differences: The effects of aging on ocular physiology have become much more clearly understood. Based on these changes, guidelines for providing lighting for the elderly have been developed. The resulting lighting installations have been shown to lead to better performance of everyday tasks and a better quality of life for the elderly.

Fatigue: There is no doubt that prolonged work in inappropriate lighting conditions can cause fatigue but whether fatigue inevitably affects task performance is unclear. This is because the effect fatigue has on task performance depends on the nature and duration of the task, and the freedom the worker has to modify how the task is done.

Discomfort: It has been established that lighting conditions that cause discomfort and alter the stimuli the task presents to the visual system will change visual performance. Lighting conditions that simply cause discomfort without affecting the stimuli presented by the task, may or may not affect task performance.

Light as an attention stimulus: It has been clearly established that light can be used to attract attention, but no progress has been made on quantifying the conditions necessary for light to have such an effect.

Light and arousal: Exposure to light can increase arousal, particularly at night, when exposure to light in sufficient quantities will suppress the hormone melatonin.

Light and mood: Mood can be influenced by many factors, lighting being just one of them. Changes in mood have been shown to affect the performance of all types of task, both visual and non-visual.

Lighting's influence on behavior: Behavior can be influenced by many factors, lighting being just one of them. Lighting can affect behavior by attracting attention, or by providing the necessary

Progress Since 1989

8-3

visual conditions in one area rather than another, or by sending a "message" about what is appropriate behavior.

Lighting and hormone balance: Understanding of the characteristics of the circadian photobiology system has grown greatly since 1989 but much remains to be determined. However, there is no doubt that exposure to light at night can have a short-term arousing effect and a longer term phase-shifting effect. To ensure a reliable phase-shifting effect, control of light exposure over twenty-four hours is necessary. Exposure to light at night can influence the performance of both visual and non-visual tasks but why some tasks are more sensitive than others is unknown.

9-1

9 RESEARCH AGENDA

9.1 "Drivers" of Research

The rationale behind this research agenda is that the provision of lighting is not an end in itself but, rather, a means to an end. Therefore, for a lighting research agenda to attract widespread support it should address the ends that are widely perceived to need attention. There are four such ends where improvements in human performance produced by lighting can be expected to have an impact. They are increasing productivity, reducing energy consumption, enhancing health and improving the quality of life. These objectives are the "drivers" of research.

Increasing Productivity

The United States is a high-wage economy. The globalization of business has meant that high-wage economies have to be more productive and more intelligent than low wage economies to keep their populations employed. Providing appropriate lighting has been recognized as an aid to productivity for many years. However, the type of work to which lighting has been applied has almost always been the production of "widgets", the increase in productivity being associated with an increase in the visibility of the "widget". Over the last two decades, the economy of the United States has moved away from being based mainly on manufacturing to being based more on the transfer of information, and away from a normal working day to twenty-four hour working. How lighting might influence the performance of "knowledge" work, done over twenty-four hours, is largely unexplored.

Reducing Energy Consumption

Long-term, there can be little doubt that there will be a need to reduce energy consumption in the United States, for three reasons. The first is the threat of global warming and the consequent climate change. The second is the ultimately limited supply of fuel, a limit that is likely to be reached sooner as some countries with large populations strive to increase their standard of living. The third is the growth in enthusiasm for all things "natural", including pristine air and water. Generating electricity burns a lot of fuel, thereby contributing to global warming and producing air and water pollution. Lighting is a major user of electricity. This suggests that in the future there will be consistent pressure to minimize the use of electric light in interiors. When responding to such pressure, it would be as well to understand what the consequences might be for human performance.

Research Agenda

9-2

Enhancing Health

In developed countries like the United States, where many forms of previously fatal medical conditions have been largely eliminated, there is increasing concern about non-life-threatening health issues. This tendency is exacerbated by the presence of an aging population, many of whom can be expected to live longer than any previous generation. They may also have to work longer, because the number of young people available to replace retiring older workers is inadequate. Lighting has a role to play in enabling older workers to perform their work effectively. It is also probable that lighting has a role to play in promoting some aspects of health for people of all ages, and workers who are not healthy are not likely to perform at their best.

Improving the Quality of Life

The quality of life means different things to different people. For many citizens of the Balkans, improving the quality of life means returning to a peaceful existence and economic growth. For many citizens of the United States, which is the focus of this research agenda, it means improving the education system and making the streets safe. For others, particularly older citizens, it means maintaining their independence in their own homes; while for yet others, it means having an attractive environment around them. Lighting can impact all these aims.

9.2 The Lighting Matrix

The "drivers" of research described above operate on a three-dimensional matrix that identifies what aspects of a lighting installation can have an effect on human performance, in different application areas, through different routes. Figure 9-1 shows the matrix.

One axis covers the routes through which lighting may influence human performance; visibility, circadian photobiology and "message". A lighting installation always affects all three of these routes - it is not possible to affect one without influencing the others. For example, a lighting installation designed to make a visual inspection task easier to perform is focussed on the visibility of the defects, but the exposure to light it provides will also be an element in determining the state of the worker's circadian system and the existence of the installation and how effective it is sends a message to the worker and to anyone else who sees it.

Another axis covers the dimensions of a lighting installation that are open to adjustment by the lighting engineer and designer. They are the amount, spectrum, distribution, timing and duration of light. In the past the emphasis of research has been very focussed on the amount of light. Recently, more consideration has been given to light spectrum, but the effects of light distribution, and timing and duration of light exposure, have largely been ignored.

The final axis covers the application areas where human performance, in its widest sense, is important. The application areas are industrial, commercial, retail, public service and hospitality/residential. Industrial applications are concerned with the production of "widgets" in some form, e.g., automobile production. Lighting installations in these areas are dominated by visibility considerations. Commercial applications are concerned with the manipulation of information, e.g., offices. Lighting installations in these areas are dominated by visibility

Research Agenda

9-3

considerations and, in some areas, by "message". Retail applications are concerned with selling. Lighting installations in these areas are dominated by "message" considerations. Public service applications, such as hospitals and police stations, undertake a very diverse range of activities, but one thing they have in common is they operate over twenty-four hours. Lighting installations in these areas are dominated by visibility and photobiology considerations. Hospitality/residential applications are concerned with creating the desired impression. Lighting installations in hospitality/residential applications are dominated by "message" considerations.

Figure 9-1 The lighting matrix showing what aspects of a lighting installation can have an effect on human performance, in different application areas, through different routes. What parts of the matrix are chosen for study is influenced by the "drivers" of research.

Research Agenda

9-4

Figure 9-2 shows an exploded version of the lighting matrix. The heavily-shaded areas in Figure 9-2 indicate high priority areas for study, i.e., conjunctions of lighting dimensions, application areas and lighting effects where little is known but there is reason to expect significant effects of lighting on human performance. The lightly-shaded areas in Figure 9-2 indicate moderate priority areas of study, i.e., conjunctions where there are likely to be significant effects of lighting on human performance but only under a limited range of conditions. The hatched areas in Figure 9-2 indicate conjunctions where the effects of lighting on human performance are well-established and there is little need for further research. The unshaded areas in Figure 9-2 indicate the conjunctions where either there is believed to be little effect of lighting conditions on human performance or the particular conjunction is unlikely to occur in practice. One thing that is clear from Figure 9-2 is that we know a lot about the effects of visibility and the little need for a lot more research in this area. It is on the circadian photobiology and "message" routes that effort should be concentrated.

The specific topics listed in the research agenda are focussed on the high priority areas of study and classified according to the route through which changes in lighting conditions can affect human performance. The specific topics are based on the preceding review of the literature. Some of the topics are fundamental in nature while others are concerned with the practicality of application. All are important if lighting is to be used to enhance human performance most effectively. Therefore, the topics are not listed in any order of priority. Different sponsors will be interested in supporting work in different areas, at different levels from basic to applied.

Research Agenda

9-5

Figure 9-2 An exploded version of the lighting matrix.

Research Agenda

9-6

9.3 Visibility: Research Agenda

9.3.1 Overview

The effect of lighting conditions on task visibility has been a subject of study for many years. Because of this, task visibility is the best understood of the routes whereby lighting conditions can affect human performance. A validated, quantitative model for on-axis visual performance exists. What is required to make it valuable is a system of task classification that identifies the importance of visual performance to task performance for any on-axis task. Further study is required to understand the effect of lighting conditions on the performance of tasks requiring off-axis vision.

9.3.2 Research Agenda

1. Develop a task classification system that enables tasks with an important visual component to be identified.

2. Investigate the effect of lighting conditions on the performance of tasks requiring different amounts of off-axis vision.

9.3.3 Justification

Having a task classification system would enable the tasks most sensitive to lighting conditions to be identified. This would encourage a wider use of the model of on-axis visual performance. It would also suggest where to apply light to most effectively increase productivity and identify where energy could be saved without sacrificing productivity.

Visual tasks come in many different forms, some requiring the use of both foveal and peripheral vision. The greater is the off-axis component, the less likely it is that current understanding of how lighting conditions affect visual performance is adequate.

9.4 Circadian Photobiology: Research Agenda

9.4.1 Overview

Knowledge and understanding of how light exposure changes the human circadian system has increased dramatically over the last twenty years, although it has not yet identified the photoreceptors involved. There is no doubt that exposure to light at the right time can shift the phase of the human circadian rhythm over the next twenty-four hours and have an immediate effect on alertness. The right time is when melatonin is being produced, which is usually at night. These circadian photobiological effects of exposure to light are important because they have the potential to affect the performance of all types of tasks, not just visual tasks. They are also

Research Agenda

9-7

important because the illuminances typically used for seeing are on the borderline of being effective for circadian photobiology, depending on the duration and timing of exposure.

9.4.2 Research Agenda:

1. Determine the spectral sensitivity of the human circadian system and then develop a suitable system of photometry.

2. Test the models of the effects on the circadian system of combining the amount, spectrum, timing and duration of exposure to light.

3. Investigate the effect of light exposure on alertness and task performance at the start and end of the normal working day.

4. Examine the interaction between cognitive task complexity and the duration of performance at night before deterioration of performance occurs. Determine if this duration can be extended by the use of light exposure at night.

5. Determine if an increase in alertness at night produced by exposure to light, is followed by greater fatigue when the light is removed.

6. Monitor developments in the understanding of health impacts of light exposure at night.

9.4.3 Justification

Items 1 and 2 on the research agenda seek to develop a basic understanding of how the conditions of light exposure affect the human circadian system. Knowledge of this type is needed before it will be possible to make accurate predictions of the effect of light exposure for any application. Given knowledge of this type it would be possible to estimate how important exposure to conventional electric lighting is to the operation of the human circadian system, or alternatively, what sort of lighting is needed to impact the human circadian system.

Items 3 and 4 on the research agenda are concerned with the impact of circadian photobiology on task performance, for both day and night, and for simple and complex tasks. If it could be shown that exposure to light had an effect on cognitive task performance during the day, this would introduce a new role for lighting in enhancing productivity in commercial, industrial and public service facilities. If it could be shown that exposure to light had an effect on cognitive work during the night, this would introduce a new role for lighting in enhancing productivity in industrial and public service facilities and anywhere else where work was maintained over twenty-four hours.

Items 5 and 6 in the research agenda are concerned with possible side effects of light exposure at night. Such understanding is necessary if the circadian photobiology route to influencing human performance is to be widely accepted.

Research Agenda

9-8

9.5 "Message": Research Agenda

9.5.1 Overview

The overwhelming impression left by the studies that have been examined in this area is one of variability in response. The aspects of lighting that can cause discomfort are known but all can be used to positive effect in the right context. Similarly, there is no doubt that different lighting conditions can produce different perceptions of objects and spaces, but whether that has an effect on mood and behavior depends on the context and culture within which those perceptions occur. This uncertain response occurs because mood and behavior are determined by many factors, lighting conditions being but one. The problem for research in this area is not in determining whether lighting conditions can alter mood and behavior but rather in what situations mood and behavior are most sensitive to lighting conditions.


1. Establish a lexicon of lighting by determining what "messages" different lighting conditions deliver in a given context.

2. Determine how sensitive mood is to lighting conditions, relative to architecture and decor, in realistic situations, and whether mood affects task performance consistently.

3. Establish the lighting conditions necessary to draw people to a retail display and if that attraction diminishes over repeated exposures.

9.5.3 Justification

Item 1 on the research agenda is concerned with the basic question of stability of "message" over time, context and culture. Unless stability can be demonstrated, it will be difficult to apply lighting to deliver a consistent "message" for any application. Identifying the lexicon of lighting should be valuable in enhancing the quality of life in general and for enhancing productivity in retail and hospitality applications.

Item 2 is concerned with the potency of lighting for changing mood relative to other aspects of the visual environment and the impact of mood changes on task performance. Understanding how lighting can be used to change mood offers the possibility of enhancing the quality of life in general. Establishing that mood changes alter task performance offers the possibility of increasing productivity in all applications.

Item 3 is focussed on retail applications. Knowledge of the lighting conditions needed to attract people to a display is important for increasing productivity and for reducing energy consumption in retail applications.

Research Agenda

9-9

9.6 Modifying Factors: Research Agenda

9.6.1 Overview

There can be little doubt that the deterioration of visual capabilities with age or eyestrain and the deterioration of cognitive abilities through fatigue or ill-health will ultimately diminish task performance. The problems for any research agenda are, first, to determine whether the deterioration of visual capabilities with age or cognitive capabilities with fatigue are important for realistic working conditions; and, second, to identify the tasks that are most sensitive to these declines.


1. Determine how to most easily ameliorate the effects of aging in the visual system and the consequences for task performance during the day.

2. Determine if exposure to light can be used to ameliorate the effect of aging on the circadian system and the consequences for task performance during the day.

3. Determine the effect of lighting conditions on the prolonged performance of skilled work.

4. Determine whether light exposure improves the quality of work of people experiencing SAD and sub-SAD during winter.

5. Using a sample of people who experience advanced and delayed sleep phase disorder, examine the effect of light exposure on absenteeism and quality of work done.

9.6.3 Justification

Items 1 and 2 in the research agenda are concerned with the effects of aging on the visual and circadian systems and how lighting might be used to ameliorate the consequences of such changes. Overcoming these effects of aging enhances health and should increase productivity where elderly people are working.

Item 3 is concerned with the effect of lighting on the prolonged performance of skilled work. Skilled work involving the manipulation of information from different sources is a common feature of commercial life today. If it could be shown that different lighting conditions prevented the deterioration in the quality of such work as work is prolonged, there would be an opportunity to use lighting to enhance productivity in commercial and public service applications.

Items 4 and 5 are concerned with enhancing the health and productivity of people who experience two commonly-found medical conditions, through lighting.

10-1

10 RESEARCH METHODS

If the research agenda are to be undertaken successfully, investigations must be conducted properly after a careful review of the literature. Unfortunately, lighting research does not have a reputation for good quality (Veitch and Newsham, 1998b). With this in mind, a brief review of the approaches that can be used and the advantages and disadvantages of each is included here.

There are several different approaches to obtaining information about the effect of lighting conditions on human performance. They are as follows:

Epidemiological approach: This approach is used to determine if two variables are correlated, e.g., if smoking cigarettes is related to the incidence of lung cancer. It is particularly useful as a method if there are many intervening factors that cannot be controlled, and / or the effect does not occur until long after exposure to the stimulus. The studies of the Heschong-Mahone group on the effect of the presence of skylights on retail sales and children’s school test performance are typical of this approach (Heschong-Mahone, 1999a and b). The overwhelming drawback of this approach is that it can only reveal whether two variables are correlated, not whether they are causally related. This means such studies are useful for determining if a relationship is worthy of further study, although such study should be undertaken only when a major effect is identified (Taubes, 1995). Practically, the main drawback is that they require extensive databases of all the relevant information, databases that often do not exist or are inaccessible.

Ecological approach: This approach is simply that of observation followed by interpretation, although it is sometimes possible to perturb the process by introducing a change in conditions. The study of Areni and Kim (1994) on the behavior of people in a wine store under “bright” and “soft” lighting is an example of this approach. This approach is most suitable where the context in which the study takes place is important and removing the activity from the context would destroy the phenomenon being studied. The main disadvantage of this approach is that it cannot provide an explanation of why effects occur. Explanations that are given when using this approach are post-hoc rationalizations. However, for some studies, such as the effect of lighting conditions on behavior, there is little alternative, because this approach provides the minimum interference with the natural condition.

Stimulus / response approach: This is the approach conventionally used in human factors research, vision research, and psychophysics. In its simplest form, a stimulus is administered to the subject under controlled conditions and a response is measured. Experiments based on this approach require decisions about three classes of variables: independent variables, dependent variables, and intervening variables. Independent variables define the conditions being examined by the experiment. Dependent variables are the measures used to quantify the response to the independent variables. Intervening variables are all those factors that may influence the relationship being investigated. There are two types of intervening variables: those that need to be controlled and those that need to be measured in order to identify the reason for any change in

Research Methods

10-2

the dependent variables. Experimental design procedures allow for several independent variables and the interactions between them to be examined in one experiment. Provided care is taken with the selection, measurement and control of independent, dependent, and intervening variables, and provided that statistical analysis of the collected data is thorough and appropriate, the stimulus / response approach can prove cause and effect. This is a great advantage over the epidemiological and ecological approaches. However, the stimulus / response approach does have one drawback; namely, that rigorous control of the intervening variables may destroy or modify the phenomenon being examined. This drawback does not confine the stimulus / response approach to the laboratory. Such experiments can be conducted in the field given the right conditions (Maniccia et al., 1999). What the drawback does mean is that where the effect being studied may be influenced by the context, e.g., where behavior is being studied, the stimulus / response approach may not be appropriate because the very act of taking part in an experiment may change the response.

It would be a mistake to think these approaches are always mutually exclusive. Rather, different approaches are appropriate for answering different questions. It is very rare for a single experiment to provide a conclusive answer to a question. Usually, multiple experiments are required, with the results from different approaches providing mutual support. This ideal is called converging operations and is much like making a case for presentation in court. In the legal situation, the prosecutor has to prove that a crime occurred and that the accused had the means and the motive to carry out the crime. In scientific research, the researcher has to prove that lighting was responsible for the measured effect. To do that, the researcher has to prove that a change in response occurred and provide a proven mechanism through which lighting might act to produce that response. This is where the cumulative nature of science is valuable. A thorough review of the literature in any field will often provide supporting evidence, or at least identify the areas where evidence is missing, as is, hopefully, shown by this report.

As an aid in planning effective research, Wyon (1996a) has introduced the idea of a linked mechanisms map (LMM). A LMM sets out all the pathways between the independent variables and the dependent variables in a specific experiment. It is only when all the steps along one or more pathways have been proven that the effect of the independent variable on the dependent variable can be said to be established. LMMs provide a rational basis for answering the question, “Why do you expect your independent variable to affect your dependent variable?” This question needs to be addressed at the planning stage of an investigation. Without a rational answer to this question, any research project is reduced to a “fishing expedition.” Before it is possible to construct a LMM, it is necessary to have some conceptual framework within which the hypothesized effects of the independent variables on the dependent variables can be organized. Conceptual frameworks for predicting the effect of lighting conditions through visibility and circadian photobiology can be readily constructed from the literature. In the absence of an extensive literature, conceptual frameworks for the impact of lighting conditions on “message” issues are more difficult to frame, although some have been proposed (Donovan and Rossiter, 1982; Gorn, 1982; Belcher and Kluczny, 1987b).

Finally, it is important to recognize that we are not alone. Analogous research has been going on for many years concerning the effects of other aspects of the physical environment on human performance (Wyon, 1996b; Milliman, 1982). It would be as well to be aware of the concepts and techniques used and the understanding gained in these fields that may be relevant to the effects of lighting on human performance.

11-1

11 REFERENCES

Adrian, W., Change in visual acuity with age, in Lighting for Aging Vision and Health, Ed: W. Adrian, Lighting Research Institute, New York, 1995.

Adrian, W., and Eberach, K., On the relationship between visual threshold and the size of the surrounding field, Lighting Research and Technology, 1, 251-254, 1969.

Adrian, W., and Gibbons, R., Visual performance and its metric, Light and Engineering, 2, 1-34, 1994.

Adrian, W., and Gibbons, R., Draft report on The Correlation of Models of Vision and Visual Performance, CIE Committee TC 1-19, 1999.

Akutsu, H., Legge, G.E., Ross, J.A., and Schuebel, K.J., Psychophysics of reading - X. Effects of age-related changes in vision, Journal of Gerontology, 46, 325-331, 1991.

Areni, C.S., and Kim, D., The influence of in-store lighting on consumer's examination of merchandise in a wine store, International Journal of Research in Marketing, 11, 117-125, 1994.

Aschoff, J., Hoffmann, K., Pohl, H., and Wever, R., Re-entrainment of circadian rhythms after phase shifts of the Zeitgeber, Chronobiologica, 2, 23-78, 1975.

Aulhorn, E., and Harms, H., Visual Perimetry, in Visual Psychophysics, Eds: D. Jameson and L.M.Hurvich, Springer-Verlag, New York, 1972.

Badia, P., Myers, B., Boecker, M. and Harsh, J., Bright light effects on body temperature, alertness, EEG and behavior, Physiol. Behav., 50, 583-588, 1991.

Bailey, I., Clear, R. and Berman, S., Size as a determinant of reading speed, Journal of the Illuminating Engineering Society, 22, 102-117, 1993.

Balazsi, A.G., Rootman, J., Drance, S.M., Schulze, M., and Douglas, G.R., The effect of age on the nerve fiber population of the human optic nerve, American Journal of Ophthalmology, 97, 760-766, 1984.

Baron, R.A., Rea, M.S., and Daniels, S.G., Effects of indoor lighting (illuminance and spectral distribution) on the performance of cognitive tasks and interpersonal behaviors: The potential mediating role of positive affect, Motivation and Emotion, 16, 1-33, 1992.

References

11-2

Bartlett, F.C., Psychological criteria of fatigue, in Symposium on Fatigue, Eds. W.F. Floyd and A.T. Welford, H.K. Lewis and Co, London, 1953.

Bartley, S.H., and Chute, E., Fatigue and Impairment in Man, McGraw Hill, New York, 1947.

Bartness, T.J., and Goldman, B.D., Mammalian pineal melatonin: A clock for all seasons, Experimentia, 45, 939-945, 1997.

Bates, B.T., Osternig, L.R., and James, S.L., Fatigue effects in running, Journal of Motor Behavior, 9, 203, 1977.

Belcher, M.C., and Kluczny, R., The effects of light on decision making: some experimental results, Proceedings of the CIE, 21st Session, Venice, CIE, Vienna, 1987a.

Belcher, M.C., and Kluczny, R., A model for the possible effects of light on decision making, Lighting Design and Application, 17, 19-21 and 54-55, 1987b.

Bennett, C.A., Chitlangia, A., and Pengrekar, A., Illumination levels and performance of practical visual tasks, Proceedings of the Human Factors Society, 21st annual Meeting, San Francisco, 322-325, 1977.

Berman, S.M., Greenhouse, D.S., Bailey, I.,L., Clear, R.D., and Raasch, T.W., Human electroretinogram responses to video displays, fluorescent lighting and other high frequency sources, Optometry and Vision Science, 68, 645-662, 1991.

Berman, S.M., Fein, G., Jewett, D.L., Saika, G., and Ashford, F., Spectral determinants of steady-state pupil size with a full field of view, Journal of the Illuminating Engineering Society, 21, 3-13, 1992.

Berman, S.M., Fein, G., Jewett, D.L., and Ashford, F., Luminance-controlled pupil size affects Landolt C task performance, Journal of the Illuminating Engineering Society, 22, 150-165, 1993.

Berman, S.M., Fein, G., Jewett, D.L. and Ashford, F., Landolt-C recognition in elderly subjects is affected by scotopic intensity of surround illuminants, Journal of the Illuminating Engineering Society, 23, 123-130, 1994.

Bertelson, P., and Joffe, R., Blocking in prolonged serial responding, Ergonomics, 6, 109, 1963.

Beutell, A.W., An analytical basis for a lighting code, Illum. Engr. (London), 27, 5, 1934.

Bills, A.G., A new principle of mental fatigue, Am. J. Psychol., 43, 230, 1931.

Birch, J., Diagnosis of Defective Colour Vision, Oxford University Press, Oxford, 1993.

Bjorset, H.H., and Frederiksen, E., A proposal for recommendations for the limitation of the contrast reduction in office lighting, Proceedings of the CIE, 19th Session, Kyoto, CIE, Paris, 1979.

References

11-3

Bjorvatn, B., Kecklund, G., and Akerstedt, T., Bright light treatment used for adaptation to night work and re-adaptation back to day life. A field study at an oil platform in the North Sea, J. Sleep Res. 8, 105-112, 1999.

Blackwell, H.R., Contrast thresholds of the human eye, Journal of the Optical Society of America, 36, 624-643, 1946.

Blackwell, H.R., Development and use of a quantitative method for specification of interior illumination levels on the basis of performance data, Illuminating Engineering, 54, 317-353, 1959.

Blackwell, H.R., and Blackwell, O.M., Visual performance data for 156 normal observers of various ages, Journal of the Illuminating Engineering Society, 1, 3 - 13, 1971.

Blackwell, H.R., and Blackwell, O.M., Population data for 140 normal 20 - 30 year olds for use in assessing some effects of lighting upon visual performance, Journal of the Illuminating Engineering Society, 9, 158-174, 1980.

Bloomfield, J.R., Studies in visual search, in Human Reliability in Quality Control, Eds. C.G.Drury and J,G,Fox, Taylor and Francis, London, 1975.

Boettner, E.A., and Wolter, J.R., Transmission of the ocular media, Investigative Ophthalmology, 1, 776-783, 1962.

Boff, K.R. and Lincoln J.E., Engineering Data Compendium: Human Perception and Performance, Harry G. Armstrong Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, OH, 1988.

Boulos, Z., Campbell, S.S., Lewy, A.J., Terman, M., Dijk, D-J, Eastman, C.I., Light Treatment for sleep disorders: Consensus report, VII Jet lag, J. Biol. Rhythms, 10, 167-176, 1995.

Boyce, P.R., Age, illuminance, visual performance and preference, Lighting Research and Technology, 5, 125-144, 1973.

Boyce, P.R., Variability of contrast rendering factor in lighting installations, Lighting Research and Technology, 10, 94-105, 1978.

Boyce, P.R., Berman, S.M., Collins, B.L., Lewis, A.L., and Rea, M.S., Lighting and Human Performance: A Review, National Electrical Manufacturers Association and Lighting Research Institute, Washington D.C., 1989.

Boyce, P.R., Lloyd, C.J., Eklund, N.H., and Brandston, H.M., Quantifying the effects of good lighting: The Green Hills Farms project, Proceedings of the IES Annual Conference, Illuminating Engineering Society of North America, New York, 1996.

References

11-4

Boyce, P.R., Beckstead, J.W., Eklund, N.H., Strobel, R.W. and Rea, M.S., Lighting the graveyard shift: the influence of a daylight-simulating skylight on the task performance and mood of night-shift workers, Lighting Research and Technology, 29, 105-142, 1997.

Boyce, P.R., and Eklund, N.H., An evaluation of office lighting options, Proceedings of the IES Annual Conference, Illuminating Engineering Society of North America, New York, 1996.

Boyce, P.R. and Rea, M.S., Plateau and escarpment: the shape of visual performance, Proceedings of the CIE 21st Session, Venice, CIE, Vienna, 1987.

Boynton, R.M., and Clarke, F.J.J., Sources of entoptic scatter in the human eye, Journal of the Optical Society of America, 54, 110-119, 1964.

Brainard, G.C., Lewy, A.J., Menaker, M., Miller, L.S., Fredrickson, R.H., Weleber, R.G., Cassone, V., and Hudson, D., Dose response relationship between light irradiance and the suppression of melatonin in human volunteers, Brain Res. 454, 212-218, 1988.

Brainard, G.C., Rollag, M.D., and Hanifin, J.P., Photic regulation of melatonin in humans: Ocular and neural signal transduction, J. Biol. Rhythms, 12, 537-546, 1997.

Brainard, G.C., Hanifin, J.P., Rollag, M.D., Greeson, J.M., Byrne, B., Glickman, G., Gerner, E., and Sanford, B., Human melatonin regulation is not mediated by the three cone photopic visual system., Journal of Clinical Endocrinology and Metabolism, 86. 433-436, 2001a.

Brainard, G.C., Hanifin, J.P., Greeson, J.M., Byrne, B., Glickman, G., Gerner, E., and Rollag, M.D., Action spectrum for melatonin regulation in humans: Evidence for a novel circadian photoreceptor, Journal of Neuroscience, 21, 6405-6412, 2001b.

Brown, I.D., Driver fatigue, Human Factors, 36, 298, 1994.

Brown, I.D., Tickner, A.H., and Simmons, D.C., Effect of prolonged driving on overtaking criteria, Ergonomics, 13, 239, 1970.

Buchanan, T.L., Barker, K.N., Gibson, J.T., Jiang, B.C, and Pearson, R.E., Illumination and errors in dispensing, American Journal of Hospital Pharmacy, 48, 2137-2145, 1991.

Bursill, A.E., The restriction of peripheral vision during exposure to hot and humid conditions, Quarterly Journal of Experimental Psychology, 10, 113, 1958.

Butler, D.L., and Biner, P.M., Preferred lighting levels: Variability among settings, behaviors and individuals, Environment and Behavior, 19, 695-721, 1987.

Cagnacci, A., Krauchi, K., Wirz-Justice, A., and Volpe, A., Homeostatic versus circadian effects of melatonin on core body temperature in humans, J. Biol. Rhythms, 12, 509-517, 1997.

References

11-5

Cajochen, C., Khalsa, S.B.S., Wyatt, J.K., Czeisler, C.A., and Dijk, D.J., EEG and ocular correlates of circadian melatonin phase and human performance decrements during sleep loss, Am.J. Physiol., 277, R640-R649, 1999a.

Cajochen, C, Zeitzer, J.M., Czeisler, C.A. and Dijk, D-J., Dose-response relationship for light intensity and alertness and its ocular and EEG correlates, Sleep Research Online, 2 (Supplement 1), 517, 1999b.

Cameron, C., A theory of fatigue, in Man under Stress, Ed. A.T. Welford, Taylor and Francis, London, 1974.

Campbell, S.S., Effects of timed bright-light exposure on shift work adaptation in middle-aged subjects, Sleep, 18, 408-416, 1995.

Campbell, S.S., Dijk, D.J., Boulos, Z., Eastman, C.I., Lewy, A.J., and Terman, M., Light treatment for sleep disorders: Consensus report III Alerting and activating effects, J. Biol. Rhythms, 10, 129-132, 1995.

Carmichael, L and Dearborn, W.F., Reading and Visual Fatigue, Houghton Mifflin Company, Boston, 1947.

Carr, S.J., and Dabbs, J.J., The effect of lighting, distance and intimacy of topic on verbal and visual behavior, Sociometry, 37, 592, 1974.

Chartered Institution of Building Services Engineers (CIBSE), CIBSE Code for Interior Lighting, CIBSE, London, 1994.

Clear, R. and Berman, S., Speed accuracy and VL, Journal of the Illuminating Engineering Society, 19, 124-131, 1990.

Clear, R, and Mistrick, R.G., Multilayer polarizers: A review of the claims, Journal of the Illuminating Engineering Society, 25, 70-88, 1996

Cogan, D.G., Lighting and health hazards, in NIOSH Symposium, The occupational safety and health effects associated with reduced levels of illumination, Cincinatti, US Department of Health, Education and Welfare, 1974.

Cole, R.J., and Kripke, D.F., Amelioration of jet lag by bright light treatment: Effects on sleep consolidation, Sleep Res., 18, 411, 1989.

Cole, B.L., and McDonald, W.A., Defective colour vision can impede acquisition from redundantly colour-coded video displays, Ophthalmic Physiological Optics, 8, 198-210, 1988.

Collins, B.L., Kuo, B.Y., Mayerson, S.E., Worthey, J.A., and Howett, G.L., Safety color appearance under selected light sources, National Bureau of Standards, NBS IR86-3403, National Bureau of Standards, Washington DC, 1986.

References

11-6

Colombo, E.M., Kirschbaum, C.F., and Raitelli, M., Legibility of texts: The influence of blur, Lighting Research and Technology, 19, 61-71, 1987.

Commission Internationale de l’Eclairage (CIE), A unified framework of methods for evaluating visual performance aspects of lighting. CIE Publication No. 19, CIE, Paris, 1972.

Commission Internationale de l’Eclairage (CIE), An analytic model for describing the influence of lighting parameters upon visual performance, CIE Publication No. 19/2, CIE, Paris, 1981.

Commission, Internationale de l’Eclairage (CIE), Technical Report: Review of the Official Recommendations of the CIE for the Colors of Signal Lights, CIE 107-1994, CIE, Vienna, 1994

Corso, J.F., The Experimental Psychology of Sensory Behavior, Holt, Rinehart and Winston, New York, 1987.

Craig, A, and Cooper, R.E., Symptoms of acute and chronic fatigue, in Handbook of Human Performance, Academic Press, New York, 1992.

Cuttle, C., and Brandston, H., Evaluation of retail lighting, Journal of the Illuminating Engineering Society, 24, 33-49, 1995.

Czeisler, C.A. and Allen J.S., Acute circadian phase reversal in man via bright light exposure: Application to jet-lag, Sleep Res., 16, 605, 1987.

Czeisler, C.A., Johnson, M.P., Duffy, J.F., Brown, E.N., Ronda, J.M., Kronauer, R.E., Exposure to bright light and darkness to treat physiologic maladaptation to night work, New England Journal of Medicine, 322, 1253-1259, 1990.

Czeisler, C.A., Chiasera, A.J., and Duffy, J.F., Research on sleep, circadian rhythms and aging: Application to manned spaceflight, Environmental Gerontology, 26, 217-232, 1991.

Czeisler, C.A., Shanahan, T.L., Klerman, E.B., Martens,H., Brotman, D.J., Emens, J.S., Klein, T., and Rizzo, J.F.III., Suppression of melatonin secretion in some blind patients by exposure to bright light, New England Journal of Medicine, 332, 6-11, 1995.

Dawson, D., and Campbell, S.S., Bright light treatment: Are we keeping our subjects in the dark, Sleep, 13, 267-271, 1990.

Devaney, K.O., and Johnson, H.A., Neuron loss in the aging visual cortex of man, Journal of Gerontology, 35, 836-841, 1980.

Dijk, D.J., Boulos, Z., Eastman. C.I., Lewy, A.J., Campbell, S.S. and Terman, M., Light treatment for sleep disorders: Consensus report II Basic properties for circadian physiology and sleep regulation, J. Biol. Rhythms, 10, 113-125, 1995.

Donovan, R.J., and Rossiter, J.R., Store atmosphere: An environmental psychology approach, Journal of Retailing, 58, 34-57. 1982.

References

11-7

Drasdo, N., The neural representation of visual space, Nature, 266, 554-556, 1977.

Drew, G.C., An Experimental Study of Mental Fatigue, Flying Personnel Research Committee Report 227, His Majesty’s Stationary Office, London, 1940.

Drury, C.G., Inspection of sheet material - model and data, Human Factors, 17, 257, 1975.

Dureman, E.I., and Boden, C., Fatigue in simulated car driving, Ergonomics, 15, 299, 1972.

Eastman, C.I., Circadian rhythms and bright light recommendations for shift work, Work and Stress, 4, 245-260, 1990.

Eastman, C.I., Stewart, K.T., Mahoney, M.P., Liu, L., and Fogg, L.F., Dark goggles and bright light improve circadian rhythm adaptation to night-shift work, Sleep, 17, 535-543, 1994.

Eastman, C.I., Boulos, Z., Terman, M., Campbell, S.S., Dijk, D.J. and Lewy, A.J., Light treatment for sleep disorders: Consensus report VI Shift work, J. Biol. Rhythms, 10, 157-164, 1995.

Elton, P.M., A study of output in silk weaving during winter months, Industrial Fatigue Research Board Report No. 9, His Majesty’s Stationary Office, London, 1920.

Eklund, N.H., Exit sign recognition for color normal and color deficient observers, Journal of the Illuminating Engineering Society, 28, 71-81, 1999.

Eklund, N.H., Boyce, P.R., and Simpson, S.N., Lighting and sustained performance: Sustained performance under three lighting installations, TR-112693, EPRI, Palo Alto, CA, and U.S. Environmental Protection agency, Washington DC, 1999.

Eklund, N.H., Boyce, P.R., and Simpson, S.N., Lighting and sustained performance: modeling data-entry task performance, TR-114913, EPRI, Palo Alto, CA, and U.S. Environmental Protection Agency, Washington, DC: 2000.

Engel, F.L., Visual conspicuity, directed attention and retinal locus, Vision Research, 11, 563, 1971.

Engel, F.L., Visual conspicuity, visual search, and fixation tendencies of the eye, Vision Research, 17, 95, 1977.

Floyd, W.F., and Welford, A.T., Fatigue, H.K. Lewis & Co. Ltd, London, 1953.

Flynn, J.E., A study of subjective responses to low energy and non-uniform lighting systems, Lighting Design and Application, 7, 6-15, 1977.

Flynn, J.E., Hendrick, C., Spencer, T.J., and Martyniuk, O., Interim study of procedures for investigating the effect of light on impression and behavior, Journal of the Illuminating Engineering Society, 3, 87-94, 1973.

References

11-8

Flynn, J.E., Hendrick, C., Spencer, T.J., and Martyniuk, O., A guide to methodology procedures for measuring subjective impressions in lighting, Journal of the Illuminating Engineering Society, 8, 95-110, 1979.

Folkard, S. and Monk, T.H., Shiftwork and performance, Human Factors, 21, 483-492, 1979.

Forger, D.B., Jewett, M.E., and Kronauer, R.E., A simpler model of the human circadian clock, J.Biol. Rhythms, 14, 532-537, 1999.

Foster, R.G., Provencio, I., Hudson, D., Fiske, S., De Grip, W., and Menaker, M., Circadian photoreception in the retinally degenerate mouse (rd/rd), J. Compo. Physiol. A 169, 39-50, 1991.

French, J., Hannon, P. and Brainard, G.C., Effects of bright illuminance on body temperature and human performance, Annual Review of Chronopharmacology, 7, 37-40, 1990.

Froberg, J.E., Sleep deprivation and prolonged working hours, in Hours of Work, Eds: S Folkard, and T.H.Monk, John Wiley and Sons Ltd., New York, 1985.

Fry, G.A., Ocular discomfort and other symptoms of eyestrain at low levels of illumination, in NIOSH Symposium: The occupational safety and health effects associated with reduced levels of illumination, Cincinnati, US Department of Health, Education and Welfare 1974.

Gifford, R., Light, decor, arousal, comfort and communication, Journal of Environmental Psychology,. 8, 177-189, 1988.

Glass, R.A., Howett, G.L., Lister, K., and Collins, B.L., Some criteria for colors and signs at workplaces, NBS IR83-2694, National Bureau of Standards, Washington, DC, 1983.

Gorn, G.J., The effects of music in advertising on choice behavior: A classical conditioning approach, Journal of Marketing, 46, 94-101, 1982.

Graeber, R.C., Sing, H.C., and Cuthbert, B.N., The impact of transmeridian flight on deploying soldiers, in The Twenty-four Hour Workday: Proceedings of a Symposium on Variations in Work-Sleep Schedules, Eds, L.C. Johnson, D.I. Tepas, W.P. Coquhoun, and M.J. Colligan, Department of Health and Human Services (NIOSH), Cincinnati, OH, 1981.

Grandjean, E., Fitting the Task to the Man: an Ergonomic Approach, Taylor and Francis, London, 1969.

Groos, G.A. and Mason, R., The visual properties of rat and cat suprachiasmatic neurones, J. Comp. Physiol A, 135, 349-356, 1980.

Groos, G.A., and Meijer, J.A., Effects of illumination on suprachiasmatic nucleus electrical discharge, in The Medical and Biological Effects of Light. Eds. R.J. Wurtman, M.J. Baum and J.T. Potts, Jr., Annals of the New York Academy of Sciences, 453, 135-146, 1985.

Hartung, J., Light, puberty and aggression, Human Ecology, 6, 273, 1978.

References

11-9

Harvey, L.O., DiLaura, D.L., and Mistrick, R.G., Quantifying reactions of visual display operators to indirect lighting, Journal of the Illuminating Engineering Society, 14, 515-546, 1984.

Hawkes, R.J., Loe, D.L., and Rowlands, E., A note towards the understanding of lighting quality, Journal of the Illuminating Engineering Society, 8, 111-120. 1979.

Heschong-Mahone Group, Skylighting and Retail Sales: An Investigation into the Relationship between Daylighting and Human Performance, Pacific Gas and Electric Company, Sacramento, CA., 1999a.

Heschong-Mahone Group, Daylighting in Schools: An Investigation into the Relationship between Daylighting and Human Performance, Pacific Gas and Electric Company, Sacramento, CA., 1999b.

Hockey, G.R.J., Signal probability and spatial location as possible bases for increased selectivity in noise, Quarterly Journal of Experimental Psychology, 22, 37, 1970.

Hockey, G.R.J., Stress and Fatigue in Human Performance, John Wiley and Sons, Chichester, 1983.

Holding, D., Loeb, M., and Baker, M., Effects and aftereffects of continuous noise and computation work on risk and effort choices, Motivation and Emotion, 7, 331, 1983.

Houpt, T.A., Boulos, Z., and Moore-Ede, M.C., Midnight sun: Software for determining light exposure and phase-shifting schedules during global travel, Physiol. Behav., 59, 561-568, 1996.

Howarth, C.I., and Bloomfield, J.R., A rational equation for predicting search times in simple inspection tasks, Psychonomic Science, 17, 225, 1969.

Inditsky, B., Bodmann, H.W., and Fleck, H.J., Elements of visual performance: Contrast metric - visibility lobes - eye movements, Lighting Research and Technology, 14, 218-231, 1982.

Illuminating Engineering Society of North America (IESNA), The IESNA Lighting Handbook, 9th Edition, Ed. M.S. Rea, IESNA, New York, 2000.

Isen, A.M., and Baron, R.A., Affect as a factor in organizational behavior, in Research in Organizational Behavior, Eds., B.M. Staw and L.L. Cummings, JAI Press, Greenwich, CT, 1991.

Jacobs, R., and Krohn, D.L., Variations in fluorescence characteristics of intact human crystalline lens segments as a function of age, Journal of Gerontology, 31, 641-647, 1976.

Jerome, C.W., The rendering of ANSI safety colors, Journal of the Illuminating Engineering Society, 6, 180 - 183, 1977.

References

11-10

Jewett, M.E., Kronauer, R.E., and Czeisler, C.A., Light induced suppression of endogenous circadian amplitude in humans, Nature, 350, 59-62, 1991.

Johnson, C.A., and Casson, E.J., Effects of luminance, contrast and blur on visual acuity, Optometry and Visual Science, 72, 864-869, 1995.

Kaplan, S., Aesthetics, affect and cognition: Environmental preference from an evolutionary perspective, Environment and Behavior, 19, 2-32, 1987.

Khek, J., and Krivohlavy, J., Evaluation of the criterion to measure the suitability of visual conditions, Proceedings of the CIE, 16th Session, Washington, CIE, Paris, 1967.

Klein, D.C., Moore, R.Y., and Reppert, S.M., Suprachisamatic Nucleus: The Mind’s Clock, Oxford University Press, Oxford, UK, 1991.

Klein, K.E., and Wegmann, H.M., The resynchronization of human circadian rhythms after transmeridian flights as a result of flight direction and mode of activity, in Chronobiology, Eds, L.E. Scheving, F.Halberg, and J.E. Pauly, Igaku Shoin Ltd., Tokyo, 1974.

Klein, K.E., and Wegmann, H.M. Significance of Circadian Rhythms in Aerospace Operations, AGARD Monograph No. 247, Advisory Group for Aerospace Research and Development, NATO, London, 1980.

Knez, I., Effects of indoor lighting on mood and cognition, Journal of Environmental Psychology, 15, 39-51, 1995.

Knoblauch, K., Saunders, F., Kusada, M, Hynes, R., Podger, M., Higgins, K.E., and de Monasterio, F.M., Age and illuminance effects in the Farnsworth-Munsell 100 hue test scores, Applied Optics, 26, 1441-1448, 1987.

Kogi, K., and Saito, Y., Rhythmic fluctuation of orientation to a continuous manual control task, J. Human Ergol. 2, 169, 1973.

Kosnik, W., Winslow, L., Kline, D., Rasinski, K., and Sekular,, R., Visual changes in daily life throughout adulthood, Journal of Gerontology, 43, 63-70, 1988.

Kronauer, R.E., A quantitative model for the effects of light on the amplitude and phase of the deep circadian pacemaker, based on human data, in Sleep “90, Proceedings of the Tenth European Congress on Sleep Research ed.

Ed. J.Horne, Pontenagel, Dussseldorf, 1990.

Kronauer, R.E., Forger, D.B., and Jewett, M.E., Quantifying human circadian pacemaker response to brief, extended and repeated light episodes over the photopic range, J. Biol. Rhythms, 14, 500-515, 1999.

References

11-11

Lack, L., and Wright, H., The effect of evening bright light in delaying the circadian rhythms of lengthening the sleep of early morning awakening insomniacs, Sleep, 16, 436-443, 1993.

LaGiusa, F.F., and Perney, L.R., Further studies on the effects of brightness variations on attention span in a learning environment, Journal of the Illuminating Engineering Society, 3, 249-252, 1974.

Lam, R.W., and Levitt, A.J., Canadian Consensus Guidelines for the Treatment of Seasonal Affective Disorder, The Canadian Journal of Diagnosis, 1999.

Landsberger, H.A., Hawthorne Revisited: “Management and the Worker”, its Critics and Developments in Human Relations in Industry, Cornell University, Ithica, NY, 1958.

Lerman, S., Lens transparency and aging, In Aging of the Lens, Eds: F.Regnault, O. Hockwin, and Y.Courtois, Elsevier / North Holland, Amsterdam, 1980.

Lighting Research Center, Lighting Answers: Multilayer Polarizing Panels, Lighting Research Center, Troy, New York, 1993.

Lipinski, M.E. and Shelby, B.L., Visibility measures of realistic roadway tasks, Journal of the Illuminating Engineering Society, 22, 94-101, 1993.

Lloyd, C., Boyce, P., Ferzacca, N., Eklund, N., and He, Y., Paint inspection lighting: Optimization of lamp width and spacing, Journal of the Illuminating Engineering Society, 28, 92-102, 1999.

Lythgoe, R.J., The measurement of visual acuity, Medical Research Council Special Report 173, His Majesty’s Stationary Office, London, 1932.

MacAdam, D.L., Visual sensitivities to color differences in daylight, Journal of the Optical Society of America, 32, 247-274, 1942.

McCann, J.J., and Hall, J.J., Effect of average-luminance surrounds on the visibility of sine wave gratings, Journal of the Optical Society of America, 70, 212-219, 1980.

McCloughan, C.L.B., Aspinall, P.A., and Webb, R.S., The impact of lighting on mood, Lighting Research and Technology, 31, 81-88, 1999.

McGrath, C and Morrison, J.D., The effects of age on spatial frequency perception in human subjects, Quarterly Journal of Experimental Physiology, 253-261, 1981.

McGuiness , P.J., and Boyce, P.R., The effect of illuminance on the performance of domestic kitchen work by two age groups, Lighting Research and Technology, 16, 131-136, 1984.

McIntyre, I.A., Norman, T.R., Burrows, G.D., and Armstrong, S.M., Human melatonin suppression by light is intensity dependent, Journal of Pineal Research, 6, 149-156, 1989a.

References

11-12

McIntyre, I.A., Norman, T.R., Burrows, G.D., and Armstrong, S.M., Quantal melatonin suppression by exposure to low intensity light in man, Life Sciences, 45, 327-332, 1989b.

McNelis, J.F., Human performance - a pilot study, Journal of the Illuminating Engineering Society, 2, 190-196, 1973.

Mackworth, N.H., The breakdown of vigilance during prolonged visual search, Quarterly Journal of Experimental Psychology, 1, 6, 1948.

Main, A., Dowson, A., and Gross, M., Photophobia and phonophobia in migraineurs between attacks, Headache, 37, 492-495, 1997.

Mangum, S.R., Effective constrained illumination of three-dimensional, light-sensitive objects, Journal of the Illuminating Engineering Society, 27, 115-131, 1998.

Maniccia, D., Rutledge, B., Rea, MS, and Morrow, W., Occupant use of manual lighting controls in private offices, Journal of the Illuminating Engineering Society, 28, 42-56, 1999.

Megaw, E.D., and Richardson, J., Eye movements and visual inspection, Applied Ergonomics, 10, 145, 1979.

Meijer, J.H., and Rietveld, W.J., Neurophysiology of the suprachiasmatic circadian pacemaker in rodents, Physiol.Rev., 69, 671-707, 1989.

Menaker, M., Commentary: What does melatonin do and how does it do it, J. Biol. Rhythms, 12, 532-534, 1997.

Milliman, R.E., Using background music to affect the behavior of supermarket shoppers, Journal of Marketing, 46, 85-91, 1982.

Mistrick, R.G., Outline of a procedure for the development of a visual performance model, Report to the Illuminating Engineering Society of North America, IESNA, New York, 1996.

Miyamoto, Y., and Sancar, A., Vitamin B2-based blue-light photoreceptors in the retinalhyothalmic tract as the photoactive pigments for setting the circadian clock in mammals, Proc. Natl. Acad. Sci, USA, 95, 6097-6102, 1998.

Mollon, J.D., “Tho’ she kneel’d in that place where they grew...” The uses and origins of primate colour vision, Journal of Experimental Biology, 146, 21-38, 1989.

Monk, T.H., Knauth, P., Folkard, S., and Rutenfranz, J., Memory based performance measures in studies of shiftwork, Ergonomics, 21, 819-826, 1978.

Moore, R.Y., Speh, J.C., and Card, J.P., The retino-hypothalamic tract originates from a distinct subset of retinal ganglion cells, J. Comp. Neurol., 352, 351-366, 1995.

References

11-13

Moore-Ede, M.C., Sulzman, F.M., and Fuller, C.A., The Clocks That Time Us, Harvard University Press, Cambridge, MA, 1982.

Mrosovsky, N., Reebs, S.G., Honrado, G.I., and Salmon, P.A., Behavioural entrainment of circadian rhythms, Experientia Basel, 45, 696-702, 1989.

Muck, E., and Bodmann, H.W., Die bedeutung des beleuchtungsniveaus bei praktische sehtatigkeit, Lichttechnik, 13, 502, 1961.

Myers, B.L., and Badia, P., Changes in circadian rhythms and sleep quality with aging: Mechanisms and interventions, Neuroscience and Biobehavioral Reviews, 19, 553-571, 1995.

Narendran, N., Vasconez, S., Boyce, P., and Eklund, N., Just-perceivable color difference between similar light sources in display lighting applications, Journal of the Illuminating Engineering Society, 29, 68-77, 2000.

Owsley, C., Sekuler, R., and Siesen, D., Contrast sensitivity throughout adulthood, Vision Research, 23, 689-699, 1983.

Page, R.A., and Moss, M.A., Environmental influences on aggression: The effect of darkness and proximity of victim, Journal of Applied Social Psychology, 6, 124, 1976.

Parsons, H.M., What happened at Hawthorne?, Science, 183, 922-932, 1974.

Poffenberger, A.T., The effects of continuous work upon output and feelings, Journal of Applied Psychology, 12, 459, 1928.

Provencio, I., Cooper, H.M., and Foster, R.G., Retinal projections in mice with inherited retinal degeneration: Implications for circadian photoentrainment, J. Comp. Neurol., 395, 417-439, 1998.

Pu, M., Physiological response properties of the cat retinal ganglion cells projecting to the suprachiasmatic nucleus, J. Biol. Rhythms, 15, 31-36, 2000.

Purcell. A., and Nasar, J., Experiencing other people’s houses: A model of similarities and differences in environmental experience. Journal of Environmental Psychology, 12, 199-212, 1992.

Rea, M.S., Visual performance with realistic methods of changing contrast, Journal of the Illuminating Engineering Society, 10, 164-177, 1981.

Rea, M.S., Behavioral responses to a flexible desk luminaire, Journal of the Illuminating Engineering Society, 13, 174-190, 1983.

Rea, M.S., The validity of the relative contrast sensitivity function for modeling threshold and suprathreshold responses, Proceedings of the Integration of Visual Performances Criteria Into the Illumination Design Process, Public Works Canada, Ottawa, pp. 76-108, 1984.

References

11-14

Rea, M.S., Toward a model of visual performance: Foundations and data, Journal of the Illuminating Engineering Society, 15, 41-58, 1986.

Rea, M.S., Toward a model of visual performance: A review of methodologies, Journal of the Illuminating Engineering Society, 16, 128-142, 1987.

Rea, M.S. and Ouellette, M.J., Visual performance using reaction times, Lighting Research and Technology, 20, 139-153, 1988.

Rea, M.S. and Ouellette, M.J., Relative visual performance: A basis for application, Lighting Research and Technology, 23, 135-144, 1991.

Rea, M.S., Ouellette, M.J., and Kennedy, M.E., Lighting and task parameters affecting posture, performance and subjective ratings, Journal of the Illuminating Engineering Society, 15, 231-238, 1985.

Rea, M.S., Ouelette, M.J., and Tiller, D.K., The effects of luminous surroundings on visual performance, pupil size and human preference, Journal of the Illuminating Engineering Society, 19, 45-58, 1990.

Reitmaier, J., Some effects of veiling reflections in papers, Lighting Research and Technology, 11, 204-209, 1979.

Roenneberg, T., and Aschoff, J., Annual rhythm of human reproduction: 1 Biology, sociology or both?, J. Biol. Rhythms, 5, 217-239, 1990a.

Roenneberg, T., and Aschoff, J., Annual rhythm of human reproduction: II Environmental considerations, J. Biol. Rhythms, 5, 217-239, 1990b.

Roenneberg, T., and Foster, R.G., Twilight times: Light and the circadian system, Photochemistry and Photobiology, 66, 549-561, 1997.

Roethlisberger, F.J., and Dickson, W.J., Management and the Worker, Harvard University Press, Cambridge, MA, 1939.

Rosenthal, N.E., Sack, D.A., Skwerer, R.G., Jacobsen, F.M., and Wehr, T.A., Phototherapy for seasonal affective disorder, J. Biol. Rhythms, 3, 101-120, 1988.

Ruberg, F.L., Skene, D.J., Hanifin, J.P., Rollag, M.D., English, J., Arendt, J., and Brainard, G.C., Melatonin regulation in humans with color vision deficiencies, J. Clin. Endocrino. Metab., 81, 2980-2985, 1996.

Russell, J.A., and Snodgrass, J., Emotion and the environment, in Handbook of Environmental Psychology, Eds: I Altman and D.Stokol, John Wiley and sons, New York, 1987.

Sanders, P.A., and Bernecker, C.A, Uniform veiling luminance and display polarity affect VDU user performance, Journal of the Illuminating Engineering Society, 19, 113-123, 1990.

References

11-15

Sanders, M, Gustanski, J., and Lawton, M., Effect of ambient illumination on noise level of groups, Journal of Applied Psychology, 59, 527 - 528, 1974.

Sasaki, M., Kurosaki, Y., Onda, M., Yamaguchi, O., Nishimura, H., Kashimura, K., and Graeber, R.C., Effects of bright light on circadian rhythmicity and sleep after transmeridian flight, Sleep Res., 18, 442, 1989.

Schonpflug, W., Coping efficiency and situational demands, in Stress and Fatigue in Human Performance, Ed. G.R.J. Hockey, John Wiley and Sons, Chichester, 1983.

Sekular, R., and Blake, R., Perception, McGraw-Hill, New York, 1994.

Shingledecker, C.A. and Holding, D.H., Risk and effort measures of fatigue, Journal of Motor Behavior, 6, 17, 1974.

Shurcliff, W.A., Polarized Light, Harvard University Press, Cambridge, MA, 1962.

Simonson, E. and Brozek, J., Effects of illumination level on visual performance and fatigue, Journal of the Optical Society of America, 38, 384, 1948.

Slater, A.L., Perry, M.J., and Crisp, V.H.C., The applicability of the CIE visual performance model to lighting design, Proceedings of the CIE 20th Session, Amsterdam, CIE Paris, 1983.

Smith, S.W., Performance of complex tasks under different levels of illumination, Part 1 - needle probe task, Journal of the Illuminating Engineering Society, 5, 235-242, 1976.

Smith, S.W., and Rea, M.S., Proofreading under different levels of illumination, Journal of the Illuminating Engineering Society, 8,47-52, 1978.

Smith, S.W. and Rea, M.S., Relationships between office task performance and ratings of feelings and task evaluations under different light sources and levels, Proceedings of the CIE, 19th Session, Kyoto, Japan, pp 207-210, 1979.

Smith S.W. and Rea, M.S., Performance of a reading test under different levels of illumination, Journal of the Illuminating Engineering Society, 12, 29-33, 1982.

Smith, S.W., and Rea, M.S., Check value verification under different levels of illumination, Journal of the Illuminating Engineering Society, 16, 143-149, 1987.

Snow, C.E., Research on industrial illumination, The Tech Engineering News, 8, 257, 1927.

Stenzel, A.G., Experience with 1000 lx in a leather factory, Lichttechnik, 14, 16, 1962.

Stenzel, A.G., and Sommer, J., The effect of illumination on tasks which are largely independent of vision, Lichttechnik, 21, 143, 1969.

References

11-16

Steward, J.M., and Cole, B.L., What do color defectives say about everyday tasks, Optometry and Visual Science, 66, 288-295, 1989.

Stone, P.T. and Groves, S.P.D., Discomfort glare and visual performance, Trans. Illum. Engng Soc, (London), 33, 9, 1968.

Stone, P.T., and Harker, S.P.D., Individual and group differences in discomfort glare responses, Lighting Research and Technology, 5, 41-49, 1973.

Taubes, G, Epidemiology faces its limits, Science, 269, 164-169, 1995.

Taylor, L.H., and Socov, E.W., The movement of people towards lights, Journal of the Illuminating Engineering Society, 3, 237-241, 1974.

Terman, M., Lewy, A.J., Dijk, D.J., Boulos, Z., Eastman, C.I., and Campbell, S.S., Light treatment for sleep disorders: Consensus report IV Sleep phase and duration disturbances, J. Biol. Rhythms, 10, 135-147, 1995.

Thapan, K., Arendt, J. and Skene D.J., An action spectrum for melatonin suppression: Evidence for a novel non-rod, non-cone photoreceptor system in humans, Journal of Physiology, 535.1, 261-267, 2001.

Tilley, A.J., Wilkinson, R.T., Warren, P.S.G., Watson, B., and Drud, M., The sleep and performance of shift workers, Human Factors, 24, 629-641, 1982.

Timmers, An effect of contrast on legibility of printed text, Institute of Perception Annual Report, IPO Eindhoven, Netherlands, 1978.

Urwick, L., and Brech, E.F.L., The Making of Scientific Management:, Vol. 3, The Hawthorne Investigations, Pitmans, London, 1965.

van den Berg, T.J.T.P., Quantal and visual efficiency of fluorescence in the lens of the human eye, Investigative Ophthalmology and Vision Science, 34, 3566-3573, 1993.

van den Berg, T.J.T.P., IJspeert, J.K., and de Waard, P.W.T., Dependence of intraocular straylight on pigmentation and transmission through the ocular wall, Vision Research, 31, 12361-1367, 1991.

Van Derlofske, J.F., Bierman, A., Rea, M.S. and Maliyagoda, N., Design and optimization of a retinal exposure detector, Proceedings of SPIE, 4092, 60-70, 2000.

Van Reeth, O., Sturis, J., Byrne, M.M., Blackman, J.D., L’Hermite-Balriaux, M., Leproult, R., Oliner, C., Retetoff, S., Turek, F.W., and Van Cauter, E., Nocturnal exercise phase delays circadian rhythms of melatonin and thyrotropin secretion in normal men, American Journal of Physiology, 266, E964-E974, 1994.

References

11-17

Veitch, J.A., and Kaye, S.M., Illumination effects on conversational sound levels and job candidate evaluation, Journal of Environmental Psychology, 8, 223-233, 1988.

Veitch, J.A. and Newsham, G.R., Lighting quality and energy-efficiency effects on task performance, mood, health, satisfaction and comfort, Journal of the Illuminating Engineering Society, 27, 107-129, 1998a.

Veitch, J.A. and Newsham, G.R., Determinants of lighting quality 1: State of the science, Journal of the Illuminating Engineering Society, 27, 92-106, 1998b.

Vingrys, A.J. and Cole, B.L., Are colour vision standards justified for the transport industry, Ophthalmic and Physiological Optics, 8, 257-274, 1988.

Vos, J.J., Disability glare - state of the art report, CIE Journal, 3, 39, 1984.

Vos, J.J., Age dependence of glare effects and their significance in terms of visual performance, in Lighting for Aging Vision and Health, Ed: W. Adrian, Lighting Research Institute, New York, 1995.

Vos, J.J., and Boogaard, J., Contribution of the cornea to entoptic scatter, Journal of the Optical Society of America, 53, 869-873, 1963.

Vrabel, P.L., Bernecker, C.A., and Mistrick, R.G., Visual performance and visual clarity under electric light sources: Part 1 - visual performance, Journal of the Illuminating Engineering Society, 24, 69-80, 1995.

Waldhauser, F., and Dietzel, M., Daily and annual rhythms in human melatonin secretion: Role in puberty control, in the Medical and Biological Effects of Light, Eds, R.J.Wurtman, M.J.Baum, and J.T.Potts Jr., Annals of the New York Academy of Sciences, 205-214, 1985.

Warren, N., and Clarke, B., Blocking in mental and motor tasks during a 65 hour vigil, Psychological Bulletin, 33, 814, 1936.

Waters, I. and Loe, D.L., Visual performance in illumination of differing spectral quality, Visual Performance Research Project, University College Environmental Research Group, University College, London, 1973.

Weale, R.A., The Aging Eye, H.K.Lewis, London, 1963.

Weale, R.A., Human lenticular fluorescence and transmissivity and their effects on vision, Experimental Eye Research, 41, 457-473, 1985.

Weale, R.A., The Senescence of Human Vision, Oxford University Press, Oxford, 1992.

Wehr, T.A., Melatonin and seasonal rhythms, J. Biol. Rhythms, 12, 518-527, 1997.

References

11-18

Wehr, T.A., Giesen, H.A., Moul, D.E., Turner, E.H., and Schwartz, P.J., Suppression of human responses to seasonal changes in day-length by modern artificial lighting, Am. J. Physiol., 269, R173-R178, 1995.

Welford, A.T., Brown, R.A., and Gabb, J.E., Two experiments on fatigue as affecting skilled performance in civilian aircrew, British Journal of Psychology, 40, 195, 1950.

Werner, J.S. and Kraft, J.M., Color vision senescence: Implications for lighting design, in Lighting for Aging Vision and Health, Ed: W. Adrian, Lighting Research Institute, New York, 1995.

Werner, J.S., Peterzell, D.H., and Scheetz, A.J., Light, vision and aging, Optometry and Vision Science, 67, 214-229, 1990.

Werner, J.S., and Steele, V.G., Sensitivity of human foveal color mechanisms throughout the life span, Journal of the Optical Society of America, A5, 2122-2130, 1988.

Weston, HC, A study of the efficiency in fine linen-weaving, Industrial Fatigue Research Board Report No. 20, His Majesty’s Stationary Office, London, 1922.

Weston H.C., The relation between illumination and visual efficiency: The effect of size of work, A report prepared for the Industrial Health Research Board and the Medical Research Council, H.M.Stationary Office, London, 1935.

Weston, H.C., The relation between illumination and visual efficiency: the effect of brightness contrast, Industrial Health Research Board, Report No. 87, H.M. stationary Office, London, 1945.

Weston, H.C., On age and illumination in relation to visual performance, Trans. Illum. Engng. Soc. (London), 16, 281-297, 1949.

Weston, HC, and Taylor, SK, The relation between illumination and efficiency in fine work (typesetting by hand), Final Report of the Industrial Fatigue Research Board and the Illumination Research committee (DSIR), His Majesty’s Stationary Office, London, 1926.

Wetterberg, L., Light and Biological Rhythms in Man, Pergamon Press, New York, 1993.

Wibom, R.I., and Carlsson, W., Work at visual display terminals among office employees: visual ergonomics and lighting, in Work with Display Units, 86, Eds. B. Knave and P.G. Wideback, North Holland, Amsterdam, 1987.

Wiggle, R., Gregory, W., and Lloyd, C.J., Paint inspection lighting, Proceedings of the International Body Engineering Conference, Detroit, MI, Society of Automotive Engineers (SAE) Paper 982315, 1997.

Wilkins, A. J., Visual Stress, Oxford University Press, Oxford, 1991.

References

11-19

Wilkins, A. J., Reading and visual discomfort, in Visual Process in Reading and Reading Disabilities, Eds: D.M.Willows, R.S.Kruk, and E.Corcos, Lawrence Erlbaum Associates, Hillsdale, NJ, 1993.

Wilkins, A.J., Nimmo-Smith, I, Slater, A.J., and Bedocs, L., Fluorescent lighting, headaches and eyestrain, Lighting Research and Technology, 21, 11, 1989.

Wilkinson, R.T. , Some factors influencing the effect of environmental stress on performance, Psychological Bulletin, 72, 260-272, 1969.

Williams, L.C., The effect of target specification on objects fixated during visual search, Perception and Psychophysics, 1, 315, 1966.

Wolf, E. and Gardiner, J.S., Studies on the scatter of light in the dioptric media of the eye as a basis of visual glare, Archives of Ophthalmology, 74, 338-345, 1965.

Wyon, D.P., Indoor environmental effects on productivity, Proceedings of the Indoor Air Quality 96 Conference, ASHRAE, Atlanta, 1996a.

Wyon, D.P., Creative thinking as a dependent variable in six environmental experiments: A review, Proceedings of the Indoor Air Quality 96 Conference, ASHRAE, Atlanta, 1996b.

Wyszecki, G.and Stiles, W.S., Color Science: Concepts and methods, quantitative data and formulae, 2nd Edition, John Wiley and Sons, New York, 1982.

Yorks, P and Ginthner, D., Wall lighting placement: Effect on behavior in the work environment, Lighting Design and Application, 17, 30-37, 1987.

Zeitzer, J.M., Dijk, D-J., Kronauer, R.E., Brown, E.N., and Czeisler, C.A., Sensitivity of the human circadian pacemaker to nocturnal light: melatonin phase resetting and suppression, Journal of Physiology, 526, 695-702, 2000.

© 2001 Electric Power Research Institute (EPRI), Inc.All rightsreserved. Electric Power Research Institute and EPRI are registeredservice marks of the Electric Power Research Institute, Inc.EPRI. ELECTRIFY THE WORLD is a service mark of the ElectricPower Research Institute, Inc.

Printed on recycled paper in the United States of America

1006415

Target:

Emerging Issues and Technologies ImpactingMarket Transformation and Energy Efficiency

EPRI • 3412 Hillview Avenue, Palo Alto, California 94304 • PO Box 10412, Palo Alto, California 94303 • USA800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

About EPRI

EPRI creates science and technology solutions for

the global energy and energy services industry. U.S.

electric utilities established the Electric Power

Research Institute in 1973 as a nonprofit research

consortium for the benefit of utility members, their

customers, and society. Now known simply as EPRI,

the company provides a wide range of innovative

products and services to more than 1000 energy-

related organizations in 40 countries. EPRI’s

multidisciplinary team of scientists and engineers

draws on a worldwide network of technical and

business expertise to help solve today’s toughest

energy and environmental problems.

EPRI. Electrify the World

lighting and human performance ii

Documents

member of epri

epri representativehas

epri customer fulfillment

corporate document

unified humanfactors

palo alto

iiicitationsthis report

performance1006415final