ED 106 659

AUTHORTITLE

INSTITUTIONSPONS AGENCY

REPORT NOPUB DACENOTE

EDRS PRICEDESCRIPTORS

DOCUMENT RESUME

Haygood, Robert C.; And OthersVisual and Auditory Informationthe Acquisition of Flying Skill.Arizona State Univ., Tempe.Air Force Human Resources Lab.,Flying Training Div.AFERL-TR-74-79Dec 7463p.

CE 003 856

Processing Aspects of

Williams AFB, Ariz.

MF-S0.76 HC-S3.32 PLUS POSTAGE*Aircraft Pilots; Auditory Discrimination; AuditoryPerception; Aural Stimuli; *Flight Training;*Learning Modalities; Models; Sensory Experience;*Sensory Integration; Sensory Training; *SkillDevelopment; Stimulus Behavior; VisualDiscrimination; Visual Stimuli

ABSTRACTThe result of a number of experieaental studies of

human auditory and visual information processing behavior and theirpossible relationship to the student pilot's acquisition of flyingskill were explored in terms of a conceptual model developed for thisstudy. The results were interpreted in terms of the potentialinterfering effects of the intake of and response to informationprocessed during flying tasks and in terms of the student pilot'snonoptimal information processing strategies during his acquisitionof flying skill. It was concluded that the experimental proceduresemployed could be adapted sucessfully for research in the area andthat the relationships found between information procesing and flyingskill warranted their further study. (Author)

AFHRL-TR-74-79

IR FORCE

P

H

UMAN

RE

S0URCE

S

AP...R 0 7 1975

VISUAL AND AUDITORY INFORMATION PROCESSINGASPECTS OF THE ACQUISITION OF FLYING SKILL

BY

Robert C. HaygoodBarry Leibowitz

Stanley R. Parkinson

Arizona Stale University

TEN IPA Arizona .62111

Edward E. Eddowes

FLYING TRAINING DIVISIONWilliams Air Force Base, Arizona 85224

U SDEPARTMENT OP

NEALTN,

EDUCATIONWELFARE

NATIONALINSTITUTE°,

EDUCATION

THIS DOCUMENTHAS SEEN REPRO

DUCEDEXACTLY AS RECEIVED

FROM

THE PERSONOR ORGANIZATION

ORIGIN

AI ING ITPOINTS OF

VIEW OR OPINIONS

STATED DO NOT NECESSARILYREPRE

SENT OFFICIALNATIONAL INSIII

UTE OF

EDUCATIONPOSITION OR POLICY

December 1974

Approved for public release; distribution unlimited.

LABORATORY

AIR FORCE SYSTEMS COMMAND2 BROOKS AIR FORCE BASE,TEXAS 78235

NOTICE

When US Government drawings, specifications, or other dataare used for any purpose other than a definitely relatedGovernment procurement operation, the Government therebyincurs no responsibility nor any obligation whatsoever,and the fact that the Government may have formulated, fur-nished, or in any way supplied the said drawings, specifi-cations, or other data is not to be regarded by implicatiinor otherwise, as in any manner licensing the holder or anyother person or corporation, or conveying any rignts orpermission to manufacture, use, or sell any patented in-vention that may in any way be related thereto.

This final report was submitted by Arizona State University,Tempe, Arizona 85281, under contract F41609-72-C-0037, pro-ject 1138, with Flying Training Division, Air Force HumanResources Laboratory (AFSC), Williams Air Force Base,Arizona 85224. Dr. Merrilyn J. Penner and Dr. Edward E.Eddowes, Flying Training Division, shared the contractmonitors hip.

This report has been reviewed and cleared for open publica-tion and/or public release by the appropriate Office ofInformation (0I) in accordance with AFR 190-17 and DoDD5230.9. There is no objection to unlimited distributionof this report to the public at large, or by DDC to theNational Technical Information Service (NTIS).

This technical report has been reviewed and is approved.

WILLIAM V. HAGIN, Technical DirectorFlying Training Division

Approved for publication.

HAROID E. FISCHER, Colonel, USAFComminder

3

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE Mon Dees Entered)

REPOR1 DOCUMENTATION PAGEREAD INSTRUCTIONS

BEFORE COMPLETING FORM

I. REPORT NUMBER

AFHRL-TR-74-79

2. GOVT ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBER

4. TITLE (and Subtitle)

VISUAL AND AUDITORY INFORMATION PROCESSING ASPECTS

OF THE ACQUISITION OF FLYING SKILL

S. TYPE cr nEPORT 4 PERIOD COVERED

Final

6. PERFORMING ORG. REPORT NUMBER

7. AUTHOR(a)Robert C. Haygood Stanley R. ParkinsonBarry 1Leshowitz Edward E. Eddowes

B. CONTRACT OR GRANT NUMBER(s)

F41609-72-C-0037

M. PERFORMING ORGANIZATION NAME AND ADDRESS

Arizona State UniversityTempe, Arizona 85281

10. PROGRAM ELEMENT. PROJECT. TASKAREA 4 WORK UNIT NUMBERS

61102E11380102

I1. CONTROLLING OFFICE NAME AND ADDRESS

Flying Training DivisionAir Force Human Resources LaboratoryWilliams Air Force Base, Arizona 85224

12. REPORT GATE

December 197413. NUMBER OF PAGES

62

14. MONITORING AGENCY NAME It ADDRESS(if different from Controlling Offree)

Hq Air Force Human Resources Laboratory (AFSC)Brooks Air Force Base, Texas 78235

IS. SECURITY CLASS. (of this report)

Unclassified

ISa. DECLASSIFICATION /DOWNGRADINGSCHEDULE

IS. DISTRIBUTION STATEMENT (of this Report)

Approved for public release; distribution unlimited.

t7. DISTRIBUTION STATEMENT (of the abstract entered In Block 20, II different Wet Report)

HI. SUPPLEMENTARY NOTES

IS. KEY WORDS (Continuo on reverse side if necessary and Identify by block number)

information processing modelvisual information processingauditory information processinginformation processing in flyinginformation processing skills

20. ABSTRACT (Continuo on reverse aide II necessary and Identify by block number)

The result of a number of experimental studies of human auditory and visual information proc behavior

and their possible relationship to the student pilot's acquisition of flying skill were explored in terms o onceptualmodel developed for this study. The results were interpreted in terms of the potential interfering effects of the

intake of and response to information processed during flying tasks and in terms of the student pilot's nonoptimal

information processing strategies during his acquisition of flying skill. It was concluded that the experimental

procedures employed could be adapted successfully for research in the area and that the relationships found betweeninformation processing and flying skill warranted their further study.

nr, FORMin. I .1" 73 1.1,7Aioa EDITION OF I NOV 61 IS OBSOLETE

Y'

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE (Mien Gel* Sneered)

SUMMARY

Problem

Research on pilot skills since World War II has fo-cused, generally, on the perceptual-motor components ofthe flying task. Recently, a conception of flying skillwhich emphasizes the pilot's information seeking and in-formation processing-functions has evolved which supersedesand incorporates the older view of flying skill as hand-eyecoordination. An information processing model was developeeto provide the basis for the present experimental analysisof flying skill. Thus, the objective of this study was touse the model to identify the information processing as-pects of the pilot's flying task and to relate them to the

student pilot's acquisition of flying ability.

Approach

An information processing analysis of the pilot's taskwas developed through review of the literature in the areasof human auditory and visual information processing, andtask analyses of the pilot's aircraft system managementfunctions. Subsequently, experiments were designed andexecuted to isolate information processing skills sensi-tive to individual differences and to determine the extentof their variability, to explore means for controlling andaccounting for such individual variability, and to ascer-tain the degree to which these information processing tasksrelate to and may be applied in improving pilot training.

Results

In a series of experiments, the importance of the fol-lowing auditory and visual information processing variableswere demonstrated: stimulus duration, stimulus similarityinterference, response-induced interference, attention,digit span, and scanning strategies. A series of studiesof audiovisual ccncept formation demonstrated that withsimple problems, auditory or visual information was equally

1

effective when scanning time was unlimited, that visualpictorial information was more effective than visualverbal !nformation when scan time was severely limited,and that there was no measurable effect of audiovisualredundancy on concept attainment performance over the rangeof test tasks studied.

Conclusions and Recommendations

The results of this exploratory investigation of in-formation processing.aspects of the pilot's flying taskindicated that experimental procedures developed could beemployed to probe further the nonoptimal information pro-cessing strategies of student pilots and to evolve moreeffective flying training methods. The results suggestthe high potential of further research in this area forimproving Air Force flying training.

2

PREFACE

This report represents a portion of the research pro-gram of Project 1138, Perceptual Motor and CognitiveComponents of the Flying Task, Dr. William V. Hagin,Project Scientist; Task 113801, Cognitive Components ofthe Flying Task, Dr. Edward E. Eddowes, Task Scientist,being carried out by the Flying Training Division, AirForce Human Resources Laboratory, Williams Air Force Base,Arizona. It is a systhesis of the results of ten experi-mental studies of human information processing accomplish-ed by the Department of Psychology, Arizona State University,Tempe, Arizona, under contract F41609-72-C-0037 with theAir Force Human Resources Laboratory. Dr. Robert C.Haygood was the Principal Investigator for Arizona StateUniversity. Dr. Merrilynn J. Penner and Dr. Edward E.Eddowes were Contract Monitors for the Air Force HumanResources Laboratory.

3

TABLE OF CONTENTS

Page

I. Introduction 7

II. The Concept of Information 8

Discrete Event Signals 11Continuous Level Signals 11Analog (Pictorial) Representation 11Logical Relationships ...; 12

Human Information Processing 13

Categories of Information Processing 13Information Storage 14Executive Processor 17Capacity Limitations 17

IV. Identification of Information ProcessingComponents of Flying Skill 17

The Information Environment 18The Normal Landing Pattern 20

V. Research Program 22

VI. The Processing of Auditory and Visual Signals 22

Measurement of Auditory Processing:Stimulus Interference 23Experimental Variables AffectingProcessing 24

Measurement of Auditory Processing:Response-Induced Interference 26Measurement of Visual Processing: APartial Report Recognition Paradigm 28Experimental Variables AffectingVisual Processing 29

4

Page

VII. Short-Term Retention of Auditory

Information 31

The Dichotic Paradigm 32

Attended and Unattended Melsages 32

Verbal and Written Monitors 33

Digit Span 33

Interference Effects in DichoticMemory 34

Digit Span and Interference 35

VIII. Audiovisual Concept Formation 35

Concept Identification in ThreeModalities 39

Comparison of Visual and AuditoryVerbal Presentation 40

The Effects of Limiting Inspectionand Response Time 41

Audiovisual Redundancy 43

Redundancy with Nonconjunctive Concepts 44

IX. Implications of Information Processing for

Flying Training 45

Visual and Auditory Scanning andInterference 45

Auditory-Verbal Information Processing 47

Audiovisual Concept Formation 48

Conclusion 49

X. References 50

5

LIST OF ILLUSTRATIONS

Figure

1 Model of the Fovironmental CompositeSystem During Flying Training

LIST OF TABLES

Table

Page

19

1 Distribution of Instances into Positiveand Negative Categories for Four BasicConceptual Rules 38

6

VISUAL AND AUDITORY INFORNATIOd PROCESSINGASPECTS OF THE ACQUISITION OF FLYING SKILL

I. INTRODUCTION

Basic research on flying skills since the second worldwar has focused primarily on the study of perceptual-motortasks such as pursuit and error tracking (Birmingham &Taylor, 1954; Briggs, 1962) and the efficiency of displayreading (Sleight, 1948). Such research efforts reflectedthe simplistic view of flying as an integrated sequence ofscanning and control movement responses. While such anapproach may have adequately characterized flying skillsome twftaty-five years ago, it is simply inapplicable tothe piloting of modern high speed aircraft. The adventof highly sophisticated navigation, flight- contrc,l, andweapons control systems has produced a dramatic change in

the pilot's role.

The military pilot of today functions primarily as anexecutive decision maker; that is, he must be able to pro-cess incoming information rapidly and accurately, and tomake crucial decisions based upon such data. The dynami-cally changing nature of the aircraft operating environmentdemands great refinement of basic information processingskills much beyond those perceptual-motor responses pre-viously considered to reflect flying skill. The relation=ship of the pilot to his informational environment--to theenvironmental sources from which he obtains informationand the information he outputs to his environment--providesthe basis for the present approach to flying skill.

The information processing paradigm would appear toprovide an heuristic conceptual framework for the experi-mental study of flying skill. In contrast to earlierorientations which have stressed the development of skillas a build-up of habits--either expressed as connectionsbetween stimulus and response, or as a build-up of responsetendencies resulting from reward--the present approachviews the organism as an information processing system.That is, the individual dynamically interacts with the

7

It

environment by gathering appropriate cue information andgenerating outputs *.n, accordance with some processingalgorithm. It shouts be readily apparent that behavioralcomponents such as short-term memory, visual scan, andconcept identification are of importance in the efficientprocessing of information. While much effort has beendevoted to the experimental investigation of these areas,there has occurred little integration of relevant findingsand even less application tc practicAl problems. It wouldappear that the virtue of the informaticn processing modelis to provide a conceptual framework, whereby the complexinterrelationships among various behavioral components canbe identified and programmatically investigated. Rigiddistinctions among sensory processes, perception, learning,memory, and verbal/motor skills do not provide the levelof integration necessary for an adequate description ofbehavior in the dynamically changing environment of themodern pilot. However, the concept of information flow- -from the energy environment through receptor systems, de-coding, gating, shunting, and short-term buffer storage,to organized long-term storage, retrieval, encoding andcomparator to language or motor response generation andfeedback guidance and monitoring--does seem to offer auseful framework for the study of flying skill.

The goal of the present effort was to clarify the roleof information processing variables in the acqui3ition offlying skill. To achieve this end, a combined analytic andexperimental program was undertaken to: (1) identifysignificant information-processing skills required in theoperation of aircraft, (2) review existing literature onlaboratory studies of information processing; and (3) de-sign and execute the necessary experiments to advanceexisting knowledge in directions relevant to pilot train-ing.

II. THE CONCEPT OF INFORMATION

At the outset, it seems necessary to precisely definewhat is meant by information. Typically, one thinks ofthe mathematical theory of communication, now commonlycalled information theory (Shannon & Weaver, 1949). Ac-

8

cordingly, information theory defines a source from which

messages originate, a channel through which they pass, and

a receiver which accepts and accumulates messages. The

channel has two important characteristics. First, it has

a limited capacity and second, it is subject to noise which

distorts the message. An attempt is made to quantify theinformation content of a message in terms of the number ofalternative messages that might have been sent and theprobabilities of the individual messages. The greater thenumber of alternatives, the greater is the receiver's un-certainty about which message will be received. The basicunit of information, the bit, represents the amount of un-certainty in the choice between two equally probable al-ternatives. The classic example is the single bit of un-certainty in the outcome of the toss of an unbiased coin,uncertainty which is resolved by receiving the informationthat the outcome was "heads" or "tails."

As the number of possible alternative messages in-

creases, so does the uncertainty. With equally likelyalternatives, the amount of uncertainty is the logarithm,to the base two, of the number of alternatives. Thus withfour alternatives, there are two bits of uncertainty; witheight alternatives, three bits, and so forth. The alter-natives need not be equally likely; if they have different

probabilities, the less probable alternatives convey moreinformation, but the average information for the whole setis reduced. in the limiting case, a message telling anoutcome that is completely certain (that is, an outcomewhich is already known because it was received before orin some other way) conveys no information. Such a message

is said to be redundant. Any message that conveys lessthan the maximum possible information is said to be partial-

ly redundant.

Much research during the past two decades has beenconcerned with the evaluation of information-theory con-structs as tools for understanding human behavior (cf.

Miller, 1956; Garner, 1962). Despite the original enthu-siasm, (Quastler, 1955), the theory and its constructs have

not lived up to their initial promise, and interest has

waned noticeably. Despite its utility in the analysis ofcommunication networks, computers, and guessing games, in-

formation theory provides at best an incomplete model ofhuman information processing. In particular, human com-munication, especially that communication which we char-acterize as instruction, cannot be described adequately inthe language of information theory.

First, except in the most trivial of cases, it isnever possible to specify the population of alternativemessages from which a particular message is drawn. Thelistener, told to standby for a radio communication ofgreat importance, may or may not have in his expectationsthe message that is subsequently sent.

Second, information theory does not deal with thequestion of meaning, which is the essential ingredient inhuman communication. No matter how well the message isreceived, it conveys no information unless the readerunderstands the words themselves. For example, much com-munication in the training situation is, in fact, devotedto the transmission of the meanings by means of definitionsand combinations of words, phrases, and other symbolicexpressions.

Third, the theory does not take into account the op-erating characteristics of the human receiver who, unlikethe idealized receiver of information theory, deliberatelyselects, rejects, filters, transforms, and forgets incominginformation.

Fourth, information theory does not address itself tothe question of the importance of messages. The variousmessages in a set are usually not of equal importance orconcern to the receiver, and the various sets of messagesthemselves differ in value.

And finally, although not directly related to humancommunication, information theory does not permit thequantification of the information content of continuoussignals. Intuitively, complex auditory signals seeminformationally richer than sine waves, but there is nodirect method of measurement.

It is apparent that information theory does not pro-vide a comprehensive definition of information. Consider-

ing the limitations which have been pointed out, a taxonomywould seem to be more appropriate. Input information tothe human operator can be categorized into several broadclasses.

1. Discrete Event Signals. This category is closestin meaning to the "information" of information theory.Three distinct classes of discrete event signals may bedefined, although there are cases in which considerableoverlap may exist.

a. Abstract (Arbitrary) Event Signals. With thistype of signal, there is no intrinsic relationship betweenthe signal itself and the event signaled. Instead, the

relationship is arbitrary. Much of aircraft communicationis of this type. Call signs which designate particularaircraft are examples of this type of event signal.

b. Intrinsic Event Signals. With this type ofsignal, there is an intrinsic relationship between thesignal itself and the event signalled. During a ground-controlled approach (GCA), the communication "turn toheading two one zero" bears an intrinsic relationship withwhat is to be done.

c. Occurrence Statements. Both types of discreteevent signals require that the receiver knows the natureof the system and the population of outcome. When the

receiver does not have prior information the signal must

present all such data.

2. Continuous Level sila12. This category refersto those signals which present the quantity or amount of

a variable, such as speed. One of the most important as-pects of skilled performance is the response to continuously

varying indications of quantity, position, direction, andvelocity. Pilots systematically monitor heading, altitude,airspeed, etc., and make corrections to the state of theaircraft based on such information. Although certain dis-plays may be digitized, the continuous nature of theinformation still remains.

3. Analog (Pictorial) Representation. One of the

most common modes of conveying information is to provide

11L

the receiver with a picture, model, or map of an object.This method is most useful in presenting information that

is difficult to verbalize; under such conditions, a pic-ture may indeed be worth a thousand words. Although the

usual mode is visual, other modalities (taste, sound,etc.) also permit this method of representation. The

basic limitation of this method is that it can be usedonly with concrete materials; pictorial representation ofabstract concepts (e.g., "freedom") may be difficult orimpossible.

4. Logical Relationships. This category of informa-tion concerns meaning rather than states or events. Three

types of logical relationships may be distinguished.

a. Equivalence Classes. The most obvious caseof equivalence class information is a verbal definition;e.g., "dc power in watts is equal to the product of the

current and the voltage." Statements of mathematicalidentities also represent equivalence inputs. For simplic-ity we will include here messages of nonequivalence, in-cluding inequalities such as "A is larger than B," andnegation.

b. Inclusive Relationships. A typical relation-ship expressed in symbolic logic is that of inclusion--arelationship that states that a particular object or eventis a member of some well-defined category or class. For

example, the statement that a T-37 trainer aircraft hastwo engines is an inclusion statement which places the T-37

in the class of all two-engine aircraft. Note, however,that the statement does not specify that these two conceptsare equivalent. Information of this type occurs frequently

in training situations.

c. Relationships of Natural Regularity. Here we

include statements af sequential and causal relationship,for example, information that "A precedes B" or "A causes

B." Such information may be derived from instruction orby an inductive or problem-solving process based onrepetitive signals.

12

III. HUMAN INFORMATION PROCESSING

The study of human information processing focuses onthe relationship of the individual to his informationalenvironment--to the sources of information on which hedraws, and the outputs he furnishes to that environment.Within the individual, the input, control, and output pro-cessing, as well as the characteristics of the information-storage systems are of central concern.

Categories of Information Processing

As indicated, human information processing can be sellOP

divided into three categories, input processing, centralprocessing, and output processing. Obviously, a cleardistinction between these three processing categories can-not be made since all three are involved in any information-processing sequence. Nevertheless, the broad spectrum ofinformation-processing research represents a series of at-tempts to clarify' the functioning of the system by focusingon one or another portion of the system, or some combina-tion of system elements.

Input Processing. As information comes into thesystem, it is first detected, then identified or recognized(categorized), and passed on to central processing. Themechanisms of identification and recognition involve com-ponent processes of gating, decoding, and filtering(selective processing or selective attention), comparing,and storage. Each of these processes utilizes informationalready stored, and involves various storage mechanisms,to be described later. Studies focusing primarily on inputprocessing include experiments on signal detection, reactiontime, speed and accuracy of display reading, pattern recog-nition, dichotic listening, visual and auditory short-termmemory, visual scanning, and psycholinguistics and readingstudies.

Central Processing. Despite its flavor of "mentalism,"most authorities agree that we cannot escape the notion ofa central processing mechanism, which includes the functions

of rehearsal, decision making, reasoning, inductive problemsolving and concept formation, organization of informationfor storage in memory, and imagery. The various functionsof gating, coding, chunking, filtering, comparing, andstorage also operate in central processing. In addition,it is usually presumed that the selection of output, thecentral monitoring of output (as opposed to monitoringwhich occurs through feedback), search control and alloca-tion of attention are under central processing control.Representative studies of central processing are reason-ably obvious from the list of mechanisms given above.

Output Processing. While the selection of output maybe generally considered as a central processing function,the actual planning and generation of responses is clearlyan` output function. Overt responses can be categorizedconveniently into four types: verbal, procedural, control,and search. These types of output have been describedpreviously and need not be repeated. Verbal output general-ly is studied under the rubric psycholinguistics. It isof special importance to note that many studies of inputand central processing also call for evaluation of thespeed and correctness of verbal responses. The ability tomake such responses is not at issue, however, and the re-sponses are used primarily as indices of the efficiency orspeed of other processes. Control outputs are generallystudied in tracking experiments. Perceptual-motor perform-ance in tracking is perhaps the best-studied outputphenomenon in human behavior. In the area of search be-havior, the primary effort has been in the allocation ofsearch time as a function of the informational "richness"of a visual pattern. More recently, efficiency of visualscanning has come under experimental scrutiny. Orientingresponses in the human have been studied primarily indetermining the factors governing "involuntary" allocationof attention.

Information Storage

Currently, multi-store models are most frequentlyused in research on memory. In typical studies, informa-tion retrieval is tested at various intervals following

input. Different functions are generally interpreted interms of different storage systems. The majority of con-temporary theorists rely on either a dual-storage (Broad-bent, 1958, 1971; Crowder & Morton, 1969; Waugh & Norman,1965) or triple-storage (Atkinson & Shiffrin, 1969) model.While it is not clear that more than a single memory storeexists (Melton, 1963, 1970) and while the utility of thetemporal interval of testing as an independent variablehas been questioned (Craik & Lockhard, 1972) the multi-store model provides an attractive structure for the frame-work of the present research.

Recall performance is probably often a joint functionof several memory systems (Atkinson & Shiffrin, 1968, 1971).First, is appears likely that memory can be divided verti-cally into (a) a sensory store which lasts only a shorttime and which apparently requires little, if any, speoialeffort on the part of the subject; (b) an active or short-term store which consists of the subject's active rehearsaland, thus, is dependent on the subject's strategy andmotivation; and (c) a long-term store which is like thesensory store in that it is passive, unlike the sensorystore, however, it extends for much longer durations.Second, it appears that memory can be divided horizontallyon the basis of the nature of the memory code; i.e., withwhich sensory modality it is most analogous.

Sensory Storage. Sensory storage is thought to be aninherently brief registration of information in a relative-ly unanalyzed form prior to the imposition of identitiesupon it. There is now a body of evidence obtained from avariety of experimental paradigms supporting the notion ofsensory storage in both the auditory and visual modalities.Neisser (1967) refers to these storage systems as echoicand iconic memory, respectively.

Information in sensory storage is subject to masking,erasure, and (or) distortion from additional stimulationto the same sensory modality--even if the subject is notattending to this additional stimulation (Averbach & Coriell,1961). Recent studies, however, show little interactionbetween storage at the sensory level and ongoing cognitiveprocesses (e.g., classifying irrelevant stimuli during the

15

presentation of the memory items and during the retentioninterval). This is to be expected if readout from sensorystorage were based on more or less unanalyzed features ofthe stimuli.

Short-Term Storage. Short-term storage refers to thesubjea7srehearsal processes. It is assumed that short-term storage is limited in capacity and that informationin this system is subject to rapid decay when rehearsal isprevented.

Once information is copied into the short-term storeit is no longer vulnerable to interference from unattendedexternal stimulation (Averbach & Sperling, 1961), but itis subject to interference from ongoing cognitive process-es (Posner & Roseman, 1965). This result would be expectedif it were assumed that both recycling of the memory traceand cognitive processing were different functions performedby the same system (Atkinson & Shiffrin, 1971).

Long-Term Storage. Long-term storage refers to in-formation which is stored on a relatively permanent basis.Information does not enter into long-term storage randomly,but rather it apparently is "carefully" placed in anorganized position so that when one references the parti-cular item in long-term storage, semantically related in-formation is available at the same time. The structureof long-term memory and the organization which it imposeson incoming stimulils one of the most intriguing yet leastundarstood areas of experimental psychology.

Information flow diagrams generally depict informationinitially in sensory storage, then transferring to short-term storage and finally into long-term storage. However,as many of the properties generally attributed to short-term storage; e.g., categorization and rehearsal, requireaccess to long-term storage, a more credible diagram wouldrequire two transfer systems from sensory storage: onegoing through short-term to long-term and one by-passingshort-term and entering long-term directly.

16

Executive Processor

Unfortunately, the specification of storage systemsdoes not address the question of how decisions are made withrespect to the material to be stored and that to be eliminated.Neisser (1967) suggests a central processing system whichoperates over and above the processing systems discussedabove. Some sort of "executive" is necessary to controlactivities, allocate attention, and to integrate sequences.In general the "executive" constitutes all of those cogni-tive activities which previously fell under the heading ofconsciousness, attention, and voluntary mental activity.

Capacity Limitation

The processing capacity of the human information pro-cessing system is limited in two ways. The first limita-tion is in terms of storage capacity; i.e., immediate mem-ory for a series of unrelated items is limited to thenumber 7 + 2. The source of this limitatior appears to berestricted to the short-term store. The sensory store andthe long-term store are thought to be high capacity sys-tems. The second limitation in capacity is in the ....ate atwhich information can be processed. The rate limitationis thought to reflect a limitation in the executive pro-cessor rather than the specific storage systems.

IV. IDENTIFICATION OF INFORMATION PROCESSINGCOMPONENTS OF FLYING SKILL

One of the major goals of the present project was theidentification of significant information-processing skillsin flying. To do so, a wide variety of documents were re-viewed and analyzed in an attempt to extract such data.These included Air Training Command materials currentlyused in the Air Force's undergraduate pilot training pro-gram as well as numerous research documents (Shannon, Waag,& Long, 1973; BaUm, Goebel, & Smith, 1972, Smode, Hall, &Meyer, 1966). Interviews vr-e conducted with numerousflight trainingpersonnel at Williams AFB, Arizona. At the

17

9t-:

Flying Training Division, Air Force Human Resources Labora-tory (AFSC), Williams AFB, Arizona, the investigators werebriefed and given practice in the operation of variousflight simulation devices. The 82d Flying Training Wing,also at Williams AFB, provided the principal investigatora familiarization flight in the T-37 training aircraft.

Early in the project, it became evident that theanalysis of individual piloting skills, even though phrasedin information-processing language, would be largely re-dundant with previous analyses of the same character (Kidd,1962; Fleishman, 1967). Instead of generating another listof skills or tasks which the pilot must perform, thedecision was made to produce a conceptual model of theflying environment which emphasizes the most salient pro-cesses. The resulting analysis represents hr attempt tostructure the field of information processing as it pertainsto flying.

The Informational Environment

Figure 1 presents a working model of the environmentcomposite informational system existing during a routinetraining flight. It is not intended to be comprehensive,but rather, an expository device only.

Input. The student pilot draws information from avariety of sources. Indications of aircraft state aregiven by visual cues from cockpit displays and controlpositions, and by kinesthetic cues from control pressuresand motion accelerations. Auditory cues relevant to thestate of the aircraft are also provided in the cockpit.The instructor pilot provides information directly in theform of instructional cues and demonstrations while exter-nal information is obtained from radio communications. Inaddition to information received from external inputsduring flight, the student also accesses his own storedinformation; that is, his own memory. The student's memorycontains cognitive and motor information concerning propermethods of performing a task, along with their criteria,emergency procedures, and other informational. items pre-viously acquired and stored.

18

22

AU

DIT

OR

Y

VIS

UA

L

KIN

ES

TH

ET

IC

INT

ER

NA

L F

EE

DB

AC

K L

OO

PS

0 z i

.J 4 gr i_

tu 0

2 is en

la 8 E

, itm 0

0

m ci....Z

z cc

e-0

EX

TE

RN

AL

FE

ED

BA

CK

LO

OP

S11

1,!1

1111

,

CO

NT

RO

LLE

D

UN

IT

VE

RB

AL

SE

AR

CH

PR

OC

ED

UR

AL

CO

NT

RO

L

Figure 1.

v:odel of the Environmental Composite

System During Flying Training.

Output. Each response of a student pilot can be con-sidered an output of information. In some cases such asstatements to the instructor and radio communications, theinformation is discrete and verbal. Information is alsoprovided to the aircraft through the direct manipulationof Controls within the cockpit. Such information is like-wise transmitted to the instructor who is observing suchactions and their effects on the state of the aircraft.Every control input will in some way alter the state ofthe aircraft, which in turn provides informational cuesconcerning the effect of such inputs. In this manner, afeedback loop is completed. Finally, the student engagesin active search behavior in order to secure necessary in-formation which will serve as input cues.

The physical actions of the pilot in controlling theaircraft have been dichotomized by Smode, Hall, and Meyer(1966) into procedural and control actions. Proceduralactions are relatively discrete, intermittent controlactivations such as moving switches and manipulating con-trols. A typical procedural action is that of loweringthe landing gear. Control actions are those which requirethe pilot to make continuous adjustments so as to controldirectly the flight path of the aircraft. The two typesof actions are distinguished more in terms of purpose thanin terms of which control is used. For example, when thethrottle position is changed to lower speed on landing,this is essentially a procedural action. In contrast, thethrottle can be used as a continuous control device tomaintain position relative to another aircraft in formationflying.

The Normal Landing Pattern

As an example of information processing requirements,consider the student returning to base under supervisionof an Instructor Pilot (IP). The student must perform acomplex sequence of actions in real time using informationfrom the sources described earlier. In addition, theexternal environment input must be entered into storage,compared to the student's knowledge of task requirements(which resides in long-term quasi-permanent storage), and

20

used to generate responses in real time so as to match therequired flight regime. While the actions required to landthe aircraft would take too much space to describe in de-tail here, major groups of task elements include theinitial overhead approach to the duty runway, the pitchoutand upwind turn, the downwind leg, the final turn, finalapproach, roundout, touchdown, landing roll, and taxi backto the flight line. The student must keep in mind theverbal information received by radio, such as landing run-way, wind and temperature information, and traffic ad-visories. As he approaches for landing, he must repeatedlymonitor his position and ground track, and such aircraftstate indicators as airspeed, rate of descent, altitude,and heading. Appropriate control actions must be generatedto satisfy maneuver requirements. Superimposed on thesecontinuous tasks are the intermittent procedural actions(e.g., lower landing gear, lower flaps) which must be ac-complished at specific prescribed points in the landing.In addition, the student must be alert continuously forindications of hazardous states or malfunctions of the air-craft subsystems.

Since human information processing capacity is limited,the st..dent pilot must shift his attention rapidly to securethe information he needs to execute the required proceduraland control actions. ` is usually called a continuouscontrol task may be cont...ouous with respect to the requiredoutput, but is usually discontinuous with respect to thepilot's behavior and his information inputs. Only inartificial settings, such as laboratory-type single-dimen-sional tracking, can the attention be focused continuouslyon one input channel. In the normal landing pattern thestudent must tine -share inputs since his capacity will notpermit simultaneous monitoring of several active informationsources. To do this, the student scans tht relevant in-struments and areas of the external world one at a time,checking each against his concept of the required statewhich must be achieved so that the future position, velocity,direction, and trim of the aircraft will match the maneuverrequirements.

21

unAgimmiNEM

V. RESEARCH PROGRAM

Within the context of an information processing ap-proach, a program of research was initiated for the investi-gation of components believed to be relevant to flyingskill. Since the research was basically exploratory innature, a broad spectrum of information processing taskswere investigated ranging from simple identification ofnonverbal stimuli to complex rule learning in conceptformation. The research program was designed to encompassthree major areas: (a) Serjsigarit2.milit--toisolate information processing skills which are sensitiveto individual differences and determine the extent of suchvariability; (b) Control--to explore means for controllingor accounting for such individual variability; and (c)Application - -to ascertain the degree to which these tasksare applicaoi to pilot training and to provide a meantfor their implementation. The present report deals pri-marily with experimental results concerned with the firsttwo areas. Although the results are based on data obtainedexclusively from a college student population, there is noreason to believe that the present findings do not holdfor pilot trainees since they are selected exclusively fromthis group. However, to insure generalizability, a longi-tudinal effort using pilot trainees at Flying TrainingDivision, Air Force Human Resources Laboratory does appearto be necessary, especially to investigate the third areaof concern, applicability to training.

VI. THE PROCESSING OF AUDITORY AND VISUAL SIGNALS

The pilot must continuously process incoming visualand auditory signals which emanated from the environment.As a result of high processing demands which occur undercondicions of task overload, it seemed that the investi-gation of the memory processes involved in the retentionof rapidly presented auditory and visual signals wouldprove to be of value. To study the processing of rapidlypresented information, an experimental paradigm was em-ployed; the observer is presented certain information which

22

he is required to retain for later recall. It was assumedthat the observer actively processed the information not

only during actual presentation of the stimulus, but also

during the time period subsequent to its presentation. To

determine the extent of this processing and its importance

for subsequent recall, an attempt to interfere with hisprocessing was made by presenting an additional interfering

stimulus. The interfering stimulus was presented subsequent

to the presentation of the target stimulus.

Consider the following thought experiment. Suppose aninterfering stimulus is presented long after the targetstimulus in a typical experiment in visual information pro-cessing. It is not unreasonable to assume that the inter-

fering stimulus will have a small or negligible effect onretention of the target stimulus. If, on the other hand,the interfering stimulus is presented shortly after the

target stimulus, it is reasonable to conjecture that the'

interfering stimulus will produce a decrement in perform-

ance. In the present work, the effect on memory perform-

ance of the delay between presentation of the targetstimulus and presentation of the interfering stimulus wasinvestigated in an attempt to obtain a measure of the time

course of processing. It was hypothesized that when thetarget stimulus was closely followed by an interferingstimulus; chat is, with no time between the interfer-

ing stimulus, a maximal decrement in performance would be

obtained. Furthermore, as the delay of the interferingstimulus was increased, performance should improve mono-

tonically. In this way, measures of the time required toprocess auditory and visual information could be obtained.

Measurement of Auditor] Processing: Stimulus Interference

Using this experimental paradigm, the processing ofsimple auditory information was studied (Leshowitz & Cudahy,1972; Cudahy & Leshowitz, 1973). A pure tone having acharacteristic pitch was presented, and the observer wasasked to remember the pitch of the target tone. Followingpresentation of the target tone, an interfering tone waspresented. The duration of the target tone was 10 msec;the duration of the interfering stimulus was 500 msec. One

second after the presentation of the interfering tone, athird tone, called a comparison tone, was presented. Thecomparison tone was either the same or different in pitchfrom the original target tone. The probability of thecomparison tone being different was .5. The subject wasasked to compare he pitch of the target and comparisontone and respond whether or not the comparison tone wasthe same or different.

The experimental variable of major interest was thedelay between termination of the target tone and onset ofthe interfering tone. The function relating the probabilityof a correct response and delay of the interfering toneprovided a graphic representation of the time course ofprocessing. It was observed that as the delay of theinterfering tone was increased, the probability that thesubject correctly identified the comparison tone increasedmonotonically. The finding that identification of thecomparison was nearly perfect for delays of the interfer-ence tone of about 100 msec is in accord with the simplenotion of interference with processing produced by asubsequently-presented irrelevant stimulus.

Experimental Variables Affecting Processing

Stimulus Duration. Several experimental variableswere found to affect the magnitude of the interferenceeffects produced by a retroactive masking tone. First andmost important was the duration of the target tone. Theinterfering tone produced a decrement in performance onlywhen the target tone was very brief. For target toneslonger than 20 msec, presentation of an interfering tonehad no effect on performance. These data can be understoodby assuming that a very brief tone is characterized by afew critical features. Presentation of an interfering tonedecreases subsequent recall of the target tone because thecritical features of the target tone are not processed whenan interference tone is presented. Longer duration tones,on the other hand, are characterized by a great redundancyin information. Thus, long-duration stimuli provide moreopportunity for the subject to discriminate than criticalfeatures. Since a retroactive interfering stimulus cannot

24

obliterate all of these critical features, interferenceeffects are less pronounced.

Channel. Another critical dimension in informationprocessing is the channel in which the interference is

introduced. It is assumed that auditory processing iscarried out in two independent channels, the left andright ears. To test the assumption of independence ofchannels, the interfering effects produced by a contra-lateral interference tone were investigated. This

means that if the target stimulus is presented to the left

ear, the interfering stimulus is presented to the right

ear. On the assumption that processing in the two channelsis independent, interference exerted by contralateral mask-ing tones should be minimal. This result should hold evenfor brief target tones which have been shown to be suscep-tible to interfering effects introduced in the ipsilateralchannel. In accord with this prediction, the obtainedresults showed that the interference produced by a tonecontralateral to the target was in all cases minimal and

thus suggested that the two auditory channels were inde-pendent.

Similarity. The similarity of the interfering andtarget stimuli is thought to be an important variable in

determining the extent to which normal information process-ing is altered. In the auditory processing task describedabove, similarity was investigated by varying the spectral

(frequency) content of the interfering stimulus. A simplenotion of interference would predict that the greater the

similarity of target and interfering tone, the greater the

interfering effects. In the present experiment, however,small effects of similarity were found. That is, as thefrequency separation between target and interfering stimuliincreased, little or no change in performance was observed.Apparently, common elements shared between target andinterfering stimuli play a very small role in interferencewith auditory information processing. The present data canbe interpreted as indicating that the role of an inter-fering stimulus is to produce a general level of interfer-ence.

The failure to find similarity effects is paradoxicalin light of the observation that not all extraneous stimuli

25

produce interference effects. Recall that a contralateralinterference tone had little or no effect on performance.Moreover, similar experiments in the literature demonstratethat interfering stimuli entering through the visual mo-dality have little effect on recognition of auditorystimuli. Apparently, interfering stimuli must be present-ed in the same auditory channel as the target tone if theyare to be effective (Leshowitz, Zurek, & Robbins, 1974).

Measurement of Auditory Processing: Response-InducedInterference

It should not be assumed that interference with in-formation processing can be produced only by stimulusevents. The subject's own responses can serve as animportant source of interference. Several investigations,both theoretical and empirical, have been devoted to astudy of interference produced by the subject's own re-sponses. In the first experiment, the effects of informa-tion retrieval on recognition memory for pitch were in-vestigated (Leshowitz & Hanzi, 1973; Leshowitz& Green,1973). The observer was presented a list of tones whichwere to be retained. The list of tones were chosen in thefrequency region between 250 and 1000 Hz. After presenta-tion of the initial stimulus list consisting of either 6or 8 tones, the subject was presented 4 or 5 test tones.The task of the subject was to determine whether a giventest tone had beeri included among those presented initially.One-half of the tones in the test set were included in theoriginal set. The probability of the subject correctlylabeling a tone as "old" is called the "hit" rate. Theprobability that the subject falsely labeled a new tone asold is the "false alarm" rate. The relative proportion ofhits and false alarms is used to compute memory strength(d') and gives a measure of recognition memory.

The purpose of using tcnes as the test material asopposed to simple vocabulary words was to control forlinguistic associations normally part of the subject'slanguage repertoire. In other words, we attempted toascertain the strength of stimulus and response - inducedinterference in a memory task in which prior experiencedoes not play a role. The attempt was to obtain a pure

26

measure of interference.

The results of this experiment were surprisinglysimilar to those presented in the vernal learning litera-ture. Specifically, both stimulus-induced and response-induced interference effects were obtained. The dataanalysis permitted isolation of precisely the amount ofinterference attributable to stimulus events from ..hatportion attributable to response interference. Probabilityof correct recognition as a function of position in theoriginal stimulus list provided a meausre of stimulusinterference. Performance as a function of position inthe test list provided a.measure of response interference.

In verbal learning, it is common to find that itemspresented at the beginning of the list and items presentedat the end of the list are rememLered better than items inthe middle of the list. Such results are the classicalserial positions effects of "primacy" and "recency,"respectively. These effects were again demonstrated forthe memory of pitch. Moreover, the magnitude of theseserial position effects was related to the amount of timeallowed the subject to rehearse the items in the stimulusset. When items were presented in rapid succession withlittle or no time allowed for rehearsal, the serial posi-tion effects differed markedly from conditions in whichthe set was presented at a slower rate. What seems to beof prime importance is the opportunity to rehearse theitems in the set.

Analysis of recognition as a function-set positionshowed that internal sources of interference may be evengreater than external sources of interference. The amountof interference produced by the subject's own responsescan be measured by comparing memory performance for itemspresented at various test-set positions. Suppose theinitial test item is recognized at near-perfect levels,independent of the item's position in the original stim-ulus set, and that items further down the test set arerecognized less well. The decrease in performance as afunction of position in the test set provides a directmeasure of response-induced interference. The present re-sults shoed a monotonically decreasing function ofrecognition performance as a function of test-set position.

These results indicate that in testing, the sheer act ofcomparing the test item and a replica of this item inmemory has a profound effect on memory for other items.In other words, the comparison process itself produces amarked decrease in recognition performance for other items.It appears that response-induced interferencc is at leastas great as stimulus-produced interference.

Interference produced by the retrieval of informationwas especially marked when little or no time was allowedfor rehearsal of the original material. For items whererehearsal was impeded by presenting items at a rapid rate,subjects had little or no difficulty in recognizing theinitial test-list items. However, recognition of allsucceeding items was drastically curtailed. It appearsthat without sufficient time for rehearsal, items in mem-ory are particularly susceptible to response-induced inter-ference. If the task does not permit sufficient processing,this information can be easily wiped out by the observer'sown responses. This finding has some important practicalimplications to which we shall return in a later section.

Measurement of Visual Processing: A Partial ReportRecognition Paradigm

In the next series of studies (Leshowitz, Hanzi, &Zurek, 1973), several experiments on visual informationprocessing were designed using general principles derivedfrom the previous studies on simple auditory informationprocessing.

The purpose of the experiment was to determine hawmuch of the array the subject could process. A secondinterest was to determine whether a visual interferencestimulus could produce decremental effects. Also, an at-tempt was made to contrast stimulus-induced and response-induced interference in processing complex visuall arrays,and to determine strategies used by observers in processingvisual arrays, specifically, do observers scan from left toright?

An objective recognition memory paradigm was employedin assessing the human information processing capacities

28

governing visual memory. The subject was presented thevisual array for a brief period. Since it has been shownthat observers have great difficulties in recalling largeamounts of visual information, a partial-report paradigmwas employed. Rather than asking subjects to report alltwelve letters of the array, they were asked to report ononly a partial sample of these letters. Subsequent tothe array, an indicator tone was presented which signaledreport of either the upper, middle, or lower row of thearray. In order to measure memory decay of the originalstimulus array, an experimental variable of interest wasthe delay between termination of the array and onset ofthe indicator tone. Following presentation of the indica-tor tone, the subject was presented a test list of letters.The test list was located to the right of the visual fieldthat contained the original stimulus array. The task ofthe observer was to indicate whether or not the test letterspresented in a given raw-column had been presented in theoriginal list. If the letter had been presented in theindicated position, the subject was instructed to vote"old." If this letter had not been presentedin the in-dicated pattern, the subject was asked to vote "new." Asbefore, hit and false alarms rates were computed as well asmemory strength scores. Recall ;.hat memory strength pro-vides a direct measure of the degree to which subjects re-tain information.

Experimental Variables Affecting Visual Processing

Scanning Strategy. A number of significant main ef-fects were obtained. First, subjects have a pronouncedproclivity to scan the visual array from left to right.In addition, subjects preferred the middle row in contrastto the upper and lower rows. As a result of these biasesfor certain positions, recognition memory for these pre-ferred stimulus positions was markedly superior to recog-nition of letters in the other locations of the stimulusarray. Of special note is the observation that thesebiases manifested themselves across all three observers inthe experiment.

Delay of Partial-Report Cue. In agreement with theclassical data on memory decay, recognition performance was

29

:43

a decreasing function of indicator tone delay. For zerodelay of indicator tone, performance was nearly perfect.When the delay was 2 seconds, performance approached nearchance levels. Thus, it appears that visual informationcontained in complex visual arrays decays in about 2seconds.

Interference Stimulus. In an attempt to interferewith normal processing of visual information, a paradigmnearly identical to the one described above in the auditoryinformation processing experiments was used. Followingpresentation of the stimulus array, a masking stimulus con-sisting of a homogeneous array of dots was superimposed onthe visual array. This retroactive masking stimul"s alwaysfollowed presentation of the visual stimulus array. Inagreement with our notions developed for interference withauditory processing, undc!r all conditions, presentation ofthe masker produced a significant decrease in subsequentmemory performance. That is, recognition memory for lettersdecreased upon presentation of the masking stimulus. Al-though increasing the duration of the original visual arrayfrom 50 to 500 cosec, and thereby permitting greater timefor processing of the original array, increased performance,the addition of a masking stimulus had a significantdebilitating effect on memory.

Response-Induced interference. The final analysis wasdevoted to an analysis of the interfering effLcts producedby the subject's awn responses. The function relatingperformance as a function of position in the test set pro-vides a direct measure of interference produced by thesubject's own responses. In marked contrast to the pre-vious set of data on recognition memory for pitch, memoryfor items at the end of the list was no less than therecognition memory for the test set items at the beginningof the set. It appears that retrieval interference is ofnegligible importance when the subject is required to pro-cess a visual array. While an immediate explanation ofthe difference between retrieval interference for visualand auditory information processing is not apparent, itdoes appear that the highly codable visual array consist-ing of English letters is less susceptible to retrievalinterference. The unidimensional, noncodable auditory

304 f.

pitch, on the other hand, is clearly susceptible to this,retrieval interference. On a codability dimension, inter-ference associated with the retrieval process becomes aless important factor in recognition memory as the informa-tion becomes more easily coded. For relatively noncodablematerial, sufficient time for rehearsal of the originalstimulus list also mitigates the interfering effects in-herent in the retrieval process. In summary, opportunityto encode a to-be-retained item is of great importance.This finding has important implications for flying training.

VII. SHORT-TERM RETENTION OF AUDITORY INFORMATION

A pilot is constantly required to process complex ar-rays of information which, in many cases, are presentedover several communication "channels" simultaneously. Thechannels might be within the same sensory modality (e.g.,two different messages, k e to the left ear and one to theright ear) or they might be between two sensory modalities(e.g., one message spoken and one message pictorially dis-played). For a more concrete example, while listening forlanding instrlctions from a control tower (auditory channel),a pilot must also maintain visual contact with his instru-ment panel (visual channel), and in addition, commence arather complicated series of movements in preparation forlanding (tactile and kinaesthetic channels).

Pilots, then, are apparently able to do several thingsat once. This leads us to some questions: (a) How arethey able to do it; i.e., what processing strategies arethey employing? (b) Are the strategies trainable; i.e.,if we find a trainee deficient in employing optimalstrategies, can we provide instructions to remedy the dif-ficulty? (c) How can we isolate the processes used so thatwe might construct devices; i.e., simulators, the operationof which would permit more efficient and inexpensive train-ing? The answers to these questions necessitate researchwith tasks involving multi-channel input.

The Dichotic Paradigm

Dichotic memory refers to a set of experimental opera-tions in which subjects receive two different lists ofitems simultaneously, one to each ear, with instructionsto reproduce as many of the items as possible followinglist presentation. In studies using these operations bothlists consist of digits and presentation rate typically is2 pairs/sec. Subjects tend to report the digits from oneear and then the other rather than alternating between earsand reporting items according to the order of their present&tion. The ear-by-ear report pattern suggests that subjectsmonitor (attend to) only one message and report it firstduring recall. They are still able, however, to reportitems from the other ear (although not as accurately), soboth digit lists (right and left ear) must have beenstored initially.

The dichotic paradigm would further appear to offera fruitful area of investigation as a result of previouswork by Gopher and Kahneman (1971) in which Israeli AirForce cadets were tested using a dichotic listening pro-cedure. The results indicatei that measured performanceon this task was significantly related to two flightcriteria: (1) success--that is, whether a cadet was're-jected from flight training and, if rejected, at whatstage; and (2) flight assignment--that is, cadets scoringhigher were assigned to high performance aircraft. Suchfindings suggest that the dichotic paradigm might providea useful laboratory analogue for studying the informationprocessing requirements of the flying task.

Attended and Unattended Messages

In the first study using the dichotic paradigm, subjectswere given a sequence of four pairs of digits on each offorty-two trials. One member of each pair was presented tothe right ear and one was presented to the left ear. Sub-jects were instructed to attend to only one ear duringpresentation and to either: (a) recall first the digitsreceived by the attended ear followed by those from theunattended ear (Attended-Unattended or A-U report), or (b)

32

recall first the unattended digits and then the attendeddigits (Unattended-Attended or U-A report).

Several interesting findings emerged from this experi-ment. The items on the attended ear were more accuratelyrecalled than those on the unattended ear regardless of theorder of report (i.e., A-U or U-A). The order of reportrequiring the least switching of attention, Attended-Un-attended, produced the most accurate retention. Whensubjects were required to attend to one ear nnd to reportthe unattended ear first, the attended message was signi-ficantly impaired.

Verbal and Written Monitors

The second experiment in this series was a replicationof the first with the exception that subjects were requiredto monitor the attended message. Two types of monitorswere employed, written and verbal. In the written condition,subjects were required to write the attended message duringpresentation, while in the verbal condition subjects wererequired to shadow (report aloud) the attended message dur-ing presentation.

While the attended message was recalled with equalaccuracy by both groups, the unattended message was moreaccurately recalled by the written monitor group. Apparent-ly, when one must encode (store) and decode (output) withthe same system (auditory-verbal), greater interferenceresults than when different systems (auditory and manual)are employed.

Digit Span

There occurred considerable between-subjects variabilityin the first experiments. The primary source of variabilitywas in errors of omissions; i.e., some subjects recalledmore items than others. Some subjects were overloaded withmessages containing 4 digits per ear while others managedto store and output the information with high accuracy.Some subjects, therefore, appeared to have a greater storage

capacity than others. In an attempt to control for storagecapacity, individual digit span tests were administered inthe next experiment. Each subject received five digitspan assessments, and groups with digit spans of 5, 6, 7,8, 9, and 10 were formed.

In this study, the correlation between digit span andoverall dichotic performance (i.e., collapsed over attendedand unattended ears) was +.891. Digit span, therefore,was identified as a potent source of variability in dichoticresearch. Further correlations were run between digit spanand attended ear and unattended ear in both A-U and U-Areports. The highest correlations were obtained on theunattended ear; i.e., where switches of attention occurredbetween input and output.

Interference Effects in Dichotic Memory

In the first three experiments, different serial posi-tion functions were obtained for attended and unattendedears. Serial position functions for the attended ear wererelatively flat (all positions were recalled with approxi-mately equal accuracy), while unattended ear serial positionfunctions showed a marked recency effect (items in theterminal positions were recalled with greater accuracy thanitems occurring in the primary serial positions).

In the next experiment, we investigated the effect ofa redundant auditory element (stimulus suffix) at the endof each dichotic list. Previous work has indicated thatsuch a manipulation interferes with information which hasnot been fully processed. In this experiment, unattendedear recall accuracy was found to be significantly impairedwhen the locution "uh" was presented binaurally at the endof the list. Furthermore, the magnitude of the effect ap-peared to be a function of digit span.

Although a relationship between stimulus suffix effectmagnitude and digit span was not expected, digit spans wereassessed and the two subjects with a low digit span werefound to have experienced more interference than the twosubjects with high digit span.

34

Digit Span and Interference

Two groups of subjects were run in the next experiment.One group consisted of subjects with high digit spans (X =9.75) while the other contained only subjects with lowdigit spans (X = 6.25). The subjects participated in twotest sessions. In the first session, subjects receiveddichotic lists consisting of five pairs of digits. In thesecond session, a stimulus suffix was presented binaurallyat the end of each dichotic list. Subjects with high digitspans showed reliably less interference than did those withlow digit spans.

In yet another condition, the high span subjects weregiven dichotic lists consisting of six pairs of digits withand without the addition of a stimulus suffix. The inter-ference produced in that experiment (6 pairs) was comparablein magnitude to that of the low span subjects in the lastexperiment (5 pairs). This relates back to the problem oftask overloading mentioned earlier. Apparently, the effectof a stimulus suffix is to add more information to the task.For those subjects whose capacity was already overloadedwithout the suffix, the effect of the suffix was to radicallydegrade recall accuracy. For subjects whose capacity matchedthe condition without the suffix, the effect of the suffixwas minimal. The implications of these findings in theimprovement of Air Force flying training will be consideredfurther in a later section of this report.

VIII. AUDIOVISUAL CONCEPT FORMATION

In ordinary language, the word concept has many mean-ings, referring among other things to inner images or ideas,to abstract relationships among environmental properties orevents, to operations whI-h an organism performs on stimu-lus inputs, and so forth. a.:nng experimental psychologists,however, the term has come to have a more precise, butlimited, meaning: a principle by which all objects orevents can be classified into two categories, examples ofa concept (positive instances) and nonexamples (negativeinstances). Concept formation concerns the learning or

35

utilization of such classification principles. Thus,studies of concept formation are usually conducted withinthe framework of a task in which the subject must sort orclassify a set of stimulus patterns into categories. The"problem;" as it is usually called, is for the subject tolearn or discover the correct principle for sorting, usinginformation provided by the stimulub patterns and by cor-rective fee-aback signals given after rAch sorting response.

Experimental procedures used in the study of conceptformation almost universally require the subject to dis-cover or infer the correct concept from a series of positive'or negative instances. While there are many variations,moseprocedures fall into one of two classes, depending onwhether the experimenter chooses the sequence of stimulusinformation (reception paradigm) or the subject has anactive role in selecting each stimulus to be examined(Lalection paradigm).

All of the studies to be re?orted here have used thereception paradigm. The distinguishing aspect of receptionii: ping is that the subject has no control over theselection and ordering cf stimuli. On each of a seriesof trials, the subject is presented with a stimulus pattern,is required to classify it (as a positive or negative in-stance), and is told 'whether his response was correct ornot (informative.feedback). Trials proceed in this fashionuntil the subject can show his knowledge of the correctclassification principle, either by verbal statement or bymeeting an arbitrary performance criterion such as 15 con-secutive correct responses. Conventional performancemeasures are the number of errors made, the number of trialsto last error, or the amount of the time taken to reach thecriterion of problem solution.

Two structural features of concepts can be identified.The first is a set of one or more specific, perceptiblestimulus properties which constitute the defining propertiesof the concept. Conventionally, these are called therelevant attributes, in contrast to other characteristicswhich vary from pattern to pattern, but have nothing to dowith the concr-t and are therefore called irrelevant at-tributes. The second feature is a conceptual rule whichspecifies how the attributes are combined or otherwise

elaborated to form the concept. For ._sample, in the con-cept "red triangle," the class of all things which are bothred and triangular, the relevant attributes are redness andtriangularity. But the concept also includes a particularrelationship between these two attributes, that of conjunc-tion or joint presence. Only those patterns which are bothred and triangular are examples of the concept. The samepair of attributes may be related by a variety of otherrules, such as inclusive disjunction (either red or squareor both).

Selection of two attributes as relevant in a concept-learning problem generates a contingency table consistingof the four categories of stimulus patterns defined by thepresence or absence of the two attributes. For example,given red and square as attributes, the four contingenciesare red square, red not-square, not-red square, and not-red not-square. This contingency table is equivalent toa truth table in symbolic logic, in which the four contin-gencies are designated TT, TF, FT, and FF (T standing fortrue and F for false). The specification of a logical.rulefurther maps these four contingencies into a two-responsesystem of examples and nonexamples of the concept. Forexample, in a conjunctive concept, the first contingency(TT, or red square) is placed in the positive categorywhile the other three are negative. The four basic two-dimensional rules are shown in Table 1; the plus and minussigns indicate that the instance is positive or negative,respectively.

Haygood and Bourne (1965) pointed out that essentiallyall studies prior to that time had used instructions whichmay be characterized as attribute identification. In at-tribute identification, the conceptual rule is carefullyexplained to the subject, and his task is to discover therelevant attributes. Haygood and Bourne developed analternative procedure, called rule learning, in which thesubject is told the relevant attributes and must discoverthe correct rule of combination. With the exception ofthe audiovisual redundancy experiment, all of the experi-ments to be reported here used attribute identificationinstructions. The rule-learning procedure will be describedlater.

The study of auditory concept formation is a relatively

37

41.

TABLE 1

DISTRIBUTION OF INSTANCES INTO POSITIVE AND NEGATIVE

CATEGORIES FOR FOUR BASIC CONCEPTUAL RULES

Stimulus Patterns

Logical Truth

Table

Conceptual Rules

Inclusive

Color

Form

Red Square

Conjunction Disjunction

Conditional

Biconditional

Red

Square

tt

Red

Not Square

TF

Not Red

Square

FT

Not Red

Not Square

FF

m

recent enterprise, beginning only about 15 years ago. Thus,little is known about the process except for a few basicstudies of abstract auditory concepts. The purpose of theseries of experiments to be reported here is to extend thestudy of audiovisual concept learning to meaningful com-parisons of learing from auditory and visual materials.

Concept Identification in Three Modalities

In the training situation, the trainer has the optionof presenting his materials in any one of three modalities.These modalities are pictorial, in which pictures of theconcepts or relationships to be learned are shown, visualverbal, in which verbal descriptions of the pictorialmaterials are shown, and auditory verbal, in which theverbal materials are presented orally or by recording.The same material can also be presented in two or allthree modalities, creating what is called audiovisualredundancy; redundancy is covered in later experiments.

Despite existing folklore to the effect that "onepicture is worth a thousand words," there is little in theway of scientific data to support such a contention, andsome recent theories of information processing suggest thatdiscrete pictorial information is recoded to verbal formfor storage and utilization in human information processing.

In any training program, the goal is to teach thetrainee concepts by having him read printed verbal materi-al..., telling him about concepts, or showing him examplesand nonexamples of concepts. The purpose of the firstexperiment was to prol.fide a direct comparison of conceptlearning in these three modalities.

The standait concept-formation procedure was used, inwhich the subject's task is to discover the correct conceptby being shown a series of positive and negative instances,guessing their clx;sifications, and receiving correctivefeedback. Each subject learned under one of the threemodalities described above (pictorial, visual verbal, andauditory verbal), a concept constructed with one of two

39

.43

rules (conjunction or inclusive disjunction) and one oftwo pairs of relevant attributes (blue-triangle or two-squares). In the visual conditions, the subject was showna card containing the stimulus pattern. After his response,feedback was given by telling the subject if his responsewas correct or not, and by placing the card face up in thecorrect one of two sorting boxes (positive and negativeinstances) placed on the table directly in front of thesubject. In the auditory condition, the subject was toldwhat the pattern was; no cards or sorting boxes were used,and the subject's feedback consisted only of being told thecorrectness of his response. Subjects worked at their ownspeed, and instructions stressed accuracy rather than speed.

The results showed clearly that learning is faster forconjunct:Ne concepts, and for problems using blue and tri-angle as relevant attributes. However, in contrast to thestriking results for these variables, modality of presenta-tion had little effect, and the effect was not statisticallysignificant overall. Post hoc analysis showed a slight, butsignificant superiority for visual verbal over auditoryverbal conditions. It was decided to suspend judgment con-cerning modality differences until a second experimentcould be performed which would eliminate the possible roleof the sorting boxes in the superiority of the visual con-ditions.

Comparison of Visual and Auditory Verbal PreseLtation

This experiment was designed to eliminate the confound-ing caused by the fact that subjects in the visual verbalconditions could see the last positive and the last negativeinstance (in the sorting boxes), while subjects in theauditory conditions did not have such a memory aid. Thistype of memory aid has been shown to increase efficiencyof learning in previous research.

The purpose of this experiment was to compare threetypes of verbal presentation. The first is the visualverbal condition of Experiment I, with the sorting boxeseliminated. The second was the auditory verbal condition

40

of Experiment 1. The third was an augmented auditoryverbal condition, in which the information gained on eachtrial was repeated at the end of each trial (e.g., "thelast pattern, one large red square, was a positive in-stance). The task and procedure were essentially the sameas those of Experiment 1, with the changes noted above.In addition, the disjunctive concepts were eliminated, andall concepts were conjunctive.

The results showed no significant differences amongthe three modality conditions, indicating that the superiorperformance in visual conditions of Experiment 1 resultedfrom the memory aid provided by the previous instancesshowing in the sorting boxes. Taken together, the two ex-periments provide a clear picture. For simple concepts,presented in a relatively uncluttered context (few irrele-vant stimulus dimensions and discrete and easily identifi-able levels of the relevant dimensions), there is essentiallyno difference between seeing the stimulus patterns, seeinga verbal description of the patterns, and being told theverbal description. This finding provides a baseline fromwhich to examine more complex visual concepts, such as theappearance of the runways when the aircraft reaches theproper point for turn to final approach, concepts for whichverbal presentation may not be as informative as visualpresentation.

The Effects of Limiting Inspection and Response Time

The first two experiments indicated that modality ofpresentation has little or no effect in this type of con-cept-formation task when unlimited time is available toexamine stimulus patterns and formulate responses. It iswell known that limiting inspection-response time has adegrading effect on performance in visual problems. Anydifferences in the rates at which pictorial..and verbal in-formation can be processed should show up in this type ofexperiment if the time allowed to examine stimuli and re-spond is limited. The purpose of this experiment was tocompare performance in visual verbal and pictorial conditionswhen time constraints are imposed. Unfortunately, a suitablemethod for controlling "inspection time" for oral presenta-

.tion has not as yet been devised, so this condition couldnot be tested.

The method was essentially the same as that used inthe first experiment, with the exception that automaticequipment with slide projection was used in place of manualpresentation of cards. This change was made to providecontrol of timing. The subjects' responses were made b7pressing pushbuttons, and feedback was given by turning ona lamp directly over the correct pushbutton. Each subjectlearned either from pictorial stimuli or printed (and pro-jected) verbal stimuli under one of three timing conditions.The three timing conditions were 1.0 sec, 3.0 sec, or un-limited time to inspect the stimulus and respond. Manysubjects were unable to respond during the 1.0 and 3.0 secintervals until the concept had been learned; however, allbut seven subjects were able to complete the task satis-factorily.

Limiting the presentation-response time severely de-graded performance, and the degradation was far worse forverbal conditions than for pictorial conditions. For thepictorial conditions, performance at 3.0 sec was actuallyslightly, but not significantly, better than performancewith unlimited time. The simplest explanation of theseresults would be that it takes longer to read an extractthe information from the verbal slides than from thepictorial slides. A follow-on pilot study supported thisinterpretation by showing that it takes subjects 0.2 seclonger to make a simple identification response (e.g., "isthis next pattern blue triangle") when the pattern is verbal,compared to pictorial. Further research with a variety ofdifferent materials is needed to substantiate this pilotfinding.

These results suggest that when discrete patterns mustbe presented quickly, as would be the case when largeamounts of discrete stimulus information must be presentedin a short period of time (as in a briefing), pictorialpresentation should be used Li preference to visual verbalpresentation. Obviously, further research is needed toestablish whether these baseline findings can be generalizedto the more complex conceptual materials used in flyingtraining.

42

-46

Audiovisual Redundimy

Redundancy exists when the same information is givenin two or more different ways. In concept-formationstudies using pictorial stimulus materials, stimulus re-dundancy has been shown to be beneficial under a widevariety of task conditions. A:though it is widely believedthat providing auditory (verbal) information which is re-dundant with pictorial information is highly beneficial inlearning, there is very little evidence to support thisbelief. The purpose of this experiment was to determineif presenting redundant auditory information would improveperformance over that found with pictorial presentationalone.

Two methods of introducing auditory redundancy wereused. The first was to describe to the subject the exactpattern he was seeing (e.g., "the pattern on this card isone large red squake"). This added redundancy informationwas not expected to aid performance, since the stimuluscharacteristics showing on the card were obvious to allsubjects. The second method of adding redundancy was todescribe to the subject the "truth table" representationof the stimulus pattern, as shown in the first two columnsof Table 2 (e.g., "this pattern is not-blue, not-triangle").Two rules were used, inclusive disjunction and biconditional.It has been proposed that much of the extreme difficulty ofthe biconditional rule results from the inability of theunskilled subject to recode the stimulus pattern to itstruth-table representation. If this is the case, thenteaching the subject its representation should help perform-ance.

Each subject learned either a disjunctive or bicondi-tional concept under one of three redundancy conditions(no redundancy, complete pattern descrption, or truth-tabledescription). The results showed no significant differencefor the three redundancy conditions. If anything, perform-ance without the ve .bal cue was slightly superior, suggest-ing the possibility that the verbal cues may have distractedsubjects from thetask. The results make it clear thatdescrJbing to the subject what he is seeing is of no helpwhen the stimulus materials are simple and schematic as they

43

are in these experiments. Even with the much more difficultbiconditional rule, there is no benefit, a finding whichtends to contradict previous interpretations of bicondition-al difficulty as stemming from inability to reduce stimuluspatterns to truth-table categories.

Redundancy with Nonconjunctive Concepts

Previous studies of stimulus redundancy in multidi-mensional concept formation had always used conjunctiveconcepts in which every combination of levels of therelevant dimensions was represented by a separate responsecategory. For example, with color and shape relevant,using two-level dimensions, the four response categoriesmight be red squares, red triangles, green squares, andgreen triangles. Three-level dimensions had never beenstudied, nor had the more common two-category responsescheme used in these experiments, in which the positivecategory is defined by one attribute combination (e.g.,red squares) and all other combinations form the negativecategory. Finally, redundancy in disjunctive concepts hadnot been studied. The purpose of this experiment was toextend previous findings to the two-choice response scheme,to disjunctive concepts, and to stimulus dimensions withmore than two levels.

The task and procedure were essentially the same usedin the first experiment. Each subject learned one of threeclassification schemes, two-choice conjunction, four-choiceconjunction, and two-choice inclusive disjunction. Redun-dancy was introduced by correlating the levels of thestimulus dimensions; for example, if size and number wereredundant, small patterns were always single, medium pat-terns double, and large patterns triple. Two levels ofredundancy were used, zero redundancy and two redundantrelevant dimensions (e.g., size redundant with number andcolor redundant with shape). The intermediate conditionof one redundant dimension was not used.

The results confirm and extend previous findings inthe study of redundancl-. There is a redundancy gain underall conditions, and thi gain is greater for the more dif-

44

As

ficult problems. These results thus provide no surprises,and it appears that previous findings can be generalizedto nonconjunctive rules and to different kinds of responseschemes.

IX. IMPLICATIONS OF INFORMATION PROCESSINGFOR FLYING TRAINING

Visual and Auditory Scannin: and Interference

The present work clearly demonstrates that eventsemanating from both the external environment and thesubject's own processing of information produce delilitat-ing effects on memory. We have shown that stimulus events,extraneous to the task, when presented in close temporalproximity to the actual information processing, cause thisprocessing to be ineffectual. Similarly, the sheer act ofevaluation and comparison of stimuli interferes with recallof previously processed material. In other words, inter-ference with processing is both a self-induced and exter-nally-induced phenomenon.

The empirical findings suggest several guidelines forpreventing interference with information processing. Forexample, if the interfering event is delayed with respectto the initial processing, the extraneous stimulus winhave little effect on performance. Similarly introductionof a potentially interfering event to a processing channelindependent of the channel involved in active processingwill hardly affect performance. In addition, it has beenshown that appreciable rehearsal of the to-be-learnedmaterial can attenuate interference with processing. Thesesummary statements appear to govern information processingof both simple and complex material presented to auditoryand visual modalities.

The obtained data have direct implications for flyingtraining. For example, in flying an aircraft, decisionsare based on information obtained from scanning a controlpanel. Can we instruct the pilot as to the optimal strategyfor processing arrays of visual information? The answer

45

appears to be affirmative. Recall that the results of thevisual information-processing experiment described aboveindicated that observers tended to have consistent stra-tegies for visual scanning. In the auditory experimentit was observed that when rehearsal of items in memory ispermitted, memory for this information was facilitated.And finally, a comparison of the auditory and visual pro-cessing experiments strongly suggested that when informationis amenable to verbal coding, as was the case for memoryfor English letters, interference emanating from retrievalprocesses is minimized.

Let us now apply the empirical, results in developingan optimal scanning procedure. Scanning should proceed ata slow but steady rate, allowing processing of such suc-cessive items of information on the control'panel. Theattempt here should be to allot enough time, say a fractionof a second, for processing and rehearsing of each readingon the control panel. Each instrument reading should befully coded before additional information is scanned. Averbal code, rather than a pictorial or figural representa-tion in visual memory, is required if subsequent recall isto be maximized. Observe that the process should be serial,with information being processed item by item. Obviously,a great deal of additional research needs to be devoted todeveloping actual teaching methods for imparting thisstrategy to student pilots. The important issue here isthat such an approach lies within the bounds of educationaltechnology.

It is further speculated that nonoptimal scanningstrategies are employed by student pilots in scanning andprocessing control panels. It is proposed that follow-upresearch be devoted to examining the actual visual scansof the eyeball when processing information in the cockpit.This can easily be carried out with existing physiologicalrecording instruments. Physiological scanning data couldbe correlated with actual flying performance measuredeither in the simulation laboratory or in actual flight.The effort would focus on defining the relationship betweenobjective measures of scanning with actual informationprocessing.

46

Auditory-Verbal Information Processing

Research by Gopher and Kahneman (1971) indicatedpotential utility of tasks employing selective attentionand short-term memory for predicting an individual's chanceof success in pilot training. Several advances were madein this area in the present research employing dichoticlistening. First, approximately 80% of the variance indichotic performance was accounted for by digit span. This

result suggests that digit spans should be assessed forpilot trainees in a longitudinal screening study. Thiscan be accomplished with minimal demand of trainee time.Twenty or thirty members of a class could be tested at atime and only 13-20 minutes would be required for test ad-ministration. The predictive value of digit span could beassessed in several classes. If it were found that traineeswith lower digit spans had a higher probability of beingdismissed from the program two approaches could be taken.First, digit span could be used as a screening device forthe program. In this approach, a minimum digit span (e.g.,7 digits) would be set up as a criterion for admissioninto the vogram. Those falling below that digit spanwould not be selected. Second, an information processingtraining program could be initiated for individuals witha low probability of completing pilot training. These in-dividuals could be given training in tasks involvingselective attention and short-term memory with the aimthat some of these abilities can be improved with practice.In this latter case, the digit span data would be used asa screening device for additional training and not fordismissal.

Another major advance in the present research was thefinding that a redundant acoustic locution at the end of astimulus list (stimulus suffix) interfered with the reten-tion of items thought to be relatively unanalyzed followingstimulus presentation. Further, the magnitude of the im-pairment was in direct proportion to the disparities betweenan individual's digit span and the number of digits in theseries presented for auditory processing.

Previous research with stimulus suffix effects hasindicated that the greatest impairment of retention occurs

'51

when the suffix is physically similar (same pitch, inten-sity, and tempo) to the stimulus list. Consistent with thepresent research these data point up some implications fortransmitting auditory-verbal information. The capacity forreceiving information is sharply limited. To be effectiveit is suggested that messages be kept short. The suffixexperiments suggest that if two messages are to be receivedin rapid succession, it would be more efficient if thesecond message were spoken in a different voice or at adifferent intensity than the first message. This enablesthe pilot to rapidly shift attention to the second messageand precludes loss of some first message information dueto second message presentation.

Audloviusal Concept Formation

The concept-formation studies conducted as part ofthis research program employed problem-solving methodscomparable to those used in the popular "learning-by-discovery" techniques. They are directly applicable toinstructional situations in which the subject is attempt-ing to learn, by trial-and-error or discovery, simple con-cepts representing compounds of discrete stimulus elements.Two major findings are evident from the present research.

The first finding stems from the experiments comparingvisual and auditory presentation. The results show clearlythat, with simple problems using discrete stimulus charac-teristicE, it makes no difference whether the subjectreceives the information through pictorial, printed verbal,or auditory-verbal methods--provided there is unlimitedtime to inspect the stimulus pattern and generate a response.When inspection/response time is severely limited, pictorialpresentation is superior to printed verbal presentation,apparently because it takes longer to read and understandthe words than to abstract the same_material from apictorial stimulus.

The second major finding concerns audiovisual redun-dancy. Describing simple visual patterns verbally to thesubject is of nJ benefit and may, in fact, distract thesubject so as to interfere with performance. Considered

4852

in the light of previous research, this finding suggeststhat "show-and-tell" is of benefit primarily with complexmaterials, or materials in which the relevant visualcharacteristics are either obscure or change rapidly overtime.

Conclusion

The present work has developed several principlesunderlying information processing of visual and auditorystimuli. We refer to the principles of channel capacity,rehearsal, coding, stimulus and response-induced interfereence. As a direct result of this work, tools have beenmade available with which to probe nonoptimal processingstrategies of student pilots and to teach more effectivetechniques for flying. The follow-up research to thepresent program will include a greater emphasis on informa-tion overload and continuity in the task.

49

53

REFERENCES

ATC Manual 51-4, rrtmary flying, jet. HQ Air TrainingCommand, Randolph Air Force Base, 1971.

ATC Syllabus P-V- 4A-A. Syllabus of instruction for under-graduate pilot training. HQ Air Training Command,Randolph Air Force Base, 1971.

Archer, E. J. Concept identification as a function ofobviousness of relevant and irrelevant information.Journal of Experimental Psychology, 1962, 63, 616-620.

Atkinson, R. C. & Shiffrin, R. M. Human memory: Aproposed system and its control processes. In K. W.Spence and J. Spence (Eds.), Advances in the psychologyof learning and motivation research and theory. Vol.

2. New York: Academic Press, 1968.

Atkinson, R. C., & Shiffrin, R. M. The control processesof short-term memory. Institute for MathematicalStudies in the Social Sciences, 1971, TechnicalReport No. 173.

Atkinson, R. C., & Shiffrin, R. H. The control of short-term memory. Scientific American, 1971, 225(2),82-90.

Atcneave, F. Application of information theory to psychology.New York: Holt, Rinehart and Winston, 1957.

Averbach, E., and Coriell, A. S. Short-term memory invision. Bell System Technical Journal, 1961, 40,309-328.

Averbach, E., and Sperling, G. Short-term storage of in-formation in vision. In E. C. ChLrry (Ed.), FourthLondon Symposium on Information Theory. London andWashington, D.C.: Butterworth, 1961.

50

54

Baum, R. A., Smith, J. S., & Goebel, R. A. Selection andanalysis of undergraduate pilot-training maneuversfor automated proficiency measurement development.AFHRL TR 72-62, AD 767580. Human ResourcesLaboratory, Flight Training Division, Williams AirForce Base, 1972.

Birmingham, H. P., & Taylor, F. V. A demonstration of theeffects of quickening in multiple-coordinate controltasks. USN NRL Report 4380, 1954.

Bourne, L. E., Jr., & Bunderson, C. V. Effects of delay ofinformative feedback and length of postfeedback in-terval on concept identification. Journal ofExperimental Psychology, 1963, 65, 1-5.

Bourne, L. E., Jr., & Haygood, R. C. The role of stimulusredundancy in concept identification. Journal ofExperimental Psychology, 1959, 58, 232-238.

Bourne, L. E., Jr., & Haygood, R. C. Supplementary report:Effect of redundant relevant information upon theidentification of concepts. Journal of ExperimentalPsychology, 1961, 61, 295-260.

Bourne, L. E., Jr., & Restle, F. A mathematical theoryof concept identification. Psychological Review,1959, 66, 278-296.

Braida, L. D., & Durlach, N. I. Intensity perception. II.

Resolution in one interval paradigms. Journal of theAcoustical Society of America, 1972, 51, 483-502.

Briggs, G. E. Pursuit and compensatory modes of informa-tion display: A review. USAF AMRL-TDR-62-93, AD-288 888.Aerospace Medical Research Laboratories, Wright-Patterson Air Force Base, Ohio, 1962.

Broadbent, D. E. The role of auditory localization inattention and memory span. Journal of ExperimentalPsychology, 1954, 47, 191-196.

51

nt

Broadbent, D. E. Immediate memory and simultaneous stimuli.Quarterly Journal of Experimental Psychology, 1957, 9,1-11.

Broadbent, D. E. Perception and communication. London:Pergamon Press, 1958.

Broadbent, D. E. Decision and stress. London: AcademicPress, 1971.

Brown, J. Some tests of the decay theory of immediatememory. Quarterly Journal of Experimental Psychology,1958, 10, 12-21.

Brown, R. Social psychology. New York: Free Press, 1965.

Bruner, J. S., Goodnow, J. J., & Austin, G. A. A study ofthinking. New York: Wiley, 1956.

Bryden, M. P. Order of report in dichotic listening.Canadian Journal of Psychology, 1962, 16, 291-299.

Bryden, M. P. The manipulation of strategies of reportin dichotic listening. Canadian Journal of Psystolom,1964, 18, 126-138.

Bryden, M. P. Attentional strategies and short-term memoryin dichotic listening. Cognitive Psychology, 1971, 2,99-116.

Bulgarella, R. G., & Archer, E. J. Concept identificationof auditory stimuli as a function of amount of relevantend irrelevant information. Journal of Experimental1-1vchology, 1962, 63, 254-257.

Corballis, M. C. Rehearsal and decay in immediate recallof visually and aurally presented items. CanadianJournal of Psychology, 1966, 20, 43-51.

Craik, F. I. M., & Lockhart, R. S. Levels of processing:A framework for memory research. Journal of VerbalLearning and Verbal Behavior, 1972, 11, 671-684.

52

FAS

Crowder, R. G., & Morton, J. Precategorical acousticstorage (PAS). Perception and Psychophysics, 1969,5, 365-373.

Denny, J. P., & Gamlin, P. Memory load and form of in-f input as determinants of proficiency inconcept formation. Paper presented at the 73d annualconvention of the American Psychological Association,Chicago, 7 September 1965 (Abstract: AmericanPsychologist, 1965, 20, 588).

Dodd, D. H., Kinsman, R. A., Klipp, R. D., & Bourne, L. E.,Jr. Effects of logic pretraining on conceptual rulelearning. Journal of Experimental Psychology, 1971,88, 119-122.

Durlach, N. I., & Braida, L. D. Intensity perception. I.

Preliminary theory of intensity resolution. Journalof the Acoustical Society of America, 1969, 46, 372-383.

Elliott, L. L. Development of auditory narrow-bandfrequency contours. Journal of the Acoustical Societyof America, 1967, 42, 143-153.

Fleishman, E. A. Performance assessment based on anempirically derived taxonomy. Human Factors, 1967,9, 149-366.

FUPT Study (Future Undergraduate Pilot Training SystemStudy). See Rusis, et al., 1971.

Garner, W. R. Uncertainty and structure as psychologicalconcepts. New York: Wiley, 1962.

Gibson, J. J. Motion picture testing and research. ReportNo. 7, Army Air Forces Aviation Psychology ProgramResearch Reports, 1947.

Gopher, D., & Kahneman, D. Individual differences in at-tention and the prediction of flight criteria.Perceptual and Motor Skills, 1971, 33, 1335-1342.

53

ZI.an 57

Gravetter, S., & Lockheed, G. R. Criterial range as aframe of reference for stimulus judgments. Psycho-logical Review, 1973, 80, 203-216.

Haygood, D. Audio-visual concept formation. Journal ofEducational Psychology, 1965, 56, 126-132.

Haygood, R. C., & Bourne, L. E., Jr. Forms of relevantstimulus redundancy in concept identification.Journal of Experimental Psychology, 1964, 67, 392-397.

Haygood, R. C., & Bourne, L. E., Jr. Attribute- and rule-learning aspects of conceptual behavior. Psycho-logical Review, 1965, 72, 175-195.

Haygood, R. C., & Devine, J. V. Effects of composition ofthe positive category on concept learning. Journalof Experimental Psychology, 1967, 74, 230-235.

Hays, W. L. Statistics for psychologists. New York:Holt, Rinehart, and Winston, 1963.

Henning, G. B. A comparison of the effects of signalduration on frequency and amplitude discrimination.In R. Plomp and G. Smoorenburg (Eds.), Frequencyanalysis and periodicity detection in hearing.Leiden: Sitjhoff, 1970.

Kepros, P. G. The identification of concepts as a functionof stimulus and response complexity. Unpublisheddoctoral dissertation, University of Utah, 1965.

Kidd, J. S. Human tasks and equipment design. In Gagne,R. M. (Ed.), Psychological principles in systemdevelopment. New York: Holt, Rinehart, and Winston,1962.

Kroll, N. E. A., Parkinson, S. R., & Parks, T. E. Sensoryand active storage of compound visual and auditorystimuli. Journal of Experimental Psychology, 1972,95, 32-38.

54

Leshowitz, B., & Cudahy, E. Masking with continuous andgated sinusoids. Journal of the Acoustical Societyof America, 1972, 51, 1921-1929.

Leshowitz, B., & Cudahy, E. Frequency discrimination inthe presence of another tone. Journal of theAcoustical Society of America, 1973, 54, 882-887.

Leshowitz, B., & Green, D. M. Comments on "Absolutejudgment and paired-associate learning: Kissingcousins or identical twins?" by J. A. Siegel and W.Siegel. Psychological Review, 1974, 81, 177-179.

Leshowitz, B., & Hanzi, R. Serial position effects fortonal stimuli. Memory and Cognition, 1974, 2, 112-116.

Leshowitz, B., Hanzi, R., & Zurek, P. M. Response-inducedinterference in recognition memory for briefly ex-posed visual arrays, 1973, in preparation.

Leshowitz, B., Zurek, P. M., & Robbins, D. Forgetting ofvisually presented words after retention intervalsfilled with detection of acoustic signals. Bulletinof the Psychonamic Societz, 1974, 3, 211-213.

Liang, C., & Chistovich, L. A. Frequency-difference limensas a function of tonal duration. Soviet PhysicalAcoustics, 1961, 6, 75-80.

Lordahl, D. S. Concept identification using simultaneousauditory and visual signals. Journal of ExperimentalPsychology, 1961, 62, 282-290.

Massaro, D. W. Preperceptual auditory images. Journal ofExperimental Psychology, 1970, 85, 411-417.

Massaro, D. W. Effect of masking tone duration on pre-perceptual auditory images. Journal of ExperimentalPsychology, 1971, 87, 146-148.

Massaro, D. W. Preperceptual images, processing time andperceptual units in auditory perception. PsychologicalReview, 1972, 79, 124-145. (a)

55

9

Massaro, D. W. Stimulus information vs processing time inauditory pattern recognition. Perception and Psycho-physics, 1972, 12, 50-56. (b)

Massaro, D. W., & Kahn, B. J. Effects of central process-ing on auditory recognition. Journal of ExperimentalPsychology, 1973, 97, 51-58.

McCormick, E. J. Human factors engineering. New York:McGraw-Hill, 1964.

Melton, A. W. Implications of short-term memory for ageneral theory of memory. Journal of Verbal Learningand Verbal Behavior, 1963, 2, 1-21.

Melton, A. W. Short- and long-term postperceptual memory:Dichotomy or continuous. In K. H. Pribram and D. E.Broadbent (Eds.), Biology of memory. New York:Academic Press, 1970.

Miller, G. A. The magical number seven, plus or minus two.Psychological Review, 1956, 63, 81-97.

Moray, N. Where is capacity limited. A survey and amodel. Acta Psychologica, 1967, 27, 84-92.

Morton, J., Crowder, R. G., & Prussin, H. A. Experimentswith the stimulus suffix effect. Journal ofExperimental Psychology Monograph, 1971, 91, No. 1,169-190.

Morton, J., & Holloway, C. M. Absence of cross-modal"suffix effects" in short-term memory. QuarterlyJournal of Experimental Psychology, 1970, 22, 167-176.

Murray, D. J., &in dichoticPsychology,

Hitchcock, C. H. Attention and storagelistening. Journal of Experimental1969, 81, 164-169.

Neisser, U. Cognitive psychology. New York: Appleton-Century-Crofts, 1967.

56

Parkinson, S. R. Variability and control in dichoticmemory. Journal of Experimental Psychology, 1974,102(1), 67-80.

Peterson, L. R., & Peterson, M. J. Short-term retentionof individual verbal items. Journal of ExperimehtalPsychology, 1959, 5C, 193-198.

Pishkin, V., & Wolfgang, A. Number and type of availableinstances in concept learning. Journal of Experi-mental Psychology, 1965, 69, 5-8.

Quastler, H. Information theory in psychology: Problemsand methods. Glencoe: Free Press, 1955.

Posner, M. I., & Boies, S. J. Components of attention.Psychological Review, 1971, 78, 391-408.

Posner, M. I., & Rossman, E. The effect of size andlocation of interpolated information reducing trans-forms upon short-term retention. Journal of Experi-mental Psychology, 1965, 70, 496-505.

Reitman, J. S. Short-term verbal retention with inter-polated verbal and nonverbal signal detection(Doctoral dissertation, The University of Michigan)Ann Arbor, Michigan: University Microfilms, 1969.No. 70-14622. Also, Communication No. 262, MentalHealth Research Institute, The University ofMichigan, 1969.

Reitman, J. S. Mechanisms of forgetting in short-termmemory. Co &nitive Psychology, 1971, 2, 185-195.

Restle, F. The selection of strategies in cue learning.Psychological Review, 1962, 69, 329-343.

Ronken, D. A. Some effects of bandwidth-duration con-straints on frequency discrimination. Journal of theAcoustical Society of America, 1971, 49, 1232-1242.

Ronken, D. A. Changes in frequency discrimination causedby leading and trailing tones. Journal of theAcoustical Society of America, 1972, 51, 1947-1590.

57

Roweton, W. E., & Davis, G. A. Effects of preresponseinterval, post-informative feedback interval, andproblem difficulty on the identification of concepts.Journal of Experimental Psychology, 1968, 78, 642-645.

Rusis, G., Spring, W. G., & Atkinson, J. M. Future under-graduatePerformance Measures. NOR 70-149, AeronauticalSystems Division, AFSC, Wright-Patterson Air ForceBase, 1971.

Shannon, C. E., & Weaver, W. The mathematical theory ofcommunication. Urbana: University of Illinois Press,1949.

Siegel, J. A., & Siegel, W. Absolute judgment and paired-associate learning: Kissing cousins or identicaltwins? Psychological Review, 1972, 79, 300-316.

Siegel, W. Memory effects in the method of absolutejudgment. Journal of Experimental Psychology, 1972,94, 121-131.

Sleight, R. B. The effect of instrument dial shape onlegibility. Journal of Applied Psychology, 1948, 32,170-188.

Smode, A. F., Hall; E. R., & Meyer, D. E. An assessmentof research relevant to pilot training. AMRL-TR-66-1966, AD-804 600. Aerospace Medical ResearchLaboratories, Wright-Patterson Air Force Base,November, 1966.

Sperling, G. The information available in brief visualpresentation. Psychological Monographs, 1960, 74(Whole No. 498),

Sternberg, S. Memory scanning: mental processes revealedby reaction-time experiments. American Scientist,1969, 57, 421-457.

58

G2

Swets, J. A. Signal detection and recognition by humanobservers. Contemporary readings. New York: Wiley,1964.

Waugh, N. C., & Norman, D. A. Primary memory.Psychological Review, 1965, 72, 89-104.

Zwislocki, J. J. A theory of central auditory maski...4and its partial validation. Journal of the AcousticalSociety of America, 1972, 52, 644-659.

59