gibson afford ances

7/30/2019 Gibson Afford Ances

1/7

Psychological Review1994, Vol. 101,No. 2, 336-342 Copyright 1994 by the American Psychological Association, Inc.0033-295X/94/$3.00

Gibson's AffordancesJames G. Greeno

Gibson developed an interactionistview ofperception and action that focused on information that isavailable in the environment. Hethereby rejected the still-prevalent framing assumption of factoringexternal-physical and internal-mental processes. The interactionist alternative, which focuses onprocesses of agent-situation interactions, is taken in ecological psychology as well as in recent re-search on conversational communication, research on complex, socially organized activity, andphilosophical situation theory. The concepts of qffordance and ability are key ideas in an interac-tionist account. In situation theory, abilities in activity depend on attunements to constraints , andaffordances for an agentcan be understood asconditions in the environment fo r constraints to whichthe agent is attuned. This broad viewofaffordances includesaffordances that are recognized as wellas affordances that are perceived directly.

In his 1954 article on visual perception of motion andmove-ment, Gibson discussed several ways in which perceptions ofmotion and movement have to be understood relationally. Ashe remarked, citing Kofflca (1935), "Just as a motion for thephysicist can bespecified only in relation to a chosen coordinatesystem, so is a phenomenal motion relative to a phenomenalframework" (J. J. Gibson, 1954, p. 310). Most of thepsychologyofperception that had been constructed was, and is, about phe-nomena that occur whenan observer is stationary. Implicitly inthis article, and more explicitly in his later, more comprehensivetheorizing (1966), Gibson argued that a psychology of percep-tion that is only about stationary observation neglects some ofthe crucial characteristics of what it claims to be about.

Anothercommitment ofGibson'scontributions was to a psy-chology of perception that avoids subsuming perceptual phe-nomena inappropriately to an apparently more comprehensivetheory. I recall a meeting in which an animal psychologist wasreporting on observations of the development of sex differencesin young dogs, and he said that one function of urinating bymale dogs was to leave messages. Gibson said, "They leave mes-sages? What do they say?" Although there surely is some inter-canine function of urination involving territory, Gibson re-sisted the casual characterization of that process in terms ofsymbolic communication.

On another issue, J. J. Gibson and E. J. Gibson, in articlesalso published in Psycho lo gica l Review (J. J. Gibson & Gibson,1955a, 1955b), argued that perceptual learning should not besubsumed by the general stimulus-response theory that wasthen at the center of scientific research and thinking about ac-tion and learning. They proposed that the psychology ofpercep-tual learning should be about learning to perceive more of the

James G. Greeno, School of Education, Stanford University, and theInstitute for Research on Learning, Palo Alto, California.This research was supported by the National Science FoundationGrantMDR-9154119.Correspondence concerning this article should be addressed to JamesG. Greeno, School of Education, Stanford University, Stanford, Cali-fornia 92305.

differentiating qualities of stimuli in the environment ratherthan acquiring associated responses that cause greater differ-entiation by enrichment ofstimuli as a result ofpast experience.Postman (1955) criticized Gibson and Gibson's (1955a) discus-sion, arguing that "descriptively, perceptual learning is the at-tachment of new responses, or a change in the frequency of re-sponses, to particular configurations or sequences of stimuli"(Postman, 1955, p. 441) and that "the need to account forchanges in response inevitably endows the problem of percep-tual learning with an associative component" (Postman, 1955,p. 442). The Gibsons were not convinced by Postman. In theirview, "the main difficulty in the way of the traditional enrich-ment theory is its implication that learning involvesa decreas-ing psychophysical correspondence between perception andstimulation" (Gibson & Gibson, 1955b, p. 448). They reiter-ated their previous contention that "perceptual learning . . .consists of responding to variables of physical stimulation notpreviously responded to. The notable point about this theory isthat learning is always supposed to be a matter of improve-mentof getting in closer touch with the environment" (Gib-son&Gibson, 1955a, p. 34).

In their discussions of perception and perceptual learning inthe 1950s, J. J. Gibson and E. J. Gibson did not present a broadtheoretical framework in which their views were encompassed.In their continued work overthe years, however, strong, system-atic theorizing was an important part of their contributions. Inthe 1960s, E. J. Gibson (1969) proceeded to develop a generaltheory of perceptual learning anddevelopment, and J. J. Gibsondeveloped a general theoretical framework for perception andsensation, which he presented in The Senses C o n s i d e r e d as Per-ceptual Systems (J. J. Gibson, 1966). Cognitive science was inthe early stage of its development around the core idea of infor-mation, and the main stream of cognitive science developed atheory of information processing. Gibson differed, and he fo-cused on the question of what information is available. In hisview, many questions about howinformation is constructed bypeople and animals could be considered better as questionsabout what sources of information there are in the environmentthat people and animals use in their activities. This frameworkis an alternative to the mainstream view, in which people and

336


2/7

SPECIAL ISSUE: GIBSON'S AFFORDANCES 337animals are thought to construct theworld that they live in andunderstand. In Gibson's view, people and animals are attunedto variables and invariants of information in their activities asthey interact as participants with other systems in the world thatwe inhabit.Gibson's view of perception has been difficult for many cog-nitive scientists (e.g., Fodor & Pylyshyn, 1981; Ullman, 1980;Vera & Simon, 1993) to understand. I believe this is becauseGibson's reasoning involves some quite general framing as-sumptions about activity and cognition that differ from those ofmainstream cognitive science. He expressed some of these moregeneral ideas in the 1970s in discussion of the concept ofqffbr-d a n c e s (L J. Gibson, 1977, 1979/1986), and the task of devel-oping a systematic general ecological psychology is being car-ried on by a growing group of investigators (e.g., Shaw, Turvey,& Mace, 1982; Turvey, 1990,1992).Situativity and the Concepts of Affordance and AbilityBickhard and Richie (1983) argued that Gibson's thinking

evolved from a view of perception as encoding features of theenvironment toward a more general view of perception as anaspect of a person's or animal's interaction with the environ-ment. The encoding view, which is still prevalent in informa-tion-processing psychology, involves analyzing cognition interms of a factoring assumption that supports analyses ofdifferent stagesof cognitive processing in relative isolation fromeach other.A claim of ecological psychology, as I understand it, is thatthe interactions among aspects of cognition and behavior aresufficiently subtle and complex that our prevalent factoringstrategy is scientifically unproductive. As Turvey has put it,"The types of phenomena that should lead the way must bedrawn from perception in the service of action and from actionin the service of perception" (Turvey, 1992, p. 86). When per-ception, motor movement, memory, reasoning, or whatever, isstudied as a separate factor, one hopes that the conclusions onetakes from those studies apply in situations where other factorshave significant roles. Although there have been occasional ob-jections to the factoring assumption (e.g., by Dewey, 1896;Lashley, 1951), factoring of processesespecially into eventsoccurring outside and inside the mental systemhas been apersistent methodological commitment of psychological re-search. Gibson was already suspicious that perception and theobserver's movement did not factor neatly in the perception ofmotion when he wrote "The visual perception of objective mo-tion and subjective movement" (J. J. Gibson, 1954). That lackof factoring became a major point of his more general theory ofperception (J. J. Gibson, 1966) and has been a key issue in thedevelopment of ecological psychology, forexample, in the anal-ysis of information that is available to a person or animal mov-ing through a spatial environment (Cutting, Springer, Braren,& Johnson, 1992; Lee, 1980). An important feature of the eco-logical view involvesa shift in situations that are taken as para-digmatic cases of cognition. Rather than building a theory ofperception on analyses of situations with stationary observers,and building a theory of action on analyses of situations wherean agent tries to reproduce a movement of an object in space,ecological psychologists are working to build a theory ofactivity,

including perception and movement, byanalyzingsituations in-volving continuous interactions, such as cascade juggling(Beek& Turvey, 1992) and pursuit of a batted fly ball by a baseballoutfielder (Michaels &Oudejans, 1992).The framing assumptions of ecological psychology are oneform of a general theoretical stance, which can be called situa-tivity theory (Greeno & Moore, 1993), in which cognitive pro-cesses are analyzed as relations between agents and other sys-tems. This theoretical shift does not imply a denial of individualcognition as a theoretically important process. It does, however,involve a shift of the levelof primary focus of cognitive analysesfrom processes that can be attributed to individual agents tointeractive processes in which agents participate, cooperatively,with other agents and with the physical systems that they in-teract with. If, in analyzing hose interactive processes, onecon-cludes that some of them factor conveniently into aspects thatcan be attributed to the environment and aspects that can beattributed to individual minds, that will be a useful and produc-tive result. Those of us who are developing situativity theory,however, believe that the factoring assumption should not betaken as a general methodological and theoretical commitment.Research in ecological psychology has focused mainly on re-lations of agents with physical systems and environments. Inother research, processes of communication and reasoning arealso being approached in ways that are inconsistent with fac-toring assumptions that have typically been made. Clark andSchaefer (1989), Schegloff (1972), and others have analyzedconversations as interactive collaborations in which contribu-tions, such as references to a place, are considered as successfuljoint actions rather than as events that occur when one personuses a referring term. Many studies of socially organized prob-lem solving and reasoning in complex environments have beenconducted, including a study by Suchman and her associates(Brun-Cottan et al., 1991) of the ground operations of an air-line, and studies by Hutchins (1991, in press) of processes ofcalculatinga military ship's position as it enters a harbor and ofremembering to adjust the wing flaps of a commercial aircraftas its speed decreases during a landing. Symbolic representa-tions of information in these situations are very tangible, andthe theoretical analyses that turn out to be productive are at thelevel of functions that are accomplished by groups of peopleinteracting with each other rather than of hypothetical mentalrepresentations constructed by and operated on by individuals.The view of problem solving that assumes a process of searchin a symbolic problem space consisting of representations ofan initial state, a goal, and a set of operators combined witha problem solver's domain-specific knowledgeand strategy forplanning, has been challenged in studies of reasoning and deci-sion making in activities of trying to get malfunctioning photo-copying machines to work (Suchman, 1987), of grocery shop-ping and food preparation by American adults (Lave, 1988); ofselling produce, candy, and other commodities by young Bra-zilian street merchants (Carraher, Schliemann, & Carraher,1988; Saxe, 1991); of making inferences about quantitativeproperties of a physical system that behaves according to linearfunctions by middle-school and high-school students (Greeno,Moore, & Mather, 1993); and of solvingor constructing expla-nations of algebra word problems (Hall, 1990). These studieshave taken an interactivist view of reasoning, considering it as


3/7

338 JAMES G. GREENOan interaction of the problem solver with material systems thatinclude meaningful symbols and considering the interpretationof the symbols' meanings as an important process to be un-derstood. An aspect of this approach was anticipated by J, J.Gibson and E. J. Gibson (1955b), whoargued, against Postman(1955), that study of learning to perceive symbols should focuson processes of differentiation. "Symbols, like natural objects,must be differentiated or identified in order to be carriers ofmeaning. They come in sets, not singly. And it is quite possiblethat the meaning of a symbol, in the mathematico-logical sense,is given by its univocality within the set" (Gibson & Gibson,1955b, pp. 449-450).

A proposal by Neisser (1992) is particularly relevant to theperception of symbols. Neisser has argued that one needs todistinguish two kinds of perceptual processes, which he callsdirect perception and recognition. Direct perception, whichprovides information for orientation and locomotion in space,occurs in dynamic interaction with the environment. Recogni-tion, which provides information for identifying and classifyingobjects and events, is more effective when the observer can ac-cumulate information about the features of an object or ar-rangement.

Ifwe choose not to factor behavior into the process categoriesof perception, memory, movement, reasoning, decision making,and so on, one then needs theoretical terms for referring to as-pects of the phenomena and systems at the level of agent-situa-tion interactions. Gibson's concept of affordance is a key pro-posal. The idea is quite straightforward. In any interaction in-volving an agent with some other system, conditions that enablethat interaction include some properties of the agent along withsome properties of the other system. Consistent with his empha-sis on understanding how the environment supports cognitiveactivity, Gibson focused on contributions of the physical sys-tem. The term affordance refers to whatever it is about the en-vironment that contributes to the kind of interaction that oc-curs. One also needs a term that refers to whatever it is aboutthe agent that contributes to the kind of interaction that occurs.I prefer the term ability, although Shaw et al. (1982) preferredto coin the term effectivity for that concept. I believe my use ofthe term ability is also synonymous with Snow's (1992) use ofthe term a pt i tud e .Affordances and abilities (or effectivities or aptitudes) are, inthis view, inherently relational. An affordance relates attributesof something in the environment to an interactive activity by anagent who has some ability, and an ability relates attributes ofan agent to an interactive activity with something in the envi-ronment that has some affordance. The relativity of affordancesand abilities is fundamental. Neither an affordance nor an abil-ity is specifiable in the absence of specifying the other. It doesnot go far enough to say that an ability depends on the contextof environmental characteristics, or that an affordance dependson the context of an agent's characteristics. The concepts arecodefining, and neither of them is coherent, absent the other,any more than the physical concept of motion or frame of ref-erence makes sense without both of them.

As Gibson's idea of affordances has been developed in re-search, it seems most productive when it is treated as a gradedproperty rather than as a property that is or is not present. Abeautifully simple example by Warren and Whang (1987) in-

volves the affordance of an aperture for a person to walk fromone side of a partition to the other. The affordance provided byan aperture is a function of its width, and the ability of a personto move through the aperture depends on the person's width.Warren and Whang's analysis of this affordance focused on aninteresting feature of behavior in the range of aperture- and per-son-widths in which the affordance varies. They measuredwhether participants walked straight through apertures, whichvaried in width, or turned their shoulders. The frequency ofshoulder-turning decreased as the width of the aperture in-creased. There were two groups of participants: one group ofgreater-than-average size and the other group of smaller-than-average size. The psychometric functions of shoulder-turningfrequency versus aperture size differed between the two groupswhen aperture size was plotted in ordinary physical units.When aperture size was plotted as a ratio of the physical widthof the aperture to the width of the participant's shoulders, thetwo psychophysical functions were nearly identical.

Affordances and Abilities in Situation TheoryA significant shift in theoretical perspective can be aided by

and sometimes requiresuse of a different formal system. Tur-vey (1992) and his associates have used the theoretical formal-isms of dynamical systems theory productively in analyzing sys-tems such as juggling and the gaits of animals.

Another system that provides theoretical and formal supportfo r an interactivist psychology is situation theory (Barwise &Perry, 1983; Devlin, 1991 ) , which includes a significant refor-mulation of logic. One example of the kind of shift that situa-tion theory proposes is its characterization of the meaning ofsymbolic expressions, such as sentences. In standard logic andlinguistics, the meaning of a sentence is a relation between thesentence and conditions in the world that the sentence asserts,generally focusing on conditions in which the sentence is true.In situation theory, the meaning of a sentence is a relation be-tween situations. The meaning of a spoken sentence is a relationbetween the situation in which the sentence is uttered and thesituation about which uttering the sentence conveys informa-tion. Analysis of a sentence's meaning includes specifying theepistemic connection that the speaker has with the situationthat the sentence refers to. Relativizing the meanings of sen-tences this way has very salutary effects, including resolution ofthe classical philosophical puzzles about reference knownas theliar paradox (Barwise & Etchemendy, 1987).

In situation theory, the concept o f constraint plays a key role.A constraint is a regularity involving situation types. A situa-tion type is a class of situations with objects that have a specifiedproperty of relation. For example, ({reading, a, b; 1)) desig-nates a type of situation where there is something (indicated bya) that is reading something else (indicated by b).For example,you are in a situation that is of this type, where a's anchor isyou and b's anchor is this article. A constraint is a dependencyrelation between situation types. Forexample,((reading, a, b;1& ((printed, b;1= > ((seeing, a, b;1,isa constraint that says that if one thing is reading another thing,and the thing being read isprinted, then the thing that is readingit is also seeing it.


4/7

SPECIAL ISSUE: GIBSON'S AFFORDANCES 339Situation theory provides a way of thinking about knowingho w to do things, in its concept of attunement to constraints ,

Barwise (1989) attributed the idea of attunement to J. J. Gib-son, and Gibson (1966/1982) attributed it to Lashley (1951).Several analyses (e.g., Barwise & Perry, 1983; Israel & Perry,1 9 9 1 ) have considered attunement to constraints as a basis formaking inferences. For example, attunement to the constraintthat smoke means fire (i.e., a situation of the type that hassmoke is also a situation of the type that has fire) will supportan inference by someone who perceives smoke that there is afire there.Attunement to constraints also can play an important role inanalyses of skilled activity. For example, in steering a car, thedriver is attuned to subtle and complex constraints that relatechanges in the direction of the car's forward motion withamounts of turning the steering wheel. The driver's actions ofexerting force on the wheel have effects of making the wheelturn more or less depending on the force that is applied, and ofmaking the direction of the car change more or less dependingon the amount the wheel is turned.

Many constraints only hold conditionally, and c o n d i t i o n a lconstraints are often used in analyses. For example, the con-straints that relate turning a steering wheel to changing the di-rection ofa car'smotion are conditional on, among other things,the car's steering mechanism being intact.The idea ofconditional constraints provides a way to charac-terize affordances and abilities quite clearly. A skilled practi-tioner's actions have effects that are functional in the activity. Inother words, there are constraints of the following form:

((action by agent}) =* ((good effects in situation))where "good effects" are outcomes that are needed or desirablefo r a broader activity to be successful.Affordances and abilitiescan be thought of as conditions in which the constraints of suc-cessful performance hold.As a simple example, consider moving from a hallway into aroom in a building.Anaction that accomplishes that is walkinginto the room, which has the desired effect that the person is inthe room because of the action. The relevant constraint is asfollows:

((walk into the room}) =* ((be in the room)).Affordance conditions for this constraint include the presenceof a doorway that is wide enough to walk through as well as apath along a supporting surface. If there is no doorway,or if thefloor is too weak to support the person's weight at some pointalong the path, or if there is a large obstacle somewhere alongthe pathway, the constraint does not apply. Ability conditionsfor the constraint to include the ability to walk along the path,including the perceptual ability to see the doorway and the co-ordination of vision with motor activity needed to move towardand through thedoorway.In the more complicated situation of driving, action-out-come constraints for the driver include movingher or his armssome amount while holding the wheel, with the outcome thatthe car's direction changes by some amount:

((movement of arms)) = > ((car changesdirection)).

Affordance conditions for this constraint include the shape andother mechanical features of the steering wheel, which is de-signed to afford rotary movement,and the mechanical arrange-ment of the steering system that translates rotary motion of thesteering wheel into turning the wheelsof the car. Ability condi-tions for this constraint include the driver's ability to grasp thewheel and to move her or his arms in the plane of the steeringwheel. If we were to focus an analysis on a more general func-tional action, such as having the car stay in a traffic lane as theroad turns, ability conditions would include perceptual abilitiesto see the lane markers or the edge of the road in relation to thecar's movement in the space and perceptual-motor coordina-tion of that dynamic visual information with the action of turn-ing the steering wheel.

Viewed in this general way, the concepts of affordance andability can be used in any analysisof activity, and some interest-ing implications are suggested. For example, participants inconversations utter words and phrases that they expect will re-sult in reference to objects, places, properties, events, and so on.People who share a linguistic practice are attuned to a greatmany constraints that include conventions of reference. Suchconstraints have the following form:

((use of a term}} =* ((refer to something}}.Constraints of this kind are obviously conditional. They onlyhold when participants in the conversation are attuned to ashared set of constraints. Many constraints also depend on theconversational setting.Affordance conditions forreference constraints include prop-erties of the language that the participants use. For example,just as a car's steering wheel, shaped so the driver can grasp andturn it, affords changing the car's direction, a term in a lan-guage, composed of phonemes that a speaker can pronounce,affords referring to an object or property that the term is usedto designate. Abilities to speak and perceive the terms of thelanguage are among the ability conditions required for con-straints of reference to apply in conversational activity.Affordances for reference constraints also are present in situ-ations where conversations takeplace.Most obviously, whentheobjects wewant to refer to are in the situation, their presence isan affordance for ostensive reference, fo r example, by pointingand saying, "That one." In Clark and Wilkes-Gibbs's (1986)study of collaborative reference, two participants were givenidentical sets of cards with tangram pictures. One of the partic-ipants had the cards arranged in a sequence, and the task of thepair was to communicate so the other participant's cards werein the same sequence. The task was made challenging by pre-venting the participants from seeing each others' cards becausea screen blocked their views.

The participants could accomplish the task because theirshared language had terms to designate properties of shapes andlocations, such as "triangle next to the square," and familiarforms that could be recognized as similar to some of the tan-gram forms. Over a fewtrials, the pairs developed conventionsthat enabled them to refer to the various cards more economi-cally. In the terms of my theoretical proposal here, they devel-oped elements of a communicative practice, including con-straints of reference that had not been available to them ini-tially.


5/7

340 JAM ES G. GREENOAnalyses of reasoning also can be framed in these terms.

Along with several students and other colleagues, I have beenworking toward a situativity-theoretic analysis of reasoningabout quantitative properties of a physical system that we de-signed for research purposes a few years ago. The system, whichwe call "winches," has two tracks in which small metal blocksare pulled by strings that wind around spools when a handle isturned. Distances along the tracks are shown by 1-in. unit rul-ers, and different sizes of spools are varied to provide differentdistances per turn of the handle. Wehaveset up situations withthe blocks in some starting positions, connected to spools ofsome sizes, andasked questions.Forexample, the red blockmaybe at 9, connected to a 3-in. (circumference) spool, with theblue block at 0, attached to a 6-in, spool. An example of thequestions we asked students is, "Will the blue block ever beahead of the red block? If so, after how many turns?" Studentparticipants were quite successful in answering such questions,and they used an interesting variety of methods. Some answerswere obtained in ways that involved direct interaction with thephysical device, by pointing to successive positions along therulers where the blocks would be after different numbers ofturns. Some answers used another physical medium, paper anda pencil, to record symbols for the positions the blocks wouldbe in after successive turns. Some answers were obtained usingarithmetic relations and operations, either spoken aloud as astudent worked them out or reported retrospectively after ananswer was given. To analyze these processes, wehavedevelopedhypotheses about ways in which inferences are made about bothordinal and numerical properties of the blocks' motions, inwhich the physical presence of the winches provide significantaffordances, as do physical symbols for numbers that are on therulers or are constructed by the participants bywriting (Greenoet al., 1993). In answeringthe questions, students provided evi-dence that they were attuned to constraints of the winches' op-eration, such as the constant distance that each block moved oneach turn, and the correspondence between the physical dis-tances moved in a sequence of turns and arithmetic relations ofadding or multiplying numbers.Affordances for the activities of reasoning were provided bythe physical presence of the winches, so students could point topositions along the track and, we hypothesize, mentally simu-late events of moving the blocks. Affordances were provided aswell by the physical resources for writing symbolic representa-tions in the form of tables. Affordances also were provided bythe conceptual entities and operations of arithmetic, which wereavailable to the student by virtue of their participation in thesocial practices of school and other settings where numericalsymbols and inferences occur.In a subsequent study, Moore (1993) observed performanceon these reasoning tasks in different situations: one with thephysical winches present, one with a computer-graphic simula-tion of the winches along with numerical symbols of the posi-tions of the blocks after each simulated turn, and one in whichtables of numerals are constructed without reference to a sys-tem to provide a physical (or simulated physical) semantic in-terpretation. Students' performance provided evidence that thedifferences among the reasoning environments affected theirreasoning activity. Apparently the physical or simulatedwinches afforded reasoning, to a greater degree, about ordinal

dependencies between variables (e.g., if the spool is larger, theblock moves farther on each turn) and relations between theblocks' positions (such as how much one block gains on theother).

As I understand it, situativity theory does not imply that hu-man activity does not use symbols, and its domain is not limitedto processes in which symbols play no role. Itdoes treat symbolsas a special kind of resource for cognition, however, and at-tempts to analyze the ways in which symbols function as com-ponents of the situations that people are in when they reasonand communicate. Hall's (1990) study of algebra students andteachers solving word problems provided particularly instruc-tive examples of ways that problem solvers construct symbolicrepresentations that facilitate their reasoning, locating symbolsthat relate to each other in convenient spatial arrangements.

Perception of AffordancesDiscussions of Gibson's concept of affordance have been

plagued by confusion about where to locate the reference of theterm. For example, is the affordance that a chair provides forsitting a property of the chair, a property of the person who sitson it or perceives that he or she could sit on it, or somethingelse? It seems clear to me that Gibson's intention was that theaffordance is a property of whatever the person interacts with,but to be in the category of properties we call affordances, it hasto be a property that interacts with a property of an agent insuch a waythat an activity can be supported.Affordances are, in this view, preconditions for activity, as Ibelieve is made clearer when they are treated as conditions forconstraints. The presence in a situation of a system that pro-vides an affordance for some activity does not imply that theactivity will occur, although it contributes to the possibility ofthat activity. Additional conditions include aspects of the activ-ity of an agent in the situation, having to do with motivation andperception. Motivation to engage in some action is related towhat the agent is doing at a more general level. For example, ifa person is engaged in an activityof going to attend a class, thenthe action of moving into the classroom is a functional part ofthat activity. That will make the person attentive to aspects ofthe environment that could provide an affordance for movinginto the classroom, such as the doorway from the hall into theclassroom.

Consistently with his view of perception as a process of pick-ing up information directly, Gibson characterized affordancesas properties of objects and layouts that are specified by infor-mation in the array of energy that is available to an agent's per-ceptual systems. This characterization works well for manyaffordances, such as the affordance of a doorway for moving tothe other side of a partition. The question ofhow the affordanceis perceived is a straightforward research problem, and experi-ments conducted by Warren and his associates provide elegantexamples of analyses that identify the information that humanperceptual systems pick up to detect affordances. In the case ofperceiving the affordance of a doorway, Warren and Whang's(1987) analysis indicated that the effective stimulus is the visualinformation that specifies the physical width of the doorway,scaled to the eyeheight of the observer.In addition to affordances for orientation and locomotion in


6/7

SPECIAL ISSUE: GIBSON'S AFFORDANCES 341space, it is likely that many affordances for individual actions inconversations and other interpersonal interactions are perceiveddirectly, that is, without need for mediating symbolic represen-tations. In conversations, pauses, facial expressions, and othergestures provide information that influences the opportunitiesfor each participant's actions. Analyses of the auditory and vi-sual energy characteristics that specify these affordances havenot been done, to my knowledge.Even so, there is strong evi-dence that mediating symbolic representationsare not requiredfor their perception because infantsare responsive to importantfeatures of these social stimuli well before they have developedthe capability of forming symbolic representations that denotethem (e.g., Tomasello,Kruger, & Ratner, in press).

In the broad meaning that I am proposing for the concept ofaffordance, it would be inappropriate to consider direct percep-tion of affordances as a denning characteristic. J. J. Gibson(1979/1986)used mailboxes asan example. He proposed that amailbox provides an affordance forposting letters, and it surelydoes that. It seems likely that formost people, in most if not allcircumstances, the process of cognizing that affordance in-cludes classifying the physical object as a mailbox. The infor-mation required for that classification has to be visually avail-able, of course, but the process of classification includes, Ishould think, a mental state that has the epistemic status of asymbol that designates the property of being a mailbox. A the-oretical possibility that seems reasonable is that mental symbolsof that kind are products of the perceptual processes that Neis-ser (1992) regarded as recognition, rather than direct percep-tion.

What of symbols that are recognized and used in processes ofmaking inferences? As J. J. Gibson and E. J. Gibson (1955b)recognized, symbols need to be identified to be used meaning-fully. In Neisser's (1992) distinction, then, perception of sym-bols is a process of recognition, rather than a process of directperception, and if my conjecture about mental symbols is cor-rect, such recognition creates mental symbols. It follows, then,that those mental symbols could be available for processes ofmaking inferences according to operations that transform sym-bolic expressions into other symbolic expressions. In our analy-ses of reasoning (Greeno et al., 1993) we have taken this as anassumption. People represent with mental symbols in differentways, just as one does with spoken or written symbols. Werep-resent with some symbols as propositions, and inferenceshavethe form of generating further representations of propositions.One can represent with other symbolsas simulations, and infer-ences have the form of operating on a model to generate stateswith properties that correspond to the represented system.

A theoretical understanding of mental symbols, includingmental states that function as simulations, could provide a basisfor understanding conceptual entities, such as numbers, arith-metic operations, and functions. Some progress has been madetoward theoretical understanding of the ability to form antici-patory mental models that simulate outcomes of activity in sit-uations (Jordan & Rumelhart, 1992). We need to understandho w that ability includes forming models that satisfy con-straints that weexpress and understand in communication withstatements of propositions. This could provide a theory oflearning that results in shared mental models that support rea-soning about the properties of conceptual entities, with the

properties represented implicitly in mental simulations ratherthan being explicitly stated in propositions.

ConclusionsJ. J. Gibson's theorizing in the 1950s, in his own writing

(Gibson, 1954) and with E. J. Gibson (J. J. Gibson & E. J. Gib-son, 1955a, 1955b), reflected two general commitments. Onecommitment was that important phenomena should not be leftout of the psychologyof perception simply because they weredifficult to study with available experimental methods. Anothercommitment was that concepts in the theory of perceptionshould not be subsumed under the apparently broader psycho-logical concept of stimulus-response association because theview of perception and perceptual learning that results fromthat subsumption is theoretically less adequate than anotherview involving differential attunement.Gibson did not, however, treat the psychology of perceptionas a compartment closed off from the rest of psychology. In hislater theorizing, he worked toward a broader view of behaviorthat would be consistent with a theoretically adequate under-standing of perception. This view focuses on interactive re-lations of behaving agents with systems in their environments.In Gibson's view, perception is a system that picksup informa-tion that supports coordination of the agent's actions with thesystems that the environment provides. This led him to developthe idea of affordances, the characteristics of objects and ar-rangements in the environment that support their contributionsto interactive activity and, therefore, the characteristics of theenvironment that agents need to perceive.

Considerable efforts are being made to developa general the-oretical account of activity considered as interactions of agentswith systems in their environments, and significant progress isoccurring. Some of these efforts, in ecological psychology, haveused Gibson's idea of affordances quite explicitly. Other efforts,focused on situations involving social interactions as well as in-teractions with physical systems, mainly use concepts andmethods of ethnographic social science. We should aspire to atheory that merges these perspectives, along with the insightsand methods of information-processing cognitive science; thesesupport analyses of information structures that are the contentsof human symbolic communication and describe the contentsof other interactions of agents in situations. I have suggested, inthis commentary, some ways in which an extended version ofGibson's concept ofaffordance is involved in one versionof thisgeneral theory of activity. Gibson's theorizing was seminal inthe development of ecological psychology and is likely to beseminal in a more general development of a theory of activityviewed as interactive relations of agents with other agents andphysical systems. Gibson's affordances, then, have already beenof great value and will support further valuable activity in ourfield.

ReferencesBarwise, J. (1989). The situation in logic. Stanford,CA:Center for the

Study of Language and Information.Barwise, J., & Etchemendy, J. (1987). The l iar: An essay on truthandcircularity. London:Oxford University Press.


7/7

342 JAM E S G, GREENOBarwise, J., & Perry, J. (1983). Situations a n d attitudes. Cambridge,M A: MIT Press/Bradford.Beek, P. J., & Turvey, M. T. (199 2). Temporal patterning in cascadejuggling. Journal ofExperimental Psychology: H u m a n Perception andPerformance, 18 , 934-947.Bickhard, M. H., & Richie, D. M. (1983). On the n a ture of representa-tion: A case study o f J a m e s Gibson's theory o f perception. New " Y o r k :

Praeger.Brun-Cottan, E, Forbes, K ., Goodwin, C, Goodwin, M., Jordan, B. ,Suchman, L., & Trigg, R. (1991) . Th e workplace project: Designingfor diversity a n d cha n ge (Videotape Report). Palo Alto, CA : XeroxPalo Alto Research Center.Carraher, T. N., Schliemann, A. D., & Carraher, D. W. (1988). Mathe-matical concepts in everyday life. In G. B. Saxe & M . Gearhart (Eds.),Children's ma thema tics (pp. 71 -87). San Francisco: Jossey Bass.Clark, H. H ., & Schaefer, E. F. (198 9). Co ntributing to discourse. Co g-nitive Scienc e, 13, 259-294.Clark, H. H., & Wilkes-Gibbs, D. (1986). Referring as a collaborativeprocess. Cognition, 22 , 1-39.Cutting, J. E., Springer, K., Braren, P. A., & Johnson, S. H. (1992).Wayfinding on foot from information in retinal, no t optical, flow.Journal of E xpe r ime ntal Psychology: G en e ra l , 121, 41-72.Devlin, K. (1991). Lo gic a n d in formation. Cambridge, England: Cam-bridge Unive rsity Press.Dewey, J. (1896). The reflex arc concept in psychology. PsychologicalReview, 3, 357-370.Fodor, J., & Pylyshyn, Z. (1981). How direct is visual perception? Somereflections on Gibson's ecological approach. Cogn ition, 9, 139-196.Gibson, E. J. (1969). Principles o f perceptual learn ing and development.New York: Appleton-Century-Crofts.Gibson, J. J. (1954). The visual perception of objective motio n and sub-jective movement. Psychological Review, 61 , 304-314.Gibson, J. J. (1966). The senses considered a s perceptual systems. Bos-ton: Houghton Mifflin.Gibson, J. J. (1977). The theory of affordances. In R . Shaw & J. Brans-ford (Eds.), Perceiving, acting, a n d k n o w in g: Toward an ecologicalpsychology ( p p . 67-82). Hillsdale, NJ: Erlbaum.Gibson, J. J. (1982).Th e problem of temporal order in stimulation an dperception. In E. Reed & R. Jones (Eds.), Rea s o n s fo r realism: Se -lected essays of J a m e s J. Gibson (pp. 171-179). Hillsdale, NJ: Erl-bau m. (Original work published 1966)Gibson, J. J. (1986). T he ecological approach to visual perception. Hills-dale, NJ: E rlbaum . (Original work published 1979)Gibson, J. J., & Gibson, E. J. (1955a). Perceptual learning: Differenti-ation or enrichment?Psychological Review, 62 , 32-41.Gibson, J. J., & Gibson, E. J. (1955b). What is learned in perceptuallearning? A reply to Professor Postman. Psychological Review, 62 ,447-450.Goodwin, C., & Goodwin, M. H . (in press). Formulating planes: Seeingas situated activity. In D. Middleton & Y . Engestrom (Eds.), Co gn i-tion a n d co mm un ica t io n a t work. Beverly Hills, CA : Sage.

Greeno, J. G., & Moore, J. L. (19 93 ). Situativity an d symbols: Responseto Vera and Simon. Cognitive Scien ce , 17 , 49-60.Greeno, J. G., Moore, J. L.,& Mather, R. (1993). A situativity-theoreticm o d e l of reasoning a n d problem solving. Manuscript submitted forpublication.Hall, R. P. (1990). M a k in g ma thema tics o n paper: Constructing repre-sentations o f stories a bo ut re la ted l inear unctions [doctoral disserta-tion]. Technical Report 90-17, Department of Information and Com-puter Science, University of California, Irvine. Also available asMonograph 90-0002. Palo Alto, CA : Institute for Research on Learn-i n g .Hutchins, E. (1991) . Organizing work by adaptation. Organization Sci-en ce , 2, 14-39.

Hu tchins, E. (in press). How a cockpit remembers its speeds.CognitiveScience.Israel, D., & Perry, J. (199 1). W h a t is in formation?(Report No. CSLI-91-145) . Stanford, CA : Center for the Study of Language an d Infor-mation.Jordan, M . I., & Rumelhart, D. E. (1992). Forward models: Supervisedlearning with a distal teacher. Cogn itive Scienc e, 16,307-354.Koffka, K . (1935). Principles of gestaltpsychology. New York: Harcourt,Brace.Lashley, K . (1951) . The problem of serial order in psychology. In L . A.Jeffress (Ed.), Cerebral mecha n is ms in behavior. New \brk:Wiley.Lave, J. (1988). Cognition in pra c t ice: M in d , m a thema tics, a n d culturein everyday l i f e . Cambridge, England: Cambridge University Press.Lee, D. N . (19 80). The optic flow field: The foun dation of vision.Philo-sophical Transactions of the Royal Society of L ondon, B 290, 169-

1 7 9 .Michaels, C. R, & Oudejans, R . R. D. (1992). The optics and actions ofcatching fly balls: Z eroing o ut optical acceleration. Ecological Psy-chology, 4, 199-222.Moore, J. L. (1993) . Co mpa ris o n s of a physica l mo d el a n d co mputerrepresentations in re asoning a n d lea rn in g a bo ut l inear unctions. Un-published doctoral d issertation, Stanford U niversity, Stanford , CA.Neisser, U. (199 2, June). Distinct system s fo r "where" and "what": Rec-onciling the ecological and representational views of perception. Paperpresented at the Fourth A nnu al Convention of the A merican Psycho-logical Society, San Diego, CA .Postman, L. (1955). Association theory and perceptual learning. Psy-chological Review, 62, 438-446.Saxe, G. (1991) . Culture a n d cognitive development: Studies in ma the-ma tica l understanding. Hillsdale, NJ: Erlbaum.Schegloff, E. A. (1972). Notes on a conversational practice: Formulatingplace. In D. Sudnow (Ed.), Studies in social interaction (pp. 75-119).New \brk: Free Press.Shaw, R., Turvey, M. T, & Mace, W. (1982). Ecological psychology:The consequences of a commitme nt to realism. In W . Weimer & D.

Palermo (Eds.), Cogn ition and the symbolic proc esses H. Hillsdale,NJ: Erlbaum.Snow, R. E. (1992). Aptitude theory: Yesterday, today, and tomorrow.E duc at ional Psychologist, 27 , 5-32.Suchman, L. (1987). P la n s a n d situated actions: Th e problem o f hu -m a n - m a c h i n e c o m m u n i c a t i o n . Cambridge, England: CambridgeUniversity Press.Tomasello, M ., Kruger, A. C., & Ratner, H. H. (in press). Cultural learn-

i n g . Beha vio ra la n d Bra in Sc ien ces .Turvey, M. T. (1990). Coordination. A m e r i c a n Psychologist, 45 , 938-953.Turvey, M. T. (1992). Ecological foundations of cognition: Inva riants ofperception and action. In H. L. Pick, Jr., P. van den Broek, & D. C.Knill (Eds.), Cognition: Conceptual and me th odolog ic al issues (pp.

85-117) . Washington, DC : American Psychological Association.Ullman, S. (19 80). Against direct perception. Behavioral a n d Bra in Sc i -en ces , 3,373-415.Vera, A. H., & Simon, H . A. (199 3). Situated action: A symbolic inter-pretation. Cognitive Scien ce , 17, 7-48.Warren, W. H., Jr., & Whang, S. (1987). Visual guidance of walkingthrough apertures: Body-scaled information for aifordances. J o u r n a lof E xpe r ime ntal Psychology: H u m a n Perception a n d Performance,13,371-333.

Received August 9,1993Accepted September 23,1993

gibson afford ances

Documents