8.5 An Ecology of Perception and Action

Perceiving is an achievement of the individual, not an appearance in the theater of his consciousness. It is a keeping-in-touch with the world, an experiencing of things, rather than a having of experiences. It involves awareness-of instead of just awareness. It may be awareness of something in the environment or something in the observer or both at once, but there is no content of awareness independent of that of which one is aware. This is close to the act psychology of the 19th century except that perception is not a mental act. Neither is it a bodily act. Perceiving is a psychosomatic act, not of the mind or of the body but of a living observer Q. J. Gibson, 1979, p. 239).

8.5. 1 Integrated Perception and Action

Dominated by information-processing theories, the recent history of perceptual psychology has emphasized research paradigms that attempt to constrain action and to isolate sensation from attention and intention. This predilection for ignoring codeterminant relations between perception and action has resulted in a relatively weak foundation for the design of new media products and a limited basis for understanding many traditional media forms.

Ulric Neisser's (1976) perceptual cycle--which frankly acknowledges the influence of both J. J. Gibson and his spouse, Eleanor Gibson--serves as a simplified framework for examining the relationship between action and perception in mediated environments. Neisser (1976) was concerned with the inability of information-processing models to explain phenomena associated with attention, unit formation, meaning, coherence, veridicality, and perceptual development. Information-processing models of the 1970s typically represented sensory organs as fixed and passive arrays of receptors. How, then, Neisser asked, would such models explain why different people attend to different aspects of the same situation? How would information-processing models help to explain why even infants attend to objects in ways that suggest the brain can easily assign to things stimuli obtained through distinct sensory modalities? How would information-processing models explain the remarkable ability of the brain to respond to scenes as if they were stable and coherent, even though the act of inspecting such scenes exposes the retina to rapidly shifting and wildly juxtaposed cascades of images?

Figure 8-1. Neisser's Perceptual Cycle (modified from Neisser, 1976, p. 21).

In the language of ecological psychologists, an organism selectively samples available information in accord with the demands of its niche. An organism's perceptions are tuned to the means that the environment offers for fulfilling the organism's intentions.

The Neisser-Gibson alternative to the information processing models adds the crucial function of exploration This addition, reflected in Neisser's Perceptual Cycle (Fig. 8-1) reflects the fact that organisms selectively sample available information in accord with the demands of their niches and, further, that organisms' perceptual capabilities are tuned to the means that their accustomed environment offers for fulfilling the organisms' intentions.

Neisser's emphasis on exploratory perception reminds us that schemata can never be entirely complete as representations of realia. Schemata are not, in Neisser's opinion, to be thought of as templates for conceptualizing experience but rather as plans for interacting with situations. "The schema [is] not only the plan but also the executor of the plan. It is a pattern of action as well as a pattern for action" (Neisser, 1991, pp. 20-21).

The idea of the action-perception cycle, which is similar in some respects to early cybernetic models (see 3.1.2.5), can also be fruitfully thought of as a dialectic in which action and perception are codeterminant. In visual tracking, for example, retinal perception is codeterminant with eye movement (see Clancey, 1993, and Churchland, 1986, on tensors as neural models of action-perception dialectics).

Neisser's use of schemata and plans echoes a multiplicity of meanings from Kant (1781, 1966) to Bartlett (1932) to Piaget (1971) to Suchman (1987). His meaning is close to what we will define as actionable mental models. An actionable mental model integrates perception of the environment with evolving plans for action, including provisions for additional sampling of the environment. Actionable mental models draw not so much on memories of how the environment was structured in the past as they draw on memories of how past actions were related to past perceptions. Rather than mirroring the workings of external reality, actionable models help organisms attend to their perceptions of the environment and to formulate intentions.

Our use of actionable mental models assumes first that mental models are rarely self-sufficient (see D. Gentner & Stevens, 1983). That is, mental models cannot function effectively (are not "runnable") without access to data about a situation. Actionable mental models, in other words, must be "situated" (Collins, Brown & Newman, 1989; Greeno, 1994) in order to operate.

Ecological psychology assumes that much if not most of the information required to guide effective action in everyday situations is directly perceivable by individuals adapted to those situations. It seems reasonable to assume that natural selection in favor of cognitive efficiency (Gatlin, 1972; Minsky, 1985; von Foerster, 1986) will work against the development and maintenance of complex MIROS if simple MIROS will contribute to survival equally well. That is, the evolution of cognitive capacities will not favor unnecessary repleteness in mental models or the neurological structures that support them even when such models might be more truthful or veridical according to some "objective" standard of representation.

In many cases, MIROS cannot serve (or do not serve efficiently) as equivalents for direct perception of situations in which the environment does the "work" of "manipulating itself' in response to the actions of the perceiver. It is usually much easier, for instance, to observe how surroundings change in response to one's movement than it is to construct or use MIROS to predict such changes.

Even when humans might employ more complete MIROS, it appears they are often willing to expend energy manipulating things physically to avoid the effort of manipulating such things internally. Lave (1988) is on point in her discussion of a homemaker responsible for implementing a systematic dieting regime. After considering the effort involved in fairly complex calculations for using fractional measures to compute serving size, the homemaker, who had some background in higher mathematics, simply formed patties of cottage cheese and manipulated them physically to yield correct and edible solutions.

There are trade-offs between elaborate and simple MIROS. Impoverished environments are likely to select against improvement of elaborate sensory and perceptual faculties and may even favor degradation of some of these faculties: We can assume that the blindness of today's cave fish evolved because eyes contributed little to the survival of their sighted ancestors. It seems reasonable to assume that, in the long run, the calculus of natural selection balances resources invested in perception against resources invested in other means of representing the environment.

In any case, for reasons of parsimony in scientific explanation (in the tradition of Occam's razor), descriptions of MIROS--which are of necessity usually hypothetical-- should not be any more complex than is necessary to explain observed facts. Accounting for observed behavior, then, with the simplest possible MIROS will assume that organisms attend to the environment directly because this is often more economical and more reliable than maintaining models of the environment or "reasoning" about it.

8.5.1.1. Perception. Gibson's seminal works (1966 and 1979, for example) established many of the theories, principles, concepts, and methods employed by contemporary ecological psychologists. Developed over a 35-year span of research on the problems of visuospatial perception, his "ecological optics" now serves as a framework for extending the ecological approach to other areas of psychology. The implications of Gibson's research extend beyond the purely theoretical: He was instrumental in producing the first cinematic simulation of flying using model airplanes and model landing fields. Gibson's novel conception of the retinal image(4)

substituted dynamic, flowing imagery of the mobile observer for the static, picturelike image of classical optics and inspired techniques of ground plane simulation and texture gradients that are the basis for many electronic games.

8.5.1.2. Invariants. In developing his radical ecological optics, Gibson (1979) focused on the practical successes of an organism's everyday behavior as it lives in and adapts to its environment. He was particularly concerned with characteristics and properties of the environment that supported such success.

Generalizing this interest, ecological psychologists investigate "the information transactions between living systems and their environments, especially as they pertain to perceiving situations of significance to planning and execution of purposes activated in an environment" (Shaw, Mace & Turvey, 1986, p. iii). Ecological psychologists focus on ordinary everyday perceiving as a product of active and immediate engagement with the environment.

An organism selectively "picks up" information in its habitat when such information is related to its ecological niche. In this context, it is useful to think of habitat as roughly equivalent to address, and niche as roughly equivalent to occupation. The perceptual capabilities of organisms are tuned to opportunities for action required to obtain enough energy and nutrients to reproduce.

"Attunement to constraints" (attributed to Lashley, 195 1, by Gibson, 1966) thus reflects the most fundamental type of information that an organism can obtain about its environment. With this in mind, ecologists such as von Foerster (1986) contend that "one of the most important strategies for efficient adjustment to an environment is the detection of invariance or unchanging aspects of that environment" (p. 82). The detection of invariances--constrained and predictable relations in the environment--simplifies perception and action for any organism. As we shall argue, detection of invariances is also critical to successful adaptation by humans to any mediated environment.

8.5.1.3. A Simple Experiment. As an example of the importance of detecting invariants, consider the human visual system as it is often presented in simple models. Millions of rods and cones in the retina serve as a receptor array that transmits nerve impulses along bundled axons to an extensive array of neurons in the primary visual cortex called V1. Neurons in VI are spatiotopically mapped, i.e., laid out in fields that preserve the integrity of the information captured by the retina. Neurons in VI transmit to specialized centers that process color, form, and motion. Yet there is much more to seeing than the processing of retinal imagery. Seeing also integrates complex systems that focus lenses, dilate irises, control vergence and saccades, and enable rotation of the head and craning of the neck.

Perception by the visual system of invariants in the environment can be thrown into complete confusion by interfering with the brain's detection of head and eye movement. You may want to try this simple experiment: Close your left eye and cock your head repeatedly to the left 2 or 3 inches. Proprioceptors in your neck muscles allow the brain to assign this jerkiness to movements of your head rather than to changes in the environment. Without this natural ability to assign movement of retinal images to selfinduced changes in head position, simply turning to watch an attractive person would "set one's world spinning."

Now close your left eye again and, keeping the right eye open, gently press on the left eyeball several times from the side. Under these abnormal conditions, your visual system now assigns roughly the same amount of eyeball jerkiness to radical movement of the environment itself.

Under normal circumstances, the brain does not attribute variation in retinal images resulting from head or eye movement to changes in the environment. Rather, an elaborate system of proprioceptive and locomotor sensors operates automatically in concert with retinal data to generate a framework of perceptual invariants against which true environmental change can be detected.

It is important to note that the concept of perceptual invariance does not necessarily imply a lack of change in the environment, but rather that the organism is able to detect reliable patterns in the change and therefore able to use the patterns as a background for less predictable

variation. Tide pool animals, for instance, are superb at detecting underlying patterns in the apparent chaos of the surf and adjusting their activity patterns to these fluctuations.

8.5.1.4. Perception of Invariants: Some Implications for Media Design. The idea of a framework of invariants is very useful in the design, management, and utilization of media environments because it reminds us that

. . . it is not necessarily better to throw more graphics techniques, more rendering power, or more artificial intelligence into the scene. These are not necessarily going to give us a better artificial world. The quantity of what goes into your artificial world is not what makes it better or more interesting; it is the quality of what goes into it. What defines quality for this activity are these environmental invariants, -which drive the human perceptual system. We can think of it as though the levels of detail had a weighting system attached, telling us how important that type of detail is or that level of detail. If these details are not associated with perceptual invariants, then the weighting factor is small. Often you can remove certain kinds of detail from an image and people cannot perceive the difference. On the other hand, if the detail has a tight link to some kind of perceptual invariant, then the weighting factor is high. .(Gardner, 1987, pp. 106-07).

While Gibson's work in the 1970s met with skepticism from his contemporary psychologists, he generated even in his day a considerable following among human-factors engineers and ergonornicists. He is now read widely by virtual-world and interface designers. The central concern for these designers is how to engineer the relationship between perceptual variants and perceptual invariants so as to optimize the user's ability to perceive and act in complex, information-rich environments.

Gibsonian psychology points to perceptual invariants that enrich our depth and distance perception. Many of the invariants relate to the ever-existent, textured, ground surface of our environment. . . . The strongest invariants are the ratios, gradients, calibration references, and optical flows tied to motion parallax, surface texture, the ground plane, and ego perception. By enabling the same perceptual invariants that people use to navigate the real world, the creator can construct a world that encourages exploration (Gardner, 1987, pp. 107-09).

8.5.2 Perceptual Learning

Gibson did not believe that sensory inputs are "filtered" or processed by propositional or symbolic schemes. Rather, he strongly favored a bottom-up paradigm in which exploratory actions rather than propositions drive processes of selective perception. Yet none of Gibson's ideas preclude learning to perceive directly, as when children come to understand that they must automatically respond to icy-slick sidewalks with flat-footed caution. Nor did Gibson deny the importance of reasoning about perceptions, as when a mountaineer carefully analyzes the complex textures of an ice-covered cliff in order to plan an ascent. Nevertheless, consistent with his view that action, not conception, drives perception, Gibson

believed that learning entails the tuning of attention and perception, not the conforming of percepts to concepts. Such perceptual learning is, in the words of Gibson's spouse, Eleanor, essentially

... an increase in the ability of an organism to get information from its environment, as a result of practice with the array of stimulation provided by the environment. This definition implies that there are potential variables of stimuli which are not differentiated within the mass of impinging stimulation, but which may be, given the proper conditions of exposure and practice. As they are differentiated, the resulting perceptions become more specific with respect to stimulation, that is, in greater correspondence with it. There is a change in what the organism can respond to. The change is not acquisition or substitution of a new response to stimulation previously responded to in some other way, but is rather responding in any discriminating way to a variable of stimulation not responded to previously. The criterion of perceptual learning is thus an increase in specificity.What is learned can be described as detection of properties, patterns, and distinctive features (E. J. Gibson, 1969, p. 77).

8.5.2.1. Propositional vs. Nonpropositional Learning. Gibson's (1979) research on visual perception in everyday situations rather than laboratory situations led him to think of perceiving as a process in which organisms acquire information directly, without the mediation of propositional reasoning. Hochberg (1974) thinks that one of Gibson's most important ideas is that

. . . there exist higher-order variables of stimulation to which the properties of the objects and events that we perceive are the direct and immediate response ... [thus] the properties of the perceived world ... are not the end products of associative processes in which kinesthetic and other imagery come to enrich two-dimensional and meaningless visual sensations with tri-dimensional depth and object meaning (Hochberg, p. 17).

Gibson sometimes used the term associative thought in ways that implied propositional reasoning. We have therefore substituted the latter in this chapter when we discuss his ideas in order to avoid confusion with current usage of the term associative, which is broadly inclusive of a variety of neurological processes. In any case, a brief review of the controversy regarding propositional and nonpropositional reasoning seems in order here (for more, see Vera & Simon, 1993, and Clancy's 1993 reply).

Cognitive psychologists and computer scientists have long used symbols and propositions to model human thought processes. Anderson's widely influential ACT* model (1983) is typical of rigorous efforts in the 1980s to use propositional logic to model learning. The ACT* model converts declarative knowledge--that is, knowledge that can be stated or described--into production rules through a process of proceduralization. The resulting procedural knowledge (roughly, skills) is highly automatic and not easily verbalized by learners.

Gordon (1994) offers this simplified example of how Anderson's (1983) notion of proceduralization might be used to model the way an agent learns to classify an object (p. 139; content in brackets added):

IF the figure has four sides

and sides are equal

and sides are touching on both ends

and four inner angles are 90'

and figure is black

THEN classify as [black] square.

Such instructions might have some value as a script for teaching students about logic, or perhaps even as a crude strategy for teaching them to recognize squares. Yet even the most sophisticated computer models fail almost entirely when they attempt to use this kind of reasoning to recognize pattern and contexts that are very easy for animals and humans.

There are other reasons to doubt assertions that the brain represents perceptual skills as propositions or production rules. While declarative knowledge (language and propositions) is obviously useful for teaching perceptual skills, the ultimate mechanisms of internal representation need not be propositional. The observation that propositions help people learn to recognize patterns could be explained, for example, by a model in which propositional frameworks are maintained by the brain merely as temporary scaffolding ("private speech"; see Berk, 1994) that supports repeated rehearsal required for perceptual development. Once the perceptual skills have been automated, the brain gradually abandons the propositional representations and their arguable encumbrance of processing speed. It then becomes difficult for learners to verbalize "how" they perceive.

Having decided that perceptual learning is not directly dependent on internalized propositions or production rules, many cognitive scientists have turned to models of nonsymbolic representation. We suspect that Gibson would have found in these emerging models considerable support for many of his ideas about indirect perception.

Kosslyn and Koeing (1992), for instance, offers an excellent treatment of the ways in which connectionist models can explain the de* tails of perceptual processing. Connectionist models (see A. Clark, 1989) employ networks of processing units that learn at a subsymbolic level. These networks (also called neural networks) can be trained, without using formal rules or propositions, to produce required outputs from given inputs, because the processing units mathematically adjust the weighting of connections through repeated trials. Neural nets are superior to proposition-based programs at learning tasks such as face recognition.

A trained subsymbolic network cannot be analyzed or dissected to yield classical rules or symbols, because the learned information is represented as weighted connections rather than as propositions. The learned information is not stored as symbols or bits of code located at specific sites. Rather, it is represented by the overall fabric of connections. Subsymbolic processing networks can, however, serve as substrates for conventional symbolic processing and therefore have some potential for modeling forms of human thought that do rely on symbols and language.

8.5.2.2. Affordances. In Gibson's view, sensory information alone is insufficient for guiding and controlling the activities of living organisms:

The variables of sensory discrimination are radically different from the variables of perceptual discrimination. The former are said to be dimensions like quality, intensity, extensity, and duration, dimensions of hue, brightness, and saturation, of pitch, loudness, and timbre, of pressure, warm, cold, and pain. The latter are dimensions of the environment, the variable of events and those of surfaces, places, objects, of other animals, and even of symbols. Perception involves meaning; sensation does not . . . (J. J. Gibson, 1974/1982, p. 35 1).

Selective perception generates much more information about an experienced event than can be obtained by sensation alone because the organism is informed during selection by traces of its activities relating to location, orientation, and other conditions. In all but extreme laboratory settings, organisms employ the natural means available to them for locomotion in and manipulation of their environment both to obtain additional information and to act on that information. For Gibson (1979), perception and action were inextricably coupled in a seamless cycle. To describe this coupling, he introduced the concepts of affordances (opportunities for action) and effectivities (capabilities for action). Affordances are functional, meaningful, and persistent properties of the environment Q. J. Gibson, 1979), "nested sets of possibilities" (Turvey& Shaw, 1979, p. 261) for activity towards which the organism is oriented by its perceptual history and heritage. In active perceiving, "the affordances of things is what gets attended to, not the modalities, qualities, or intensities of the accompanying sensations. . . " (J. J. Gibson, 1977/1982, p. 289).

If ecological information specifies the affordances of things . . . then it does not specify abstract physical properties of the classical sort (for example, the three Cartesian dimensions), but rather ecologically relevant properties such as texture, resistance to deformation, and manipulability. Both kinds of properties may be real, but it is the functional properties, the affordances, that we animals are aware of directly ... (Reed, 1988, p. 232).

Thus, an affordance is, roughly speaking, a pathway for action that enhances the survivability of an organism in its niche--an opportunity for action

... such as support by a firm surface, grasping by a limb of a tree, or mating by an animal of the opposite sex. Gibson claimed that affordances such as these are specified by the structure of light reflected from objects, and are directly detectable. There is therefore no need to invoke representations of the environment intervening between detection of affordances and action; one automatically leads to the other (Bruce & Green, 1990, p. 382).

Affordances simultaneously enable some possibilities and constrain others, and they make actions more predictable and replicable.

Speaking more humanely, we do not in any sense reduce the statistical variety of nature when we are engaged in learning the fixed patterns, the constancies of nature. Rather, it is nature that reduces the statistical variety in us.Our subsequent behavior becomes more predictable to an outside observer who knows the order of nature because it comes to be more closely coupled to, and defined by, that order. We have acquired, in Spinoza's phrase, more "aptness of the body" because our ideas are less "mutilated and confused" (Hawkins, 1964, p. 23.9).

Mediated habitats encompass a range of affordances and effectivities, related to cognitive artifacts such as the book, the calculator, and the television. Such artifacts can do some of the work of storing and transforming information, and this Work may therefore lessen the user's need to construct or maintain more complex MIROS. In addition, such artifacts can provide

... affordances for reasoning.... [which] are properties of representation in relation to a person's or group's abilities to use the representations to make inferences. Reasoning is an activity that transforms a representation, and the representation affords that transformational activity. Abilities for reasoning activities include knowing the operations to perform on the notational objects in the representation and understanding the semantic significance of the objects and operations (Greeno, Moore & Smith, 1993, p. 109).

In the Gibsonian (1979) paradigm, affordances are opportunities for action rather than physical artifacts or objects. Nevertheless, it is useful to think of sets or suites of affordances as bundled in association with tools or devices (see 24.2). The affordance of "browse-ability" is itself composed of clusters of affordances; one exploits the turnability of a book's pages in order to exploit the readability of their text. We

can characterize the phone by its "handle-ability," "dial-ability," "answer-ability .. .. listening-to-ability," and "talking-into ability," affordances that in some cases serve multiple goals or ends. The complete action pathway for realizing the opportunity afforded by the telephone for talking to someone at a distance must be perceived, though not necessarily all at once, and "unpacked" through the effectivities of a human agent. Interface designers refer to this unpacking as entrainment.

One of the reasons Gibson argued that direct perception is independent of reasoning is because, by definition, the properties of an affordance are persistent, even invariant. They are the knowns of the problem: the "climb-ability" of a branch for the squirrel, the "alight-ability" of a rock for the seagull, the "grab-ability" of a deer for the wolf. Such affordances are perceived automatically as the result of repeated engagement with consistent circumstances, "hard

wired" in the form of durable connections between dendrites (see Crutcher, 1986; Kupferman, 1991).

It may seem peculiar or contrived to use climb-ability as an alternative to the familiar forms of the verb to climb. The grammar of most human languages is, after all, centered on action in the form agent-action-object or agent-object-action. Organizing propositions in terms of action, however, is a serious limitation if one wants to describe mediated environments as complex fields of potentialities. The language of affordances and effectivities refocuses attention on how the environment structures activity rather than on descriptions of activity per se.

In the calculus of planning and action, detection of the invariant properties of affordances allows some aspects of a problem to be stipulated or assumed, freeing cognitive resources to attend to the unknowns, those aspects of the environment subject to change: Is this branch thick enough? Are the waves too frequent? Is the buck too big?

The capacity to detect and respond to affordances results from repeated engagement with sets of circumstances that over time-in the life of the individual or the species-are consistent enough to induce automaticity (Sternberg, 1977) in perception and action. Affordances influence the interaction of the organism with its environment, not only by enabling and constraining action but also by entraining the organism's perceiving and acting in predictable and repeatable sequences.

As a general rule, it can be assumed that organisms will not squander sensory or cognitive resources on aspects of the environment that have no value as affordances, because natural selection (or learning) will have effectively blinded them to objects and phenomena they cannot exploit. "We see the world not as it is but as we are," in the words of the Jewish epigram. To paraphrase this from a Gibsonian perspective, we see the world not as it is, but as we can use it.

8.5.2.3. Effectivities. Effectivities (or capabilities; Greeno, Smith & Moore, 1993) are intentional, meaningful properties of a perceiving organism that trigger, guide, and control ongoing activities directed towards exploitation of the inherent possibilities of affordances (Turvey, Shaw, Reed & Mace, 1982). An effectivity encompasses the structure, functionality, and actions that might enable the organism to pursue what is, roughly speaking, in human terms, a goal (see 3.1.4.5). Using its "climber things," the squirrel exploits the climb-ability of the branch to escape a predator. Using its "alighter things," the seagull exploits the alightability of the rock for rest. Using its "grabber things," the wolf exploits the grab-ability of the deer to obtain nutrients.

Taken together, geometrical, kinetic, and task constraints constitute a description of a person's effectivities.... Geometrical and kinetic constraints are intrinsic to physical properties of the actor and can be directly measured/ calculated with respect to an external frame of reference (i.e., height, weight). Task constraints are more functional and "psychological." They include all of the intentional, goal-directed considerations that encourage the person to perform one action rather than another (Mark, Dainoff, Moritz & Vogele, 1991, pp. 484-85).

Affordances and effectivities are neither specific organs of perception nor specific tools of execution. Rather, they are emergent properties produced by interactions between the perceiver and its environment. A well-tuned relationship between affordances (opportunities) and effectivities (abilities) generates a dialectic that, Csikszentmihalyi (1990) argues, is experienced by humans as highly satisfying. He calls this dialectic the "flow experience" (p. 67). Affordances and effectivities are mutually grounded in and supported by both the regularities of the physical structure of the environment and by the psychosomatic structure of the perceiver. It is meaningless to consider whether an object affords action without also considering the nature of corresponding effectivities that some organism might employ to exploit that affordance to achieve the organism's intentions: A flat, 3-feet-tall rock affords convenient sitting for a human but not for a bull elephant.

Indeed, meaning, of perhaps a quite fundamental sort, is extant in the relationship of organisms to their environments. Here is our working definition of ecological meaning:

Those clusters of perceptions associated with the potential means--that is, affordances and effectivities--by which an organism might realize its intentions.

Our definition does not assume that organisms are conscious or that they use semantics or syntax, nor does it necessarily assume that organisms are purposeful. Our definition does assume, however, that all organisms engage in activities that can be characterized as intentional or goal oriented.

Many biologists and psychologists would criticize these notions of intentionality or goal orientedness, especially when they are applied to simpler forms of life. Intentionality implies teleological thinking, and such critics typically hold teleology in disrepute because it has been associated with doctrines that seek in natural systems evidence of deliberate design or purpose-vitalism and creationism, for example. A narrower conception of intentionality and goal orientedness, however, is very convenient in the study of selforganizing and cybernetic systems (see 3.1.2.5) in which

. . . feedback mechanisms are characterized by the fact that the input is controlled by the output, and thus stabilizes the output, or makes the performance relatively independent from disturbing influences. One can consider the stability of the output as the "goal" of the system. To turn the argument around, whenever a behavior in biology is described as goal directed, i.e., teleological, it is very likely that a feedback mechanism is involved (Gregory, 1989, p. 176).

When ecological psychologists attribute intentions and goals to nonhumans, they typically do so in the more limited sense associated with functional maintenance of homeostasis--or in Maturana's (1980) terms, autopoesis--rather than as an attribution of deliberate design or purpose (see 3.1.3.5).

8.5.2.4. Unification of Effectivities and Affordances. A curious phenomenon emerges in humans when effiectivities engage with affordances: The affordances often seem to disappear from awareness. Winograd and Flores (1986) cite Heidegger's example of hammering a nail:

To the person doing the hammering, the hammer as such does not exist. It is part of the background of readiness-to-hand that is taken for granted without explicit recognition or identification as an object. It is part of the hammerer's world, but is not present [to awareness] any more than are the tendons of the hammerer's arm (p. 36).

The disappearance of an affordance from awareness signals the psychological unification of the effectivity with the corresponding affordances. One can think of this unification as an extension of the effectivity by the affordance or as a "path" for action-perception. In everyday activity, the very "routin-ness" and familiarity of such paths makes them invisible to the organism. Thus, another metaphor for directness of action-perception is transparency. The organism perceives and acts through the unified effectivity-affordance (arm and hammer) and is therefore only aware of the object of perception and action (the nail).

The hammer presents itself as a hammer only when there is some kind of breaking down or readiness-to-hand. Its "hammernness" emerges if it breaks or slips from grasp or mars the wood, or if there is a nail to be driven and the hammer cannot be found.... As observers, we may talk about the hammer and reflect on its properties, but for the person engaged in ... unhampered hammering, it does not exist as an entity (Winograd & Flores, 1986, p. 36).

In terms of ecological psychology, we can think of Heidegger's concepts of breakdown and resulting unreadiness-to-hand as a partial decoupling of an effectivity from its corresponding affordance. Breakdowns "serve an extremely important cognitive function, revealing to us the nature of our practices and equipment, making them 'present-to-hand' to us, perhaps for the first time. In this sense they function in a positive rather than a negative way" (Winograd & Flores, p. 78).

8.5.2.5. Everyday Learning and Media Environment. For J. J. Gibson, the ordinary world of everyday learning and perception is

... not the world of physicists. It is far larger than atoms and far smaller than galaxies. It is the geological environment of ground, water, earth, and sky, and it is the evolved world of flora and fauna. Ile substances and surfaces of the ground and the animate and inanimate things above it need to be described at their own level, if we are to understand what animals are aware of. Equally important, the energy fields in the medium of the air (or water) exist at an ecological level and contain information about the furnishings of our habitat. These sources of information in stimulation must therefore be studied ecologically, not physically. Psychology *must begin with ecology, not with physics and physiology, for the entities of which we are aware and the means by which we apprehend them are ecological (Reed, 1988, p. 230).

The popularity of Donald Norman's (1988) book, The Psychology of Everyday Things, which shares key ideas with Gibson's work, testifies to an increased awareness by the general public that media engineers and scientists must look beyond the merely physical properties and attributes of systems. In an age of knowledge workers and postindustrialism, human habitats and artifacts must accommodate mentality as well as physicality, and support creativity as well as consumption. Cognitive ergonomics (Zucchennaglia, 1991) is just as important as corporal ergonomics (Mark, Dainoff, Moritz & Vogele, 1991): Both depend to a considerable extent on understanding fundamental human capabilities that were tuned long ago by ecological circumstances. Yet if new media are to support the development and utilization of our uniquely human capabilities, then we must acknowledge that the most widely distributed human asset is the ability to learn in everyday situations through a tight coupling of action and perception.

Covariations of tactile, auditory, visual, and other sensory inputs are normal concomitants of most action, and must be presumed to provide most of our normal clues for the integrated perception of our environments. The detection and analysis of covariation is thus the main function required of the relevant cortical network. Given this, the system has all it needs by way of an internal representation of the tactile world-as-perceived for the organization of relevant action. The state of conditional readiness for action using other dimensions of the effector system, such as walking, can be derived directly from this representation, without any need for an explicit "map" (MacKay, 1991, p. 84).

Emerging media systems and technologies appear headed toward a technical renaissance that could free media products from constraints that now limit the agency of end users: the limited symbology and dimensionality of paper and ink, the shadows captured and cast from a single point of view in photographs and films, the fixed sequences and pacing of analog broadcast technology.

Saba (1988), for example, argues that the virtual contiguity afforded by integrated telecommunications systems (incorporating venues such as two-way video conferences and distributed or networked multimedia) transforms possibilities for participation in communities of production and learning. In his view, the convergence of media technologies, techniques for multitasking, and sharing of tools for communication reduces the transactional distance between participants and reduces dependence on communication through explicit discourse.