situated

Moreover, because sensorimotor contingencies will differ depending on the idiosyncratic nature of sensory apparatuses and skills necessary for their use, perceptual experience is likely to differ for different kinds of perceivers. “If perception is in part constituted by our possession and exercise of bodily skills … then it may also depend on our possession of the sort of bodies that can encompass those skills, for only a creature with such a body could have those skills. To perceive like us, it follows, you must have a body like ours” (Noë 2004, 25).17 The body relativity of perception, Noë thinks, requires a method of investigation distinct from that to which cognitivism is committed. Even were one to retain the strategy of explaining perception as the achievement of algorithmic processes, the body must enter this explanation in ways cognitivists never foresaw. “The point is not that algorithms are constrained by their implementation, although that is true. The point, rather, is that the algorithms are actually, at least in part, formulated in terms of items at the implementational level. You might actually need to mention hands and eyes in the algorithms!”18 (Noë 2004, 25, emphasis in original).

Successful perceivers, according to O’Regan and Noë, will have mastery, or tacit knowledge, of the sensorimotor contingencies particular to their own sensory apparatuses. But it is not just motion of the eye that reveals sensorimotor contingencies. Often these contingencies show themselves only when the entire organism is in motion, moving around objects, picking them up, seeing them against a background of other objects, etc. The actions an organism takes in its efforts to perceive display a kind of skill—a skill that in turn reflects familiarity with the sensorimotor contingencies that impose order on patterns of changing stimulation that would otherwise appear as an uncrackable cipher—a code waiting to be broken.

emergentist Russellian panpsychism

Moreover, there is good reason to think that the combination problem is more tractable than the explanatory gap faced by the physicalist. The concepts involved in articulating the physical facts are very different from the concepts involved in articulating the consciousness facts: the former concepts are third-personal and quantitative, the latter concepts are first-personal and qualitative. This radical difference provides grounds for thinking there could never be a priori derivations from the physical facts to the consciousness facts, and hence that zombies would remain conceivable even for an ideal reasoner. There is no such support for the conceivability of micro-experiential zombies, given that in this case first-person qualitative concepts are employed in the articulation of both the fundamental and the higher-level facts.

microexperiential zombie

What kind of consciousness is mass, as opposed to charge or spin? What is it like to be a quark? Panpsychism is a broad theoretical framework, and it will take time to fill in the details. Compare: it took decades of hard work to bridge the gap between the basic principles of Darwinian evolution by natural selection and modern genetics.

to explain human consciousness, in terms of more basic forms of consciousness

There is, then, a good case for panpsychism even before we get to thinking about the need to explain human and animal consciousness in more fundamental terms. But, of course, the Russellian panpsychist also aspires to do this. Physicalists believe that consciousness can be explained in terms of processes that do not involve consciousness. There is a general consensus that no one has yet worked out how to do this. The Russellian panpsychist proposes an alternative research programme: Instead of attempting to explain consciousness in terms of non-conscious elements, try to account for the consciousness of humans and other animals in terms of more basic forms of consciousness, basic forms of consciousness that are postulated to exist as essential properties of basic material entities. It is still early days in the panpsychist research project, but the history of failure of physicalist solutions to problem of consciousness makes it rational to explore other options.

Furthermore, it is arguable that panpsychism is the most theoretically virtuous theory of matter consistent with both the data of physics and our first-person awareness of the reality of consciousness. This is what I have called ‘the simplicity argument’ for panpsychism (Goff 2016, 2017). Assuming the falsity of dualism, we know that some material entities, that is, living brains, have a categorical nature that involves consciousness. Neither physics nor introspection give us any clue as to the categorical nature of material entities outside of brains, or indeed of the categorical nature of the components of brains. And therefore, the most simple, elegant, parsimonious hypothesis is that the categorical nature of the stuff outside of brains is continuous with that of brains in also being consciousness-involving. Or to put it another way: we would need a reason for thinking that matter has two kinds of categorical property rather than one. Special relativity is not entailed by the empirical datum that light is measured to be the same in all frames for reference, but it is arguably the simplest hypothesis consistent with that datum. Similarly, panpsychism is not entailed by the datum of consciousness but it is arguably the simplest hypothesis consistent with that datum.

When one first hears about the view that quarks are conscious, it is natural to interpret what is being claimed dualistically. That is to say, one imagines that the quark has its physical properties and its experiential properties sitting side by side, as it were. However, this would not be a Russellian form of panpsychism. For the Russellian panpsychist, the physical properties of the quark—such as mass and charge—are forms of consciousness. Those very properties that physics characterizes behaviouristically are, in their categorical nature, forms of consciousness.11 In this way, the Russellian panpsychist avoids the dualist’s difficulties reconciling the efficaciousness of consciousness with the causal closure of the physical world.

You might actually need to mention hands and eyes in the algorithms!”

Notice, however, that recognition of this point does not require that cognitivists give up their central doctrine, that is that cognition is computation over symbolic representations. Lakoff and Johnson's conclusions, for instance, are consistent with the possibility that basic concepts are encoded symbolically, and that more abstract concepts evolve in virtue of computational operations on the representations of concepts such as up, down, back, and front. Thus, research on Conceptualization might better be viewed not as a challenge to cognitivism, but as a splash of cold water in the faces of cognitivists who have resisted the need to consider how an organism's body might shape and limit the representations that are available to computational processes.

Focus on Conceptualization marks a departure from traditional cognitivism insofar as it takes very seriously the body's role in concept acquisition. Whereas cognitivists have tended toward a view of mind that allows its investigation to proceed without detailed investigation of the properties of the body that forms a shell around it, work on answering the question of Conceptualization shows the danger in this strategy. A more complete understanding of how organisms conceive their world, and why they develop the conceptions of the world that they do, demands careful attention to the nature of interactions with the world that an organism's body permits. Because cognition takes place in a body, one should not be surprised to find that the properties of the body have a significant impact on the products of cognition. It is recognition of this point that, some in embodied cognition believe, is lost in traditional cognitivism.

In a series of clever experiments, Boroditsky and Ramscar (2002) show an effect that might be taken as evidence for the kind of embodied conceptualization that Lakoff and Johnson describe. Boroditsky and Ramscar exposed subjects to the ambiguous sentence “Next Wednesday's meeting has been moved forward two days.” The sentence is ambiguous because, depending how one conceives of “moving forward,” the new meeting day may be either Monday or Friday. Subjects were then asked “What day is the meeting now that it has been rescheduled?” Boroditsky and Ramscar were interested in whether primes that suggested to subjects motion through space would influence the subjects’ interpretation of the ambiguous sentence. The data show that exposure to primes suggesting thoughts of forward motion would make “Friday” the more probable response; when primes suggested thoughts of something moving toward the subject, the subject would more likely respond “Monday.” In explaining these data, Boroditsky and Ramscar argue that subjects conceive of time in spatial terms. When already prompted to think about forward motion, subjects are inclined to see the rescheduling of the meeting as moving away (toward Friday), and when subjects are compelled to reflect on objects moving toward them, they tend to understand the meeting as coming toward them—moving from Wednesday to Monday.

Lakoff and Johnson argue that many of the basic concepts are spatial: a reflection of those properties of the human body that determine motion through space. Human concepts of up and down, for instance, emerge from the vertical orientation that human bodies typically assume in wakeful moments. Similarly, the concepts front and back have their origin in the fact that sensory organs for sight, smell, and taste are on the same plane of the human body. The ascent from these basic concepts to more abstract ones follows a complex route, but the details are not important in this context. An organism's understanding of its world is embodied, Lakoff and Johnson contend, in the sense that the concepts through which it views the world are grounded in properties of its body.

These two claims of Clark's, that a cognizer's need to build models of the world might often (or on occasion) be satisfied instead by access to information already present in the environment, and that organisms’ access to information will differ depending on how particular organisms are able to interact with their environment and which interactions with the environment are important for them, are part of what embodied cognition theorists mean when they claim that cognition is situated. The picture of the situated cognizer differs considerably from that of the symbol processor. The former is interacting with its environment in a manner designed (or intended) to extract information that is relevant to its needs. The latter uses its sense organs to, in effect, take a picture of the world, which it then submits for processing, awaiting the final result: a new and more refined picture, suitable for guiding action. The situated cognizer is the bicyclist who crashes to the ground if he stops moving; the symbol processor is the archer who holds steady, closes her eyes, and visualizes the target.

Clark's second point about embodied cognition reflects another Gibsonian theme: researchers in embodied cognition, because they reject or at least harbor misgivings about the need for world modeling, prefer to investigate the organism-environment “fit” that facilitates an organism's ability to access information external to it. Different species of organism, owing to differences in the properties of their bodies (including their sensory organs) will fit the environment in distinct ways. Differences between the bodies of a koala and a quoll account for differences in how each kind of organism interacts with a tree. As these interactions differ, so too the variety of information that becomes available to each organism differs.8 But also, because koalas and quolls use trees for different ends, the information to which each is tuned to respond may not be the same.

Andy Clark, who among philosophers has doubtless done most to advance and refine conceptions of embodied cognition, emphasizes two aspects of embodied cognition that he draws from converging fields of research (1997).7 First is the idea that embodied cognition eschews “excessive world modeling” (1997, 23). This idea, although distinct from Varela et al.'s of “world making,” also has a Gibsonian ancestor. As we have seen, Gibson believed that the presence of information in the environment obviated a need for inferential operations on impoverished representations. An object's size does not need to be calculated from representations of visual angle and distance because information about object size is already specified within the optic array. Embodied cognition theorists have developed this Gibsonian idea, seeking to discover how various cognitive activities that cognitivists have sought to explain in representationally rich terms might be better or as well explained using representationally scanty resources.

Randall Beer, whose support for embodied cognition arises from his painstaking and insightful studies of simulated robot behavior, says that “embodiment emphasizes the role of an agent's own body in its cognition … An embodied nervous system can utilize the natural biomechanics of its body, the geometry of its sensory surfaces, and its ability to actively control these surfaces to simplify many problems” (2003, 211). Beer's research shares the dynamical systems perspective that Thelen also champions, according to which cognitive behavior is the product of interactions between brain, body, and world that become intelligible through the lens of dynamical systems theory. In this conception of embodiment, behavior that the cognitivist would have assumed to be the product of purely computational processes taking place within the brain becomes instead the output of a dynamical system consisting of brain, body, and environment, whose changes over time can be modeled with differential equations. The salient point is that the tools of dynamical systems theory can often (or, at least, sometimes) explain, and often (or, at least, sometimes) explain better how cognition arises than would a traditional cognitivist account.

Thelen's characterization of embodiment thus echoes Varela et al.'s. Embodied cognition emphasizes the special form cognition takes as a result of the idiosyncratic properties of the cognizer's body.

[t]o say that cognition is embodied means that it arises from bodily interactions with the world. From this point of view, cognition depends on the kinds of experiences that come from having a body with particular perceptual and motor capabilities that are inseparably linked and that together form the matrix within which reasoning, memory, emotion, language, and all other aspects of mental life are meshed. (2001, 1)

To summarize, the ecological school of psychology Gibson developed brought with it a conception of psychology quite distinct from the cognitivist school that within Gibson's lifetime came to predominate. For cognitivists, cognition consists in computational operations on informationally meager symbolic representations in order to produce new representations that serve cognitive functions such as perception, language comprehension, problem solving, and so on. For Gibson, perception is the detection of information that, with no further embellishment, suffices to specify features of an observer's world. According to Gibson, the active observer could, by collecting and sampling the wealth of information contained within the optic array, know its world in terms relative to its needs. We will see these Gibsonian themes recurring time and again as we now turn toward a discussion of embodied cognition.

We come finally to the most controversial (and maligned3) element of Gibson's theory of perception: affordances. Gibson believed that information in the optic array sufficed to specify opportunities for action, thus providing observers with an ability to perceive e.g., the chair as something that can be sat upon, the surface as something that can be walked on, the food as nourishing, and so on. Because the present interest in Gibson is as an inspiration for embodied cognition, this is not the place to consider whether Gibson could mount an adequate defense of this view. The significance of affordances for current purposes concerns their agent-relativity. Whether a chair can be sat upon, and thus whether an observer correctly perceives it as something that can be sat upon, depends on the properties of the observer as well as those of the chair. The perception of a given chair as being something on which one can sit is thus a perception that is tied to a particular observer. Similarly, to use another of Gibson's examples, a branch affords a resting place for a bird, but not a person. Accordingly, a bird can perceive the branch as affording a resting place for itself, but a person cannot. The perception of affordances, in short, is an observer-relative form of perception.4 The very same chair might be perceived as “sittable upon” by one observer but not another.5 Gibson believed that how the world is perceived, and what it is perceived to afford, depends on the kinds of interactions an observer of a particular sort—having a body of a particular kind—can make with its environment.

Before explaining Gibson's account of affordances, it is worth pausing to contrast Gibson's view of perception with the cognitivist's. For Gibson, perception is strongly coupled with action, because most of the invariants within the optic array do not appear except against a background of change. Perceivers actively scan, move within, or otherwise manipulate the environment. For the cognitivist, however, perception is a process that begins and builds from static images that flutter across the retina. The vision scientist Richard Gregory neatly encapsulates the antithesis of Gibson's view: “Perceptions are constructed, by complex brain processes, from fleeting fragmentary scraps of data signaled by the senses … Current sensory data (or stimuli) are simply not adequate directly to control behavior in familiar situations” (1972, 707).1 Gregory's characterization of the perceptual process motivates typical cognitivist accounts of, for instance, the perception of object size: the size of an object must be computed from a representation of the amount of visual angle it subtends together with a representation of its distance from the observer. In contrast, Gibson denies the need for computations over representations to determine object size. Object size is already specified by invariants in the optic array such as horizon-cuts and the amount of graded texture an object's base obscures. These invariants render the need for computation otiose.2

An example of an invariant is the nondisruption of edges in covering surfaces (Gibson 1979, 76). An observer's motion relative to one surface that is in front of another will cause the edges of the distal surface to change while the edges of the proximal surface remain constant. This invariant specifies the relative distances of the objects to the observer. Similarly, objects at different distances from an observer will, if the same size, be cut by the horizon in the same proportion (178). Thus, the fact that two trees have their top thirds above the horizon specifies to the observer that they are the same height, regardless of their relative distances from the observer (and, of course, regardless of whether the observer has conscious access to this invariant).

The information about which Gibson speaks is contained in the optic array. Converging on every point in an illuminated environment, including the sense organs of a perceiver, is light reflected from surrounding surfaces. Preserved in this array of light is information about surfaces’ shapes, textures, and colors. This is because the intensity and wavelength of light from one surface will differ from the intensity and wavelength of light reflected from other surfaces, in effect causing a projection onto an organism's sense organs of a group of solid visual angles, the edges of which mark the boundaries of surfaces in the environment. As the organism moves, the elements of the optic array undergo lawful transformations. Some visual angles expand, others contract, new ones appear, and so on. However, within this changing pattern of visual angles will be a variety of constant features. These invariants, Gibson argued, specify, or carry information, about structure in the environment. Perception, according to Gibson, is the process by which this information is detected.