CURRENTOmcriONSm PSYCHOidG1CAi>SaRNa

Sun-Compass Learning in Insects:Representation in a Simple MindFred C. Dyer and Jeffrey A. Dickinson

Although most research in cog-nitive psychology focuses primari-ly on human beings—or on verte-brate animals used as models forstudying human b e h a v i o r -insects and other invertebrates po-tentially have an important role toplay in the study of cognition. Thebehavioral capacities of such or-ganisms present a rich array ofphenomena that can test the gen-erality of theories of informationprocessing by biological systems.Invertebrates have long served asmodels for elucidating basic sen-sory and neural mechanisms.More recently, some cognitive sci-entists/'^ defying conventionalwisdom, have suggested that in-sects may also offer insights intohow nervous systems are orga-nized to acquire and use complexinternal representations of eventsor relationships in the outsideworld.

This article examines a phenom-enon of learning in honeybees thatat least partially vindicates thisgenerous view of insect cognition.The phenomenon in question isthe ability, which is widely sharedby other animals, to memorize thesun's pattern of movement over

Fred C. Dyer is Associate Profes-sor in the Department of Zoologyat Michigan State University. Jef-frey A. Dickinson is a graduatestudent in zoology and in theProgram in Ecology and Evolu-tionary Biology at MichiganState. Address correspondenceto Fred Dyer, Department of Zo-ology, Michigan State Univer-sity, East Lansing, MI 48824;e-mail: fcdyer@msu.edu.

the day and to use this stored pat-tern for navigation. We review ev-idence that this ability constitutesa truly representational form oflearning, and then discuss recentstudies that provide new insightsinto processes that underlie thedevelopment of the bee's internalrepresentation of the sun's course.We then point out some generalinsights that could emerge fromfurther studies of this system in in-sects.

ORIENTATION USING ACELESTIAL COMPASS

A wide variety of vertebrate andinvertebrate species can use theazimuth of the sun (its compass di-rection) to steer paths betweenfixed points on the earth's surface.Honeybees also refer to the sun(and sun-linked patterns of polar-ized light in blue sky) to commu-nicate the compass direction offood in their dance language.̂ ^ Us-ing the sun as a compass requiresan ability to compensate for itsmovement during the day. Thisability implies in turn a mecha-nism that integrates informationabout the sun's changing direction(which in bees and many otherspecies is measured relative tolandmarks) with informationabout time of day (which in mostanimals is provided by an internalclock synchronized to the light-dark cycle).

Although many animals uselandmarks to measure changes inthe sun's direction, most speciescan navigate using the sun in un-familiar terrain. Such an ability im-

plies that the pattern of solarmovement, once learned, pro-vides a directional reference inde-pendent of the reference used tolearn it.

An additional challenge is thatthe azimuth changes at a variablerate over the day, shifting rela-tively slowly in the early morningand late afternoon and relativelyquickly near midday (Fig. 1). Fur-thermore, the daily pattern ofchange, which astronomers callthe solar ephemeris, changes withseason and with latitude (Fig. lb).Animals that are physiologicallyequipped to be active over a rangeof seasons and latitudes would dowell to learn the current localephemeris and use it as the basisof their celestial compass.

'::::-r THE HONEYBEE'S•'•^^ '̂•^^H; CELESTIAL COMPASS y

Since the 1950s, scientists haveknown that honeybees use a celes-tial compass that has all of theseproperties^ and that the bees'knowledge of the current localephemeris develops throughlearning.* The frame of referenceby which bees detect and memo-rize the sun's pattern of movementis provided by landmarks sur-rounding the nest. Experiencedbees draw upon this memory todetermine the solar azimuth oncompletely overcast days, whenclouds obscure both the sun andpatterns of skylight polarization.^Their memory allows them to per-form dances communicating thedirection of food relative to the so-lar azimuth even when they havenot seen celestial cues during theflight to the food. Bees performdances in the nest on the verticalsheets of comb, encoding flight di-rection in the orientation of thebody relative to gravity (where theupward direction represents thecurrent direction of the sun; see

Copyright © 1996 American Psychological Society 67

00-E

10 ^2 14 16Local Solar Time

Fig. 1. The solar ephemeris. The sun'scourse on the summer solstice in EastLansing, Michigan (=43°N) is shownin (a). The circles show the hourly po-sitions of the sun from 0500 hr to 1900hr as it makes its transit across the sky.The sun moves at a constant rate (157hr) along its arc, but the rate of changeof the azimuth (its compass direction)varies over the day as the sun risestoward the zenith (cross) and then de-scends again. Shaded sectors show thechange in azimuth over two equal (2-hr) time periods. The rate of change inthe morning period, when the sun islow in the sky, is 9.57hr; the rate ofchange at midday, when the sun ishigh in the sky, is 37.57hr. An alter-native method for displaying changein solar azimuth over the day (the so-lar ephemeris) is shown in (b). Theheavy line (a) plots the sun's courseshown in (a). The other curves repre-sent the ephemerides at the equatoron the December solstice (b), when theazimuth shifts clockwise (left to right)across the southern horizon, and onthe June solstice (c), when the azimuthshifts counterclockwise (right to left)across the northern horizon.

Fig. 2). A bee that has returnedfrom a flight under a cloudy skybases her dance on the solar azi-muth encoded in memory for thattime of day. Honeybees' dancesincidentally provide human ob-

servers a window on the contentsof the bees' memory of the sun'scourse. For example, over severalhours of cloudy weather, bees'dances to a fixed feeding sitechange in accordance with thechange in the azimuth of the sun,indicating that the bees can storethe details of the current solarephemeris in memory.

How could a bee form an inter-nal representation of the solarephemeris as a result of her expe-rience seeing the sun moving rela-tive to the terrain? In pioneeringstudies of sun-compass learning,Lindauer''^ excluded the obviouspossibility that bees simply mem-orize a list of time-linked solar po-sitions, thus compiling a sort oflookup table recording their expe-rience. He found instead that beescan estimate the sun's position attimes of day when they have neverseen it. In his experiments, hetrained incubator-reared bees tofly to a feeding place in the southfrom noon onward for severaldays, but otherwise deprived thebees of an opportunity to leave thenest. When Lindauer tested thebees' ability to find the compassdirection of the food (in a newlandscape, so that they could notuse familiar landmarks), he did soin the morning. The bees searchedpredominantly in the trained com-pass direction. This finding im-plied that they knew somethingabout the course of the sun in themorning (as the azimuth movesfrom east to south), although theyhad previously seen it only in theafternoon (as the azimuth shiftsfrom south to west). Other studieshave strengthened this conclusionby showing that insects can esti-mate the sun's position at night.^

PREVIOUS MODELS

Thus, whatever learning mech-anism underlies the development

of the sun compass in insects, itmust perform some sort of compu-tation that fills gaps in an animal'sexperience of the sun's course,and hence derives from incom-plete sensory information a repre-sentation that encodes, at leastroughly, the sun's entire course.This is the sort of phenomenonthat gave rise to the cognitive rev-olution in psychology: a behaviorthat can be understood only byconsidering the internal processesthat organize sensory informationand encode representations of ex-ternal events.

Several studies have examinedhow an insect could fill a gap in itsexperience of the sun's course.The general assumption has been

Fig. 2. Communication of flight direc-tion in honeybee dances. The top ofthe figure shows the sun at three suc-cessive positions along its course, andthe resulting change in the angle be-tween the solar azimuth and the flightdirection from a bee hive to a fixedfeeding site. At the bottom of the fig-ure are three dances corresponding tothese three positions of the azimuth.In the dance, a bee repeatedly turns afigure-eight pattern. The central ele-ment in this pattern is the "wagglingrun," in which the bee runs in astraight line while waggling her bodyvigorously from side to side. Thedancer's orientation relative to gravityduring the waggling run encodes thedirection of the food relative to thecurrent solar azimuth.

Published by Cambridge University Press

CURRENT DIRBCTlONSm PSYCHOLOGICAL SCIENCE

that the insect records a set oftime-linked solar azimuth posi-tions at times of day when she seesthe sun, and then derives fromthis information a single rate ofcompensation for estimating so-lar movement at other times ofday. Three distinct computationalmechanisms have been proposed,each with experimental or obser-vational support. First, insectsmight estimate unknown posi-tions of the sun by interpolating ata uniform rate between temporallyadjacent known positions of thesolar azimuth.^ Second, theymight extrapolate forward fromthe most recent known position ofthe sun, using the rate of azi-muthal shift observed at thattime.^'^ Third, they might extrap-olate backward, using the sun'sazimuth and rate of movement ob-served at a later time of day.^ Mostprevious evidence favored the in-terpolation hypothesis, but the ex-trapolation mechanisms could notbe excluded as explanations forcertain observations. All three hy-potheses could account for Lin-dauer's original results.

A NEW APPROACH

We developed a new experi-mental approach^ in the hope ofdeciding which of these hypothe-ses best accounts for bees' abilityto estimate unknown positions ofthe sun. To our surprise, our re-sults led us to reject all three pre-vious models of sun learning inhoneybees. Instead, our data wereconsistent with a completely dif-ferent model that resembles oneproposed independently to ac-count for sun learning by ants.^°

Following Lindauer,** we usedincubator-reared bees and allowedthem to see only a small portion ofthe sun's daily course for at least 7days. During their daily flighttime, in the 4 hr before sunset, we

trained bees to find a dish ofscented sugar water in a particularcompass direction, and labeledthem individually so that we couldkeep track of their experience.Then we examined how these af-ternoon-experienced bees esti-mated the course of the sun whenthey flew for the first time in themorning and middle of the day. Todo this, we allowed the bees toleave the hive only when the skywas overcast, and then observedtheir dances when they returnedfrom the food. As mentioned, theorientation of a dancer relative togravity signals the angle that shehas flown relative to the sun (Fig.2). Because we observers knew thedirection the bees had flown, wecould infer from each bee's dancewhere she had determined the sun

to be relative to the line of flight.By observing bees on cloudy days(when they could not see the sun'sposition during the flight) at timesof day when they had never flownbefore (so they could not rely on aremembered position of the sun),we hoped that the dances wouldtell us how the bees filled the largegap in their experience of the sun'scourse.

We performed this experimentwith two colonies of incubator-reared bees and obtained similarresults for both colonies.^ Figure 3shows the data from one colony.Each data point is the solar azi-muth inferred from a single danceby a bee visiting the hive betweentrips to the food. We observed 554dances by 46 different bees in thiscolony. The data fall into a striking

Training period

360-

315-

270-

C 2 2 5 -

I 180-

•^ 1 3 5 -

90-

45 -

0 T I I I I I I 1 I I I I I6 9 12 15 18

Local Solar Time21

Fig. 3. Solar positions estimated by 46 partially experienced (incubator-reared) beesduring an entire cloudy day, as inferred from 554 dances performed in the hivebetween visits to a feeder.^ The bees had previously been exposed to a smallportion of the sun's course during the late afternoon (shaded area). The cloud coverobscured celestial cues, and hence required bees to remember (or estimate) thesun's position. Each symbol is the solar azimuth inferred for a single dance. Theopen symbols (circles and triangles) show data from 2 bees that departed qualita-tively from the pattern exhibited by the rest. The curved line shows the solarephemeris at the time of this experiment. The dashed lines show the predictions ofthree different computational mechanisms that had been proposed to underlie theability of bees to estimate unknown positions of the sun (see the text): 1, linearinterpolation at a uniform rate between the solar azimuth at the end of one dailytraining period and the azimuth at the beginning of the next; 2, forward extrapo-lation from the position and rate of movement of the azimuth at the end of tine dailytraining period; 3, backward extrapolation from the position and rate of movementof the azimuth at the beginning of the daily training period (as observed by the beeson days prior to the test).

Copyright © 1996 American Psychological Society

pattern that was highly consistentamong bees (with only two quali-tative exceptions, discussed in thenext section), and an:iong differentdances by individual bees. In themorning, the bees used a solar az-imuth that was almost exactly 180°from the average solar azimuth ex-perienced during the daily trainingperiod. They continued to use thisangle until a short interval nearmidday (well before the normalstarting time of the training pe-riod), when they changed theirdances by 180° and began usingthe evening azimuth of the sun.The linear interpolation and ex-trapolation mechanisms proposedin previous studies do a very poorjob of describing this pattern. Inparticular, because these hypothe-ses assume that bees derive a sin-gle linear rate of compensation toestimate unknown portions of thesun's course, none predicts thebees' abrupt midday transitionfrom the morning angle to the af-ternoon angle. If anything, thebees' estimate of the sun's coursemost closely resembles the actualcourse of the sun, but the thick-ness of the cloud cover, as well asother evidence, precluded the pos-sibility that the bees had detectedthe (approximate) position of thesun during the flight.^

We suggest that our partiallyexperienced bees based theirdances on an internal representa-tion that approximated the actualpattern of solar movement overthe day, even at times when thebees had never seen the sun. Inboth the internal ephemeris andthe real ephemeris, the sun risesopposite where it sets (as mea-sured relative to features of the lo-cal landscape), with the azimuthchanging relatively slowly duringthe morning and during the after-noon, and moving from one sideof the local meridian to the other atmidday. In the conditions of ourexperiment, the bees' approximateephemeris was well described by a

step function that switched instan-taneously by 180° at midday, butbees that did several successivedances during the midday transi-tion actually changed their orien-tation progressively (if rapidly),implying that the underlying rep-resentation would be better de-scribed by a continuous function.We also found, as in previousstudies, "• that bees whose flight ex-perience spanned a larger portionof the day developed an internalephemeris that more closelytracked the current solar ephem-eris.

AN INNATE MODEt OFTHE SUN'S COURSE

Our experiments imply thatbees are innately informed of theapproximate dynamics of solarmovement over the day. As beesbegin to acquire flight experience,this innate information allowsthem to develop an internal repre-sentation that reflects portions ofthe sun's course previously seenand approximates the pattern ofsolar movement at other times ofday. With further experience, thisinnate template is modified to con-form more closely to the actualcourse of the sun.̂ *^

As mentioned, 2 bees deviatedqualitatively from the pattern thatled us to this interpretation, andyet these exceptions conform toour general hypothesis. Thesebees, whose dances are indicatedby open symbols in Figure 3, usedthe afternoon azimuth early in theday and the morning azimuth latein the day. One bee switched atmidday (when all the other beesswitched from using the morningazimuth to using the afternoon az-imuth). The other switched late inthe afternoon. These 2 bees resem-ble the 44 others that we observedin using internal representations

encoding (approximately) the en-tire pattern of solar movement, buttheir ephemerides appear to havebeen out of phase (either spatiallyor temporally) with the one usedby the other bees.

The hypothesis that insects areequipped with an innate templateencoding the sun's approximatecourse not only explains our re-sults^ and recent data from ants^°better than do the linear interpola-tion and extrapolation hypothe-ses, it is also consistent with muchof the data that was considered tosupport these earlier hypothe-ses '̂̂ and thus may lead to a moregeneral account of the develop-ment of the sun compass, at leastin insects.^

The mechanism implied by ourhypothesis would offer certain ad-vantages to a small-brained, short-lived organism. The bee's task is to"discover" the current local solarephemeris, and presumably to doso as quickly as possible: Honey-bee foragers have only about 10days in which to collect food forthe colony before they die, andthey make only a few flights fromthe nest prior to beginning forag-ing. The bees' default knowledgeof the sun's general pattern ofmovement could simplify (andspeed up) the process of discover-ing the actual ephemeris. Our re-sults suggest, for example, that the"initialized" state of the neuralnetwork that encodes the ephem-eris approximates the state at-tained after an actual ephemerishas been learned. Intriguingly, theaverage solar ephemeris experi-enced by organisms on Earth (i.e.,the ephemeris averaged over alllatitudes and all days of the year)is a 180° step function (Fig. 4). Aneural network that by default en-codes an ephemeris of this shape(as appears to be the case in ourbees) would presumably have tochange less on average to conformto any given actual ephemeris thanone that by default estimates the

Published by Cambridge University Press

CURRENT DlRECTIOtsIS IN PSYCHOLOGiCAL SCltNCd

B 10 12 14 16 ISLocal Solar Time

Fig. 4. Actual patterns of solar move-ment on Earth. The shaded area cov-ers all possihle solar ephemeris func-tions that could be observed on theEarth's surface (i.e., all latitudes andall days of the year). The mean of allthese functions is the step functionshown by the heavy line (the dashedvertical line at noon indicates that themean azimuth at noon is indetermi-nate). The mean ephemeris essentiallymatches the approximate ephemerisdeveloped by our partially experi-enced bees (see Fig. 3). Shown forcomparison are the actual solarephemeris functions for three latitudeson the summer solstice: a, 43''N (EastLansing, Michigan); b, 25°N; c, 15° N.

solar azimuth at random at differ-ent times of day.

Furthermore, an approximateephemeris in the shape of a 180°step function would allow a bee tobegin to use her sun compass withrelatively little error throughoutthe day even if she had sampledonly a small portion of the sun'scourse (Fig. 4). This is especiallytrue during much of the year atlow latitudes, where the actual so-lar ephemeris resembles the step-shaped representation developedby our afternoon-experiencedbees: The azimuth changes littlethroughout the morning, shifts byabout 180° at local noon, thenchanges little throughout the after-noon. By contrast, the interpola-tion and extrapolation mecha-nisms hypothesized previouslycould achieve similar accuracyonly after a bee pieced together aninternal ephemeris through exten-sive sampling of the sun's courseat different times of day.

INSECTS,

Sun-compass learning in insectsappears to rely upon a purpose-built learning mechanism appro-priate for a small-brained animal,but we believe that further studiesof this mechanism may illuminatebroader questions about learningand cognition. For one thing, acritical property of the bee's mech-anism is that general features ofthe pattern to be learned are in-nately encoded. As ethologistshave long known, learning pro-cesses in many animals, includingvertebrates, have this property. Arelatively simple example is songlearning in birds: Certain speciesseem predisposed to learn soundsthat are patterned like their spe-cies-specific song.^^ A more com-plex example is language learningin children, which appears to beguided by innate rules for recog-nizing speech sounds and organiz-ing them into words and grammat-ical sentences.^^ Sun-compasslearning in insects may provide anopportunity to identify neurobio-logical mechanisms by whichinnately specified information me-diates the integration of newly ac-quired information by the nervoussystem.

Another issue that might be il-luminated by further studies ofsun-compass learning is the role oftime as a variable in the neuralcode by which the brain representscertain environmental patterns.Many species share with bees andants the ability to learn the patternof solar movement relative to timeof day. Also, many animals, in-cluding bees, can learn the occur-rence of events such as the avail-ability of food in relation to time ofday, as read from their internalclocks.^ Although molecular neu-robiologists are closing in on themechanisms underlying biologicaltimekeeping, scientists have virtu-ally no idea how the clock pro-

vides input about time of day toneural circuits involved in learn-ing. Conceivably, the bees' sun-learning mechanism could serve asa model system for exploring thisquestion.

Finally, further studies of sun-compass orientation may provideinsights into the evolutionary de-sign of learning mecharusms. Thecorrespondence between the bee'sdefault knowledge and the sun's ac-tual pattern of movement bespeaksan important role for natural selec-tion in shaping the underlyingmechanisms. The key evolutionaryquestions concern how these mech-anisms arose, how they differ frommechanisms underlying other learn-ing phenomena (including those inother species), and why (in terms ofsurvival value) they have the specificproperties that they do. Exploring theadaptive specialization of learningmechanisms, an area of intense theo-retical interest,̂ '̂ '̂ ^ requires combin-ing a broad comparative perspectivewith focused study of specific mech-anisms. Many animals orient usingthe sun as a compass, and yet mustsolve very different specific tasks inlearning the sun's course. Consider,for example, that a long-lived andwide-ranging animal sud\ as a birdneeds to recalibrate its internalephemeris throughout its life to trackthe varying patterns of solar move-ment that it experiences,^^ whereas ashort-lived and sedentary animalmay need to learn only the solarephemeris experienced early in itslife. Conceivably, such comparisonscould provide insights into hownatural selection has adapted thelearning mechanism underlyingsun-compass orientation to fulfilldifferent navigational demands.

Acknowledgments—The research de-scribed in this article was supported bygrants from the U.S. Nationai ScienceFoundation.

Copyright © 1996 American Psychological Society

Notes

1. C.R. Gallistel, The Organization of Learning(MIT Press, Cambridge, MA, 1990).

2. P.S. Churchland, V,S. Ramachandran, andT.J. Sejnowski, A critique of pure vision, in Large-Scale Neuronal Theories of the Brain, C. Koch andJ.L. Davis, Eds. (MIT Press, Cambridge, MA,1994).

3. K. von Frisch, The Dance Language and Ori-entation of Bees (Harvard University Press, Cam-bridge, MA, 1967).

4. M. Lindauer, Angeborne und erlernteKomponenten in der Sonnenorientierung der Bi-enen, Zeitschrift fUr vergleichende Physiologie, 42,43-62 (1959).

5. F.C. Dyer, Memory and sun compensationby honey bees, journal of Comparative Physiology A,160, 621-633 (1987),

6. F.C. Dyer, Nocturnal orientation by the

Asian honey bee. Apis dorsata. Animal Behaviour,33, 769-774 (1985), and publications by Lindauer,Edrich, and Wehner cited therein.

7. D.A.T. New and J.K. New, The dances ofhoney bees at small zenith distances of the sun,journal of Experimental Biology, 39, 271-291 (1962);R. Wehner, Himmelsnavigation bei lnsekten:Neurophysiologie und Verhalten, Neujahrsblattder Naturforschenden Gesellschaft in Zurich, 184,1-132 (1982): R. Wehner and B. Lanfranconi,What do ants know about the rotation of the sky?Nature, 293, 731-733 (1981).

8. J.L. Gould, Sun compensation by bees. Sci-ence, 207, 545-547 (1980).

9. F.C. Dyer and J.A, Dickinson, Develop-ment of sun compensation by honeybees: Howpartially experienced bees estimate the sun'scourse. Proceedings of the National Academy of Sci-ences, USA, 91, 4471^*474 (1994).

10. R. Wehner and M. Miiller, How do ants

acquire their celestial ephemeris fimcbon?wissenschaften, 80, 331-333 (1993).

11 P. Marler, Song-learning: The interface be-tween behavior and neuroethology. PhilosophicalTransactions of the Royal Society of London B, 329,109-114 (1990).

12. S. Pinker, The Language Instinct (MITPress, Cambridge, MA, 1994); N. Chomsky, Onthe nature, use, and acquisition of language, inReadings in Philosophy and Cognitive Science, A.I.Goldman, Ed. (MIT Press, Cambridge, MA,1993).

13. D.F. Sherry, L.F. Jacobs, and S.J. Gaulin,Spatial memory and adaptive specialization of thehippocampus. Trends in Neurosciences, 15, 298-303(1992).

14. K, Schmidt-Koenig, J.U. Ganzhom, andR. Ranvaud, The sun compass, in Orientation inBirds, P. Berthold, Ed. (Birkhauser-Verlag, Basel,Switzerland, 1991).

Visual Perception of Locationand DistanceJack M. Loomis, Jose A. Da Silva, John W. Philbeck, andSergio S. Fukusima

Visual perception of three-dimensional space is one of theclassic problems in philosophyand experimental psychology. It isimportant for two reasons. First,explaining the phenomenology ofvisual experience, of which spaceperception is central, is one of themajor concerns of research on con-Jack M. Loomis and John W.Philbeck are, respectively. Pro-fessor and graduate student inthe Department of Psychology atthe University of California,Santa Barbara. Jose A. Da Silvaand Sergio S. Fukusima are Pro-fessors in the Department of Psy-chology at the University of SaoPaulo, Ribeirao Preto, Sao Paulo,Brazil. Address correspondenceto Jack M. Loomis, Departmentof Psychology, University of Cal-ifornia, Santa Barbara, CA 93106-9660, or Jose A. Da Silva, Depart-ment of Psychology, FFCLRP,University of Sao Paulo, RibeiraoPreto, SP, Brazil, CEP 14050-901.

sciousness, mental events, andhuman cognition. Second, visualspace perception plays an essentialrole in the control of much of hu-man spatial behavior.

WHAT NEEDS TOBE EXPLAINED?

Research on the topic has ad-dressed two quite distinct empiri-cal domains: the psychophysics ofvisual space and the visual controlof action. Researchers concernedwith the former have been inter-ested in such issues as the map-ping between physical and visualspace; the intrinsic geometry of vi-sual space; the stimulus cues andinternal constraints that determinevisual space; the interrelationshipof perceived direction, distance,size, and motion; and the sensorymechanisms and neural compu-tations involved in perceivingspace.^ Researchers concerned

with the latter have focusedlargely on how visual informationis used by the observer in the con-trol of spatial behavior, such asreaching, ball catching, running,or driving.^

It is unfortunate that the pro-grams of research on visual spaceand on the control of action havebeen conducted so independentlyof one another. The majority of vi-sual perception researchers proba-bly believe they are justified in in-vestigating the psychophysics ofvisual space in isolation hecausethe process of visual perceptionacts as a module functionally dis-tinct from other modules involvedin controlling action (e.g., themodule that specifies the com-mands to the extraskeletal mus-cles). Furthermore, the output ofthe perception module—visualspace—exists independently ofany of the actions to he controlled.In this view, once the process ofvisual perception and the structureof visual space have been workedout, this knowledge can be readilyapplied to the problem of control-ling action.

Taking issue with this view,however, are those researchersworking on the control of actionwithin the ecological framework.Their opposing view is that action

Published by Cambridge University Press