Lighting Res. Technol. 2013; 45: 124-132

Flicker can be perceived during saccades atfrequencies in excess of 1 kHzJE Roberts MSc and AJ Wilkins DPhilDepartment of Psychology, University of Essex, Colchester, UK

Received 30 November2011; Revised 31 December 2011; Accepted 1 January 2012

When driving at night, flickering automobile LED tail lights can appear as multipleimages. The perception of a flickering source of light was therefore studied duringrapid eye movements Isaccades) of 20-40 0 amplitude in an otherwise dark room!<1Iuxl. The temporal modulation appeared as a spatial pattern known as a'phantom array' during the saccade. The appearance of the pattern enabled thediscrimination of flicker from steady light at frequencies that in 11 observersaveraged 1.98 kHz. At a frequency of 120 Hz, the intrasaccadic pattern wasperceptible when the contrast of the flicker exceeded 10%. It is possible thatintrasaccadic stimulation interferes with ocular motor control.

1, Introduction

When driving at night behind a car with LEDtail lights, it is possible to experience a trail oflights with each rapid movement of the eyes(saccade). The trail occurs because the LEDlamps are lit intermittently to control heatbuild-up. The frequency of operation is aboveflicker fusion, varies from one manufacturerto another and may not be the same for bothtail lamps, to judge from videos posted on theweb. I Each flash of the lamp is imaged on adifferent part of the retina during the sac­cades, forming a pattern dubbed a 'phantomarray' by Hershberger and Jordan. 2

Normally, flicker is not perceived at fre­quencies greater than the so-called criticalflicker fusion frequency. 'Classic data fromKelly for modulation of large (30 0 radius)visual fields at different li/fht levels from0.03cd·m-2 to 5000cd·m-- showed thatbeyond lOO Hz, even flicker with 100% mod­ulation was hardly ever directly visible, eithercentrally or peripherally.,3 Critical flicker

Address for correspondence: A Wilkins, Department orPsychology, University of Essex. Colchester C04 3SQ. UKE-mail: arnold@csscx..ac.llk

© The Chartered Institution of Building Services Engineers 2012

fusion frequency is, however, usuallymeasured with a spatially unstructured field.It is rarely measured during eye movements,and when it is, estimates of the criticalfreq uency increase.4 When each flash of aspatially structured light source is imaged ona different part of the retina during a rapideye movement, the light source forms a'phantom array', and the above example oftail light flicker shows that the array issometimes perceptible under everyday view­ing conditions.

The chat forums contain complaints ofdiscomfort from the 'phantom array' whendriving at night. Some drivers are aware ofthe phenomenon, others are not. In an initialsmall-scale survey, respondents (20 studentsand staff at the University of Essex) wereasked the following question verbatim: 'Whenyou are driving at night and the car in front isa new one with long red lights, if you changethe direction you are looking does anythinghappen?' Despite the deliberately non-specificwording of the question, three of the 20respondents reported a 'trail of lights'. Thispaper shows that the perception of flicker asintrasaccadic patterns is possible at frequen­cies in the kilohertz range, more than 10 times

10.1177/1477153512436367

Flicker pal/erns above 1kHz during saccades 125

the typical estimates of critical flicker fusionfrequency.

Normally, we are unaware of the retinalimage during a saccade. Mechanisms ofsaccadic suppression thought to be largelycentral and cortical in origins.6 prevent theprocessing of the image during the saccade.The structured images before and after thesaccade normally mask the intrasaccadicimage. However, as the example of tail lightflicker shows, there are circumstances inwhich the pre- and post-saccadic maskingfails because at night the image before andafter the saccade is spatially unstructured. Itmay then be difficult to distinguish intrasac­cadic percepts such as the phantom arrayfrom those percepts that occur before andafter the saccade.

During steady gaze (with microsaccadesand drifts), spatially repetitive patterns can bedistinguished at spatial frequencies as high as40-50 cycle·deg- and can most readily bedetected at spatial frequencies of around 4cycle.deg-17 During a saccade, the angularvelocity of the eye reaches a peak that can beas high as 700 deg.s- I depending on theamplitude of the saccade.8 It might thereforebe possible in principle to distinguish f1icker­induced intrasaccadic patterns at frequenciesof up to 50 x 700 = 35 kHz. Two consider­ations make this unlikely. First, contrastsensitivity may be reduced to some extentduring a saccade. Volkmann et al.9 measuredthe ability to perceive the contrast of a gratingduring saccades. The gratings were flashed for10 ms to the steadily fixating eye and alsoduring 6° horizontal saccades. Contour mask­ing before and after the saccade was mini­mised by a diffuse unpatterned field of view.Relative to contrast sensitivity during steadyfixation, contrast sensitivity during the sac­cade was reduced by a factor of more thanfour. More recently, however, Garcia-Perezand Peli 10 have investigated contrast sensitiv­ity for spatial patterns that appear only

during a saccade and have shown that visualcontrast processing is largely unaltered duringsaccades. Some of the reduction observed byVolkman et alY may have occurred becausethe saccade trajectory was not perfectlyaligned with the stripes of the grating, therebyreducing the contrast of the retinal imageduring the saccade. Indeed, this might explainwhy Volkmann et al. observed that thereduction in sensitivity increased as the spatialfrequency of the gratings increased. Be that asit may, the physical reduction in contrast dueto poorly aligned eye trajectories is irrelevantwhen flicker is perceived as an intrasaccadicpattern. If there is a reduction in contrastsensitivity during a saccade, it will reduce themaximum frequency at which flicker can beperceived. If contrast sensitivity is reduced bya factor of four, then, on the basis of theclassic data from Campbell and Robson,7 themaximum perceptible spatial frequencywould be reduced from 48 cycle.deg- I to 38cycle·deg- I

, a relatively modest change.The second consideration that might make

a 35 kHz limit unlikely is the integration timeof the retinal cells. Retinal cells integrate theirsignals over a period of time during whichintensity and duration trade off against eachother. Bloch's law of temporal summationstates that 1:,1 x I:,T=constant where 1:,1 andI:, T are the light pulse intensity and temporalduration, respectively. At low light energylevels with large high-velocity saccades andhigh-frequency flicker, there may be insuffi­cient variation in energy during the flight ofthe eye to stimulate the retina cells. However,Sam Berman in a memorandum to the IEEEPARl789 committee dated 11 January 2011combined data from Hart I I with that fromFukuda4 and showed .that typical lightsources can provide the luminance incrementof sufficient magnitude such that temporalsummation would not occur until I:,T wasgreater than approximately 'h msec, corre­sponding to 2 KHz flicker frequency.

Lighrillg Res. Tee/mol. 2011; 45: 124-132

126 I E Roberls and Al Wilkins

Most of the existing literature on the'phantom array' has been concerned withissues surrounding the location of the intra­saccadic percep,ts jn relation to the timing ofthe saccade, _-I, although Hershbergerel al. 'l reported that a phantom array wasvisible by most of their subjects even at500 Hz, the maximum frequency they used.This observation raises the possibility of (I)visibility of the intrasaccadic percept at fre­quencies greater than 500 Hz and (2) individ­ual differences in its perceptibility. Thepurpose of this study was, therefore, toinvestigate in a range of observers the upperfrequency limits below which flicker can beperceived as a phantom array and abovewhich no such perception is possible. Wechose viewing conditions likely to enhanceintrasaccadic perception in order to provideworst-case estimates of use to engineersresponsible for designing signal lights.

2. Experiment 1: 400 saccades with 1, 2, 3and 5 kHz flicker

2.1. Method2.1.1. Apparatus

An analogue oscilloscope (ISO-TECHISR620) was operated on its side so thatwhen the time base was set at its maximum of21ls per sweep of the screen (0.5 MHz), avertical green line was visible (CrE chroma­ticity x= 0.307, y =0.529). The z-input of theoscilloscope controlled the brightness of theline and was connected to the output of acomputer-controlled function generator(Velleman PCSUIOOO 2 MHz). The functiongenerator was programmed to generate a sinewave at a frequency of I, 2, 3 or 5 kHz or asteady signal with similar time-averaged volt­age and luminance. The luminance of the linevaried over time between a minimum of0.02cd·m-2 and a maximum of 3l0cd·m-2

.

The oscilloscope display was visible through acircular aperture (diameter 50 mm) in a matt

Lighting Res. Tecllllol. 2011; 45: 124-132

black screen (width 2 m) on which were twowhite discs that served as fixation points.These were positioned 600 mm horizontally tothe left and right of the line and were visibleas a result of the low ambient lighting of theroom « I lux).

2.1.2. Panicipal/lsParticipants, four males and seven females

aged 22-65 years, mean 34 years, with normalor corrected to normal visual acuity wererecruited from staff, students and acquain­tances of the authors at the University ofEssex. Three wore corrective lenses during thetesting. One female, aged 26, used a greenoverlay when reading (but not for theexperiment).

2.1.3. Pracedl/reParticipants were seated 1.7 m from the

display, which was at eye level. The line wasilluminated for 3 s in two immediately succes­sive trials, during which participants wererequired to make saccades back and forthbetween the targets, and then decide in whichof the two presentations, a pattern of spatiallyperiodic lines could be seen. On one of thetrials selected at random, the function gener­ator modulated the brightness of the line; onthe other trial, the line was not modulated.Ten such pairs of trials, separated by a 3 sinterval, were given at each frequency in aconstant random order.

2.2. ResultsThe average data are presented in Figure 1.

A cumulative normal was fitted to the datausing least squares. The 75% threshold basedon the fitted curve was 1.67 kHz (standarddeviation 0.52 kHz). A cumulative normalwas also fitted to each individual participant'sdata using least squares and the mean of theindividual thresholds was 1.98 kHz (standarddeviation 1.13 kHz).

Flicker pal/ems above 1kHz during saccades 127

4. Interim discussion

3. Experiment 2: corroboration usingdifferent apparatus

Figure 1 Experiment 1: Percentage of trials in whichflicker was correctly detected. shown as a function offlicker frequency

Light from a 12 V tungsten halogen lampcontrolled by a stabilised DC supply wasdirected via an optic fibre to the aperture of aPrinceton Applied Research (Oak Ridge, TN,USA) light chopper (model 187). The end ofthe optic fibre measured I mm wide by 7 mmhigh and was observed by the authors from adistance of I m through the rotating shutter ofthe chopper in an otherwise darkened room«I lux). When the shutter was operated atI kHz, the intrasaccadic pattern was invari­ably perceived; at 2 kHz, it was occasionallyperceived and at 3 kHz, the pattern could notbe distinguished.

5.1. MethodThe method was the same as for

Experiment I except that the fixation pointswere repositioned 300 mm horizontally to theleft and right of the light. Three participantsfrom Experiment I aged 25-38 took part.

5. Experiment 3: 20° saccades with 1,2,3and 5 kHz flicker

the estimate of 2 cycle.deg- ' based on (I) thetemporal frequency of the line, (2) the ampli­tude of the saccade ~40°) and (3) the peakvelocity of 500 deg·s- for a saccade of thisamplitude expected on the basis of the dataquoted in Leigh and Zee. '6 Halving theamplitude of the saccade to 20° would beexpected to change the velocity to 400deg·s- 1

and increase the spatial frequency of theintrasaccadic pattern from 2 cycle.deg- I to2.5 cycle·deg- at I kHz. At 2 kHz, the changewould be from 4 cycle·deg-I to 5 cycle.deg- I

and at 3 kHz from 6 cycle·deg- I to 7.5cycle.deg- I

. On the basis that the contrastsensitivity function peaks at about 4cycle.deg- I and the fact that the thresholdwas between I kHz and 2 kHz, one mightanticipate that the threshold frequency atwhich the intrasaccadic pattern is perceptiblewould be little affected by reducing thesaccade amplitude to 20°. On the otherhand, the amount of saccadic suppressionhas been shown to increase monotonicallywith the size of a saccade. I? Experiment 3 wastherefore designed to determine the thresholdfor a saccade of 20° amplitude.

65

•

2 3 4Frequency (kHz)

100

8013~00

60C'"~'"a.

40

200

Using the apparatus from Experiment I, theauthors estimated the spatial frequency of theintrasaccadic pattern by counting the numberof lines in a known angular displacement atlOO Hz intervals between lOO Hz and 500 Hz.The data were used to estimate the spatialfrequency at I kHz. The average estimate of1.8 cycle.deg- I agreed reasonably well with

5.2. ResultsFigure 2 shows mean data for the three

participants. The 75% threshold based on themeans of the group was 2.35 kHz (standarddeviation 0.42 kHz). The mean of the individ­ual thresholds was 2.47 kHz (standard devia­tion 0.57 kHz). The thresholds did not differconsistently or significantly from those inExperiment I.

Lighting Res. Teclmol. 2011; 45: 124-132

128 JE Roberts and AJ Wilkins

•

100~- ~-----------,

6. Interim discussion

7. Experiment 4: modulation thresholdsat 120Hz

headache. If so, the conditions known tohave these untoward effects (e.g. fluorescentlighting) might be those that sustain theintrasaccadic percept and those not so asso­ciated (e.g. incandescent lighting) be thosethat fail to sustain the intrasaccadic percept.This possibility was investigated in the nextstudy.

Both fluorescent and incandescent lightingflicker at twice the frequency of the electricitysupply, but the modulation depth of thevariation in light from an incandescent lamp(defined as (Lmax - Lm;n)/(Lmax +Lm;n)) isusually less than 10%. The modulation fromfluorescent lighting varies between about 5%and 100% depending on the control circuitryand the phosphors23 and was typicallybetween 30% and 50% until high-frequencyelectronic ballasts became widespread. Thefollowing experiment was designed to deter­mine the modulation depth at which theintrasaccadic pattern from flicker at 120 Hzbecomes perceptible.

7.2. ResultsThe average data are presented in Figure 3.

A cumulative normal was fitted to the datausing least squares. The 75% threshold basedon the fitted curve was 10% (standard devi­ation 7.2%). A cumulative normal was alsofitted to each individual participant's data

7.1. MethodThe method was the same as in Experiment

1 except that the z-modulation of the oscillo­scope had a square-wave luminance profileand a frequency of 120 Hz. The modulationdepth defined as (Lmax - Lm;n)/(Lmax + Lm;n)was 5%, 10%, 20% or 40%. A total of 40trials were presented, 10 for each modulationdepth, selected in random order. Four partic­ipants from Experiment 1 aged 25-65, mean40, took part.

65234Frequency (kHz)

20.J--~-~-~-~--~-----1

o

U80

~0<.>

60C'"\,>'"0-

40

Figure 2 Experiment 3: Percentage of trials in whichflicker was correctly detected, shown as a function offlicker frequency

As anticipated on the basis of the spatialfrequency of the intrasaccadic pattern, theamplitude of the saccade in the range 20--400

had little effect on the upper frequency limitat which the intrasaccadic pattern could beperceived.

Flicker at frequencies of 100 Hz and 120 Hz(above flicker fusion) is known to interferewith the control of eye movements. 18.19

Flicker from fluorescent lighting at 100 Hzand 120 Hz is known to impair visual perfor­mance20 and cause headaches in some indi­viduals20

.21 Brundrett22 showed that

individuals who complained of discomfortfrom fluorescent lighting had a higher criticalflicker fusion frequency than those who didnot, and corroborated these observationselectrophysiologically. One of the participantsin Experiment I used a green overlay to treatvisual stress when reading. This participanthad a threshold of 4.9 kHz, well above that ofthe other observers. The intrasaccadic per­ception of flicker as an anomalous pattern inthe form of a phantom array provides aputative mechanism for the interference witheye movements, and the possibly consequentdisruption of visual performance and

Lighfil1g Re.\". Tee/mol. 2011; 45: 124~132

Flicker palleflls above I kHz during saccades 129

100 • •90

13ID580 -<.>

C~70ID"-

60·

50.'---__- __-_-__-,

o 5 10 15 20 25 30 35 40Percent modulation

Figure 3 Experiment 4: Percentage of trials in whichflicker at 120 Hz was correctly detected, shown as afunction of the modulation of the flicker

using least squares and the mean of theindividual thresholds ranged between 3%and 14% with a mean of 9% (standarddeviation 4%).

8. Interim discussion

The average modulation depth at whichflicker at 120 Hz gave rise to a perceptibleintrasaccadic pattern was 10% and thereforegreater than the modulation depth of incan­descent lighting and lower than the modula­tion depth typical of gas discharge lightingand some LED lighting. The modulation wassquare-wave. The abrupt change in brightnessmay have made the pattern more perceptiblethan it would have been had the modulationbeen sinusoidal. Gas discharge lighting andLED lighting can have abrupt transitionsreminiscent of those in a square wave.

With a saccade of 400 amplitude andvelocity of about 500deg.s-', the spatialfrequency of the intrasaccadic pattern ata frequency of 120 Hz is only about0.24 cycle.deg- ' , well below the peak ofthe contrast sensitivity function at ~4

cycle·deg- 17 It is therefore possible that theintrasaccadic pattern would be perceived at

lower modulation with a smaller and there­fore slower saccade because the spatial fre­quency would then be higher and closer to thepeak of the contrast sensitivity function. Thesmallest saccades are the involuntary micro­saccades that have velocities of about50 deg.s- ' .'4 and the smallest voluntary sac­cades are probably those that occur duringreading, which are typically 1_20 in amplitudewith a velocity that ranges from 80 deg.s- l to120deg·s- 1 (S. Jainte, personal communica­tion). With velocities such as these, onlyabout one cycle of a pattern would beavailable during the saccade, and the visibilityof a grating is known to decrease when fewerbars are present'S When the authors used a10 saccade, they were unable to see anyperiodic intrasaccadic pattern at 120 Hz evenwhen the modulation was 100%.

9. General discussion

The neurological effects of saccadic suppres­sion appear in motion-sensitive neurons 26

The flashed presentation during a saccademight be expected to have little effect on anymechanisms of suppression that are due toimage motion. This may be one reason whythe intrasaccadic stimulation was clearlyperceived.

The above experiments were undertakenlooking at a light source in a darkened room.The measurements, therefore, apply to con­ditions in which light sources such as signallamps are visible in low ambient lighting. Oneexample of such conditions is that with whichthis paper began: driving at night behind taillights of cars. These are conditions in whichthe apparent displacement of the location ofthe lights due to intrasaccadic perception canbe a distraction - one that mayor may notimpact on road safety.

Recently, it has been shown that thehorizontal spatial periodicity of words inter­feres with reading by prolonffng the timetaken for vergence movements.- It is possible

Lighting Res. Tee/mol. 2011; 45: 124-132

130 JE Roberts and AJ Wilkills

that the intrasaccadic simulation of spatialperiodicity may interfere further with thisprocess and help to explain the effect of high­frequency flicker in disturbing eye movementcontrol, particularly during reading. ,s.19 Themeasurements obtained here reflect the con­scious perception of spatial periodicity, andit is possible that noil-conscious effects occurat parameter ranges beyond those exploredhere.

The perception of the phantom arraywould appear to be predictable from consid­erations of the spatial frequency, contrastand extent (number of cycles) of the intra­saccadic stimulation. With this in mind, it isof interest that the contrast limit for percep­tion of the array at twice the electricitysupply frequency (120 Hz) averaged 10%,which is above the modulation typical ofincandescent lighting but below that typicalfor gas discharge and LED lighting. In otherwords, even under conditions in which theintrasaccadic stimulation ('phantom array')is maximally visible, as here, a modulationdepth of 10% is insufficient for its reliableperception. It remains to be seen whether thecomplaints of discomfort from gas dischargelighting can be attributed to the 'phantomarray' (perceptible or otherwise) and anyresultant interference with saccadic control.

The experimental conditions used in thisstudy might be expected to enhance theperceptibility of the intrasaccadic pattern,because after the retina was exposed to aflash, little further light reached the stimu­lated location on the retina. In a well-litroom, the contrast of the image of each flashwould presumably be reduced by the addi­tional light reaching the part of the retinastimulated by the flash later in the saccade.The perceptibility of the intrasaccadic patternmight, therefore, be less when a dark line isviewed on a white background that flickers.Bullough et 01.3 have reported the percepti­bility of flicker when subjects observed aminimally contoured wall of a room lit with

Lighting Res. T(·cJmo/. 2011; 45: 124-132

flickering light from an LED luminaire andmade a 40° saccade between two distanttargets; 30% were able to detect flicker at300 Hz. In a subsequent study,2S participantswere required to wave a white rod and reportany stroboscopic effects. Under these condi­tions, the upper frequency limit was stillhigher. The participants in the originalstudy3 did not include those with migraineor epilepsy, and it is possible that the greatercortical hyperexcitability in these partici­pants29 would render the flicker more visible.Our observation that an observer who usedcoloured overlays for reading was able toperceive the intrasaccadic pattern at higherfrequencies than those for the remainder ofthe sample is of interest because colouredfilters have been shown to reduce corticalhyperactivation in migraine.3D

In summary, temporally modulatinglighting can result in intrasaccadic percep­tion of spatial periodicity for parameterranges in which the lighting has beenassociated with impaired ocular motorcontrol, impaired visual performance, dis­comfort and headaches. The upper fre­quency limit of such effects is well abovethe frequency at which the modulation canbe appreciated as flicker.

Funding

This research received no specific grant fromany funding agency in the public, commercial,or not-for-profit sectors.

Acknowledgements

We thank Sam Berman and Brad Lehman forideas that contributed to this study and NinaR Wolinski for conducting the survey con­cerning car tail lights.

Flicker pat/erns abo!'e I kHz during saccades 131

References

Strange tail light effect. Retrieved 29December 2011, from http://www.youtube.­com/wa tch?v=TIACrzzleos.

2 Hershberger WA, Jordan JS. The phantomarray: a pcrisaccadic illusion of visualdirection. Psychological Record 1998; 48:21-32.

3 Bullough JD, Sweater Hickcox K, Klein TR,Narendran N. Effects of nicker characteristicsfrom solid-state lighting on detection, accept­ability and comfort. Lightillg Research alldTechllology 20 11; 43: 337-348.

4 Fukuda T. Erfeet of eye movement on per­ception of nicker. Perceptual alld Motor Skills1979; 48: 943-950.

5 Brcitmcyer BO, Ganz L. Implications of sus­tained and transient channels for theories ofvisual pattern masking, saccadic suppression,and information processing. PsychologyRe!'iew 1976; 83: 1-36.

6 Chahine G, Krekelberg B. Cortical contribu­tions to saccadic suppression. PLoS Olle 2009;4(9): e6900.

7 Campbell FW, Robson JG. Applicationof Fourier analysis to the visibility of gratings.Joul"l1al of Physiology 1968; 197: 551-566.

8 Eizenman M, Frecker RC. Precise noncon­tacting measurement using the corneal reOex.Visioll Research 1984; 24: 167-174.

9 Volkmann FC, Riggs LA, White KD, MooreRK. Contrast sensitivity during saccadic eyemovements. Vision Research 1978; t8:1193-1198.

10 Garcia-Perez MA, Pcli E. Visual contrastsensitivity is largely unaltered during saccades.Frontiers in Psychology: Perception Science2011;2:247.

11 Hart Jr WM. The temporal responsiveness ofvision. In Moses RA, Hart WM, editors.Adler's Physiology of the Eye, ClillicalApplication. SI. Louis, MO: The C. V. MoshyCompany, 1987,429-457.

12 Hershberger WA, Jordan JS, Lueas DR.Visualizing the perisaccadic shift of spatiotopiccoordinates. Perception and Psychophysics1998; 60: 82-88.

13 Jordan JS, Hershberger WA. Timing the shiftin retinal local signs that accompanies a

saccadic eye movement. Perception andPsychophysics 1994; 55: 657-666.

14 Noritake A, Vagi A. Is the phantom array anevidence for Discrete-EPI model? Joul"llal ofVision 2002; 2: Article 165.

15 Watanabe J, Noritake A, Maeda T, Tachi S,Nishida S. Perisaccadic perception of contin­uous nickers. Vision Research 2005; 45:413-430.

16 Leigh RJ, Zee DS. The Neurology of EyeMo!'emems, 3rd edn, Oxford: OxfordUniversity Press, 1999.

17 Stevenson SB, Volkmann FC, Kelly JP, RiggsLA. Dependence of visual suppression on theamplitudes of saccades and blinks. VisionResearch 1986; 26: 1815-1824.

18 Wilkins AJ. Intermittent illuminationfrom visual display units and OUOI'es­cent lighting affects movements of theeyes across text. Human FaClOrs 1986; 28:75-81.

19 Kennedy A, Murray W. The effects of nickeron eye movement control. QuarterlyJoul"I1al of Experimemal Psychology. Sectioll AHuman Experimental Psychology 1991; 43:79-99.

20 Veitch J, McColl S. Modulation of nuoreseentlight: Flicker rate and light source effects onvisual performance and visual comfort.Lightillg Research and Techllology 1995; 27:243-256.

21 Wilkins AJ, Nimmo-Smith IM, Slater A,Bedoes L. Fluorescent lighting, headaches andeye-strain. Lighting Research and Technology1989; 21: 11-18.

22 Brundrett GW. Human sensitivity to flicker.Lighting Research alld Technology 1974; 6:127-143.

23 Wilkins AJ, CIark C. Modulation of light fromfluorescent lamps. Liglllillg Research andTechnology 1990; 23: 103-109.

24 Dimigen 0, Valsecchi j\1, Sommer W, KlieglR. Human microsaccade-related visual brainresponses. Journal oJ Neuroscience 2009; 29:12321-12331.

25 Kelly DH. How many bars make a grating?Vision Research 1975; 15: 625-626.

26 Thiele A, Hellning P, Kubischik M, HoffmannK-P. Neural mechanisms of saccadic suppres­sion. Science 2002; 295: 2460-2462.

Lighting Res. Tee/lIlol. 201 [; 45: [24-132

132 JE Roberls and AJ Wilkins

27 Jainta S, Jaschinski W, Wilkins AJ. Periodicletter strokes within a word alTect fixationdisparity during reading. Journal of Visioll2010; 10: 2, 1-\1.

28 BlIllollgh JD, Sweater Hickeox K, Klein TR,Lok A, Narcndran N. Detection and accept­ability of slroboscopic effects from nicker.Lighting Research and Technology 2011; inpress.

Lighting Res. Tee/lllol. 2011; 45: 124-132

29 Aurora SK, Wilkinson F. The brain ishyperexcitable in migraine. Cephalalgia 2007;27: 1442-\453.

30 HlIang J, Zong X, Wilkins A, Jenkins B,Bozoki A, Cao Y. FM RI evidence that preci­sion ophthalmic tints reduce cortical hypcr­activation in migraine. Cephalalgia 2011; 31:925-936.