The Bounds of Cognition

Frederick Adams, Kenneth Aizawa

• Publisher: Wiley
• Year: 2011
• Print ISBN: 9781444332704, 9781405149143
• eBook ISBN: 9781444357301
• Edition: 1
• Topic: Philosophy
• Subtopic: Philosophy History & Theory

An alarming number of philosophers and cognitive scientists have argued that mind extends beyond the brain and body. This book evaluates these arguments and suggests that, typically, it does not.

• A timely and relevant study that exposes the need to develop a more sophisticated theory of cognition, while pointing to a bold new direction in exploring the nature of cognition
  
• Articulates and defends the "mark of the cognitive", a common sense theory used to distinguish between cognitive and non-cognitive processes
  
• Challenges the current popularity of extended cognition theory through critical analysis and by pointing out fallacies and shortcoming in the literature
  
• Stimulates discussions that will advance debate about the nature of cognition in the cognitive sciences

Chapter 1

Introduction

In the Monadology, Wilhelm Gottfried Leibniz claimed that

7. Further, there is no way of explaining how a Monad can be altered or changed in its inner being by any other created thing, since there is no possibility of transposition within it, nor can we conceive of any internal movement which can be produced, directed, increased, or diminished there within the substance, such as can take place in the case of composites where a chance can occur among the parts. The Monads have no windows through which anything may come in or go out. The attributes are not liable to detach themselves and make an excursion outside the substance, as could sensible species of the Schoolmen. In the same way neither substance nor attribute can enter from without into a Monad. (Leibniz, 1979, pp. 251–2)

The Leibnizian idea is that monads, or minds, do not causally interact with things outside of themselves. They are, as we might say, closed causal systems. Instead, the apparent causal relations between one monad and the rest of creation are simply due to the infinite power of God to pre-establish a harmony among the internal workings of individual monads so that they appear to causally interact.

It is an understatement, however, to say that few philosophers or psychologists these days take seriously the idea that human or animal minds work as do Leibnizian monads.¹ The orthodox view in cognitive science maintains that minds do interact with their bodies and their environments. Cognitive processes within brains are not causally isolated from the rest of the world. Cognitive processing depends on the environment in ways too numerous to mention. The developing fetal brain can be poisoned by maternal alcohol consumption. Dense congenital cataracts can impair the development of normal visual processing, especially during a critical or sensitive period of child development. Years of practicing the violin can shape the amount of cortical material dedicated to representation of the fingers of the left hand. Humans and other animals causally interact with the world in order to perceive it by vision, olfaction, and audition. Cognitive processes are influenced by low oxygen concentrations at high altitudes and high nitrogen concentrations at great underwater depths. Cognitive processes are also influenced by any number of psychoactive drugs from alcohol to nicotine to Δ⁹-tetrahydrocannabinol. Cognitive processes clearly depend on the body and environment. In short, contemporary cognitive psychology is anti-Leibnizian: cognitive processes do causally depend on bodily and environmental processes.

Under the influence of the phenomenological tradition in philosophy, dynamical systems theory, and mobile robotics, the extended cognition movement has sought to move beyond mere anti-Leibnizianism. The extended cognition movement maintains that cognitive processes depend on bodily and environmental processes, but not merely causally. It is not just that bodily and environmental processes causally influence cognitive processes; they literally constitute or realize cognitive processes. Cognitive processes do not occur exclusively within brains; they span brains, bodies, and environments. Cognitive processes extend from brains into their surrounding bodies and physical environs. A handful of examples will illustrate the kinds of cases used to challenge orthodoxy.

A common method for finding the product of 347 and 957 is to write the problem down on a piece of paper, aligning the “3” in the hundreds place of the first numeral with the “9” in the hundreds place in the second numeral, aligning the “4” in the tens place of the first numeral with the “5” in the tens place of the second numeral, and so on.² This facilitates the application of the partial products algorithm in several ways. Since the numbers are written one above the other, one can rely on vision to keep the ones, tens, and hundreds places coordinated. One does not have to devote special attention or burden memory in order to accomplish this coordination. In addition, since one can write down a number that has to be carried above the column to which it will be carried, this removes the burden of remembering the number to be carried. Further, by recording one’s work on paper at each step, one is spared the task of remembering where one is in the calculation and the results of past bits of computation. It is because the use of pencil and paper generally provides a faster and more reliable method of computing the products of large numbers that one so frequently turns to it.

Surely the best known example in the extended cognition literature is the Inga–Otto thought experiment, developed by Clark and Chalmers (1998). In this story, Inga is a normal human subject who hears from a friend that there is an interesting exhibit at the Museum of Modern Art and decides to go see it. She thinks for a moment before recalling that MOMA is on 53rd Street, and then sets off for 53rd Street. In contrast to Inga, Otto suffers from Alzheimer’s disease and has numerous memory lapses. To help him compensate, he must rely upon cues in his environment. In order to handle addresses, Otto relies on a notebook in which he writes this kind of information. Thus, when he hears his friends talking about the interesting exhibit at the Museum of Modern Art, he reaches for his notebook to look up the address. Finding that the museum is on 53rd Street, he sets off.

Other examples involve the role of the body and movement in cognitive life.³ During the course of normal human activity, the head and eyes typically move through space. This happens any time a person walks, drives a car, or turns her head. During these activities, the light entering the eye carries information about the relative distances of objects. The light projected from more distant objects changes differently than does the light projected from less distant objects. In a simple case, there is what occurs when one fixates on objects on the distant horizon. Here, nearer objects appear to be displaced farther in the direction opposite to the motion than do more distant objects. Humans are extraordinarily good at using this motion parallax as a guide to the relative distances of objects. As vision scientists often put it, motion parallax is a powerful monocular cue for relative depth.

One hypothesis concerning these cases is that they are all instances in which human cognitive processes in the brain take advantage of non-cognitive tools found in the body and environment. On this orthodox construal of tool use, humans have a more or less stable set of cognitive capacities for learning, remembering, perceiving, and attending. Learning and training make for greater or lesser degrees of stability. Learning and training can yield dramatic changes in the cognitive processes involved in such abilities as playing a violin, tasting fine wine, and speaking natural languages. In many situations, however, humans do not seek to modify their cognitive apparatus. Instead, they live with the cognitive mechanisms they have and complement them with tools that enable them to compensate for their cognitive shortcomings. It is because of limitations on human short-term memory that humans use pencil and paper for computing the products of large numbers. Because the information about whether or not to carry a one is on the paper, it need not be kept in memory. Because of the alignment of the columns of numerals on the page, one need pay less attention to being sure that tens are added to tens and hundreds are added to hundreds. In the Inga–Otto case, the reason Otto uses the notebook to store information is obviously that his long-term memory is failing him. He lacks the normal memory resources that Inga possesses. The notebook enables him to compensate for this lack. Otto’s use of the notebook is not exactly like Inga’s use of normal long-term memory. The notebook is a tool that he uses, in conjunction with his spared cognitive capacities of seeing, reading, and writing, in order to achieve some tolerable level of functionality. The different ways in which bodily motions influence the play of light from objects near and far is a potentially useful tool for determining relative distance. Humans use this tool, among many others, because they cannot directly perceive the relative distance of objects.⁴

Recent work advancing the hypothesis of extended cognition offers radically new and different analyses of these cases. Advocates of extended cognition complain that orthodox cognitive science is in the grip of a picture of the locus of cognition. Orthodoxy maintains, without justification, so the story goes, that cognitive processing occurs within the brain. Advocates of extended cognition take the foregoing cases to show, or make plausible, the view that cognitive processing literally extends from the brain into the bodily and environmental tools that humans exploit. The manipulation of pencil and paper in the computation of large products becomes a literal part of one’s cognitive processing. The notes in the notebook that Otto keeps with him constantly constitute part of Otto’s memory and the physical basis of part of his stock of beliefs. The use many animals make of bodily motions to induce motion parallax constitute part of their perceptual processing. In short, according to the hypothesis of extended cognition, the tools many organisms use (often) become part of their cognitive processors. This view is so radical that one might well be skeptical that anyone really means to assert such a thing. Yet there are many clear and simple assertions of it:

Cognitive processes span the brain, the body, and the environment. (van Gelder and Port, 1995b, p. ix)

Cognitive processes are not located exclusively inside the skin of cognizing organisms. (Rowlands, 1999, p. 22)

What I am claiming is that not only thoughts, but also feelings and sensations, must be seen as supervening on the entire brain–body–world nexus. (Rockwell, 2005, p. 71)

Cognitive processes are partly constituted by physical and bodily movements and manipulations of objects in real-world environments. (Gibbs, 2006, p. 12)⁵

Not satisfied with noting the causal dependencies between cognition and bodily and worldly processes – not satisfied with simply rejecting Leibnizian monadology – the advocates of extended cognition champion a constitutive dependency.

What motivates this bold new hypothesis? In our reading of the literature, we have come across essentially five distinct types of arguments for the hypothesis of extended cognition. The most pervasive type focuses attention on the way in which structures outside of the brain causally interact with parts of the body and external world. We group these arguments under a broad category of “coupling arguments.” They invoke one or another type of causal connection or coupling relation between the brain and the body/external world in order to make the case that the non-brain components should be understood as realizing cognitive processes. According to Mark Rowlands,

cognitive processes are not located exclusively inside the skin of cognizing organisms because such processes are, in part, made up of physical or bodily manipulation of structures in the environments of such organisms. (Rowlands, 1999, p. 23)

The psychologist Raymond Gibbs, discussing intentions rather than cognitive processing per se, reasons in essentially the same way:

The windsurfer continually affects and is affected by the set of the rig, so the behavioral intention to successfully windsurf emerges as a result of the interaction between the person and environment. Focusing on the agent alone, or on how the agent responds to the environment, fails to capture the complex nuances of windsurfing behavior. Just as it is important to understand the significance of paper and pencil when one does long division, where the cognition of doing long division is in part “offloaded” into the environment, the intentionality in windsurfing is best understood as a distributed cognitive behavior involving a person, a device, and the environment. (Gibbs, 2001, pp. 117–18)

These examples are among the more succinct presentations of this argument.⁶

Another type of argument might be thought of as a version of a coupling argument. These arguments begin by drawing attention to causal connections between the brain and parts of the body or environment, but then, rather than concluding that cognition extends into these parts of the body or environment, they conclude that the brain and the body, and perhaps the environment, constitute a cognitive system. They conclude that there is an extended cognitive system. From this conclusion, there is a tacit shift to the conclusion that cognitive processing extends from the brain into the body and the environment. Part of Haugeland’s articulation of embodied and embedded cognition involves this two-step argumentation.⁷ Clark and Chalmers may also have some version of this in mind. After describing some examples where they believe there is extended cognition, they write

In these cases, the human organism is linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right. All the components in the system play an active causal role, and they jointly govern behavior in the same sort of way that cognition usually does. If we remove the external component the system’s behavioral competence will drop, just as it would if we removed part of its brain. Our thesis is that this sort of coupled process counts equally well as a cognitive process, whether or not it is wholly in the head. (Clark and Chalmers, 1998, pp. 8–9)

Notice that Clark and Chalmers move from a claim about a brain and an external object constituting a cognitive system – the cognitive system hypothesis – to the claim that cognitive processing is not wholly in the head – the extended cognition hypothesis. That is, the argument has the implicit two-step structure we mentioned above: first infer from the existence of certain causal interactions that there is a cognitive system involving brain, body, and environment; and then infer from the cognitive system hypothesis that there is extended cognition.

A third pattern of argument supposes that there are cases in which processes that span the brain and body, or brain, body, and environment, are in all relevant respects just like cognitive processes that occur within the brain. Add to this the tacit premise that if there is this equivalence, then the processing spanning the brain and body, or brain, body, and environment, is cognitive processing. This yields a simple modus ponens argument for extended cognition.⁸

The fourth type of argument sits uneasily with the third. These are “complementarity arguments.” The cognitive equivalence arguments rely on putative equivalences between cognitive processes thought to occur in the brain and processes occurring in the brain, body, and environment. This is the kind of thinking one finds underlying the claim that Inga is, in all important and relevant respects, exactly like Otto. By contrast, the complementarity arguments rely on the fact that, because brain processes are of one character and bodily and environmental processes are of another, brain processes and bodily and environmental processes work well together. The combination of intracranial and extracranial processes achieves results that are in some sense superior to those achieved by just the brain alone. It is the fact that the brain alone figures out large products relatively slowly and with relatively low reliability, where the brain – in conjunction with sensory and motor skills and pencil and paper – more quickly and more reliably computes large products, that argues for the view that cognition extends into the arms, hands, pencil, and paper. What makes for a tension between these two lines of thought is – to put matters crudely – that, in the first line, one is making the case that Otto and Inga are cognitively the same, but in the second that Otto and Inga are not cognitively the same.

The fifth, and most distinctive, of all the arguments contends that the theory of evolution by natural selection supports the view that cognition extends into the environment. The major premise of the argument is that, if some cognitive feature is adapted to work in conjunction with some feature of the environment, then that feature of the environment is really part of the cognitive apparatus of the mind. This is an argument developed in considerable detail in Rowlands (1999) and briefly reviewed in Rowlands (2003).

Given that there is so much to be said in favor of the hypothesis of extended cognition, one might wonder what could possibly sustain the old-fashioned hypothesis of brain-bound cognition. The advocates of extended cognition have a short answer: mere prejudice. Haugeland begins his discussion of embodied and embedded cognition by paying homage to René Descartes’s enduring influence on contemporary cognitive science orthodoxy.⁹ Descartes, of course, did not maintain that the mind is causally isolated from the material world. He was a two-way interactionist, famously believing that mind and body interacted by way of the pineal gland. What Descartes did maintain was that reason is constituted by a distinct thinking substance that survives bodily death. One way to be anti-Cartesian would, thus, be to endorse some form of physicalism and maintain that reason, or the mind, or cognition, is realized or constituted by the brain. Something like this is cognitive psychological orthodoxy. Haugeland, however, champions a more radical course. He proposes that the mind is constituted not just by the brain, but by the brain, body, and environment. The mind is embodied in flesh and blood and the larger causal nexus of the world. Rowlands (1999, 2003) and Rockwell (2005), in their own ways, also conjecture that the current demarcation of the boundaries of cognition is a remnant of a largely discredited Cartesian view of cognition.¹⁰ The charge appears again in a plainer form in Clark and Chalmers (1998), Clark (2003), and Clark (2005). There, the idea is simply that the hypothesis that cognition is brain-based is merely an unjustified prejudice. Rockwell (2005), for his part, provides a somewhat different diagnosis of the prejudice: “But I also maintain that to say a mind must be embodied only by the brain of an organism is a hangover from a justly discredited epistemology that builds its foundation on atomism and sense-datum theory” (Rockwell, 2005, p. 49).

Despite the growing popularity of the hypothesis of extended cognition, we remain defenders of orthodoxy. We argue that there are principled reasons for believing that the kind of cognitive processing cognitive psychologists care about is, essentially without real-world exception, intracranial. Two principal hypotheses about the nature of cognitive processes support this. In the first place, we maintain that cognitive processes involve non-derived mental representations; that is, cognitive processes involve representations that mean what they do in virtue of naturalistic conditions that do not include the content-bearing states, properties, or processes of other entities. Because these representations are typically found inside, but not outside, the brain, cognitive psychologists have one principled reason to think that cognition is typically intracranial. Second, cognitive psychologists attempt to distinguish the cognitive in terms of its underlying mechanisms. Cognitive processes are those that take place in virtue of certain mechanisms. Although these mechanisms could (conceptually, metaphysically, and physically) occur outside of the brain, they typically do not. In general, these mechanisms are often poorly understood, but they have features that are familiar to any serious student of cognitive psychology. For example, there is Miller’s (1956) discovery that short-term memory has some sort of “size capacity.” Consider a task such as listening to a string of distinct letters of the alphabet presented one per second, and then repeating the sequence. Normal human subjects are generally quite capable of performing this task for strings of five, six, and seven letters. But for eight-, nine-, and ten-letter strings, recall falls off dramatically. The standard hypothesis is that short-term memory has a fixed capacity of seven, give or take two, items. Seven items fit comfortably in memory, where more tend to “fall out” and be forgotten. We do not mean to propose that in order to be short-term memory, something must respect Miller’s rule. Rather, we propose that findings such as this should guide us in determining what memory is like and what really differentiates cognitive processes and mechanisms from non-cognitive processes and mechanisms. Our empirical hypothesis, the one we think is embraced by the majority of cognitive psychologists,...

Table of contents

• Cover
• Title page
• Copyright page
• Contents
• Preface
• Acknowledgments
• 1 Introduction
• 2 Refining the Issues
• 3 Original Content
• 4 Cognitive Processes
• 5 The Mark of the Cognitive, Extended Cognition Style
• 6 The Coupling-Constitution Fallacy
• 7 Extended Cognitive Systems and Extended Cognitive Processes
• 8 Cognitive Equivalence, Complementarity, and Evolution
• 9 Inference to the Best Explanation and Extended Cognition
• 10 Future Directions
• Bibliography
• Index