Outsourcing the mind

This review by Dan Lloyd, Professor of philosophy at Trinity College, Connecticut on two new books on the embodied mind appeared in the American Scientist.

SUPERSIZING THE MIND: Embodiment, Action, and Cognitive Extension. Andy Clark. xxx + 286 pp. Oxford University Press, 2008.

OUT OF OUR HEADS: Why You Are Not Your Brain, and Other Lessons from the Biology of Consciousness. Alva Noë. xvi + 214 pp. Hill and Wang, 2009.

Sum res cogitans. “I am thinking substance.” With these words, written in about 1640, René Descartes simultaneously created the modern mind and gave it a huge philosophical headache. Cartesian dualism opened an abyss between mind and matter, which was good news for mechanistic physics. But “thinking substance” was thereby expelled from nature, and psychology has labored ever since to bring the mind back into the scientific fold—an effort that has culminated with the rise of cognitive neuroscience. A modern-day Descartes would perhaps say, “I am synaptic substance,” or, to be more accurate, “I am the information transmitted across neural networks.” Sum cerebrum.

Swapping brain for mind bridges the metaphysical gulf, but lesser dualisms still haunt cognitive science. The popular thought experiment of a “brain in a vat” captures the intuition that cognition and consciousness depend exclusively on the machinery between our ears. In the standard vat tale, one is asked to imagine that one’s brain has been removed from one’s body and placed in a vat of nutrient fluids, and that all of its normal neural inputs and outputs are being simulated by a supercomputer. The brain has no way of knowing whether it is in a skull or in a vat. Can we be sure that this is not our current situation? How do we know that anything beyond our brains is real rather than virtual? The moral of the thought experiment seems to be that the neural representations of body and world are only indirectly related to real external things. Is this state of affairs an anachronistic “Cartesian materialism,” with a neural computer on one side, and the body and world on the other?

A contemporary movement in cognitive science looks beyond this lingering dualism, promoting “extended cognition” and “embodiment” as crucial components of the science of mind. Andy Clark, author of Supersizing the Mind, and Alva Noë, author of Out of Our Heads, are preeminent expositors of extended and embodied cognition, and their two books represent the state of the movement, complete with its internal tensions.

Clark critiques what he calls the “brainbound” model, which depicts the mind “as essentially inner and, in our case, always and everywhere neurally realized.” He puts forth a contrasting model, which he refers to as EXTENDED, “according to which thinking and cognizing may (at times) depend directly and noninstrumentally upon the ongoing work of the body and/or the extraorganismic environment.” He further characterizes this model as follows:

According to EXTENDED, the actual local operations that realize certain forms of human cognizing include inextricable tangles of feedback, feed-forward, and feed-around loops: loops that promiscuously criss-cross the boundaries of brain, body, and world. The local mechanisms of mind, if this is correct, are not all in the head. Cognition leaks out into body and world.

The first section of Supersizing the Mind surveys work in which considerations of embodiment and extended informational resources have transformed theories of perception, cognition and motor control. Consider the problem of walking—easy for us, but a challenge for robots, especially if their walking is highly engineered via exact mechanical control of every joint, precalculated in a central controller. Such highly motorized and micromanaged movement is inefficient both physically and computationally. Biological walking, in contrast, exploits the “passive dynamics” of the material body. We ride on springy, free-swinging limbs. Once set in motion, animal bodies like ours saunter on their way with minimal shoving and shaping from the brain.

Our bodies lighten the load for our brains in many other ways as well. Expressive gestures, including words, Clark observes, are not merely communicative output but may also “function as part of the actual process of thinking.” Gestural information can interact with language. As we talk (to others and to ourselves), we also listen, using our bodies and words as reminders and abbreviations. Outsourcing is truly powerful, however, when we exploit the myriad cognitive scaffolds of the world around us, particularly the world of artifacts. In general, when information is available in the environment, we will use it instead of framing a “brainbound” thought. For example, to play the video game Tetris, one must anticipate whether moving shapes will fit together. To test for a match, one can manipulate the shapes mentally or try out the rotations on screen. Skilled players use on-screen manipulation rather than tax their minds.

The picture of mind that emerges in Clark’s treatment, although not “brainbound,” remains neurocentric. He portrays the brain as a lazy genius at the center of a loose confederacy of tricks and tweaks. Some of the outsourcing involves symbol manipulation and some involves shortcuts that eliminate the need for language (or other symbols) altogether. Clark’s vision loosens up cognitive science itself: Good old-fashioned computational models still have a place, albeit a diminished one. We need to be alert to every kind of computation (including dynamical systems of distributed representations) and, more important, to the diversity of vehicles for computation, many of which are outside the head.

One consequence of the extended approach is a “hypothesis of cognitive impartiality”:

Our problem-solving performances take shape according to some cost function or functions that, in the typical course of events, accord no special status or privilege to specific types of operation (motoric, perceptual, introspective) or modes of encoding (in the head or in the world).

Cognition doesn’t care how or where it occurs! Extended theorizing in the spirit of this hypothesis could reshape cognitive science, embedding embodied human life in an ecology of useful and symbolic objects, a flow in which neural activity is one eddy among many.

Extended cognition entails a supersized mind, and much of the second part of Clark’s book defends the philosophical idea that mind itself leaks into the world. The core argument is really “Well, why not?” Worldly activity with cognitive scaffolding accomplishes many of the same ends as neural computation and evidently saves the brain a lot of bother, so why not let the mind be where the work is done? In particular, philosophical views about beliefs, regarding both what they are and how they fit in the life of the mind, seem neutral about where a belief is located. It might be spread among the synapses, or on a microchip hardwired into the brain, or in a handy notebook—any of those media could preserve all the features of the belief. Critics search for some “mark of the mental” that will keep thinking inside the skull, but Clark counters that in some cases these lines in the sand are the result of mistaken analyses of the mental, and in other instances, they are lines easily crossed by extended minds.

Alva Noë’s target is consciousness, and in broad terms his position is compatible with Clark’s. Noë writes that in Out of Our Heads his central claim is that to understand consciousness—the fact that we think and feel and that a world shows up for us—we need to look at a larger system of which the brain is only one element. Consciousness is not something the brain achieves on its own. Consciousness requires the joint operation of brain, body, and world. Indeed, consciousness is an achievement of the whole animal in its environmental context. I deny, in short, that you are your brain.

Even in this passage we see an ambiguity that runs throughout the book. (Unlike Noë’s thoughtful and thorough Action in Perception [The MIT Press, 2004], Out of Our Heads is a manifesto of hyperbolic claims resting on sketches of argument.) Is it that we are not merely our brains (which is Clark’s view as well), or that we are not our brains at all? How completely “out of our heads” are we? The radical possibility runs through passages like this one, in which Noë describes what he refers to as a “sensorimotor, enactive, or actionist approach”:

Seeing is not something that happens in us. It is not something that happens to us or in our brains. It is something we do. It is an activity of exploring the world making use of our practical familiarity with the ways in which our own movement drives and modulates our sensory encounter with the world. Seeing is a kind of skillful activity.

This theory of perception has strong echoes of J. J. Gibson’s “ecological approach” to perception, along with the embodied phenomenology of Maurice Merleau-Ponty. Seeing is certainly skillful activity, but is that activity mediated by anything like an inner representation, or by a state of conscious awareness that either guides the activity or results from it? Noë repeatedly edges toward elimination of inner states, only to hedge:

We ourselves are distributed, dynamically spread-out, world-involving beings. We are not world representers. We have no need for that idea. To put the point in a provocative way, we are, in Merleau-Ponty’s memorable phrase, “empty heads turned toward the world.” And as a result of this, our worlds are not confined to what is inside us, memorized, represented. Much more is present to us than is immediately present. We live in extended worlds where much is present virtually, thanks to our skills and to technology.

We have no need for representation—at all? Or is it that our worlds are not confined to what is inside us, those innards nonetheless enacting cognitive, computational and conscious processes?

There is much to be gained by recognizing the intricate embedding of consciousness in the body and the world. Cognitive science (including the study of consciousness) has been shackling itself with its brainbound assumptions. Consciousness depends on its embodied embedding, but should it be identified entirely with the myriad couplings and loops the brain surely exploits?

This question is in play between Clark and Noë. Each disputes the other in passing. For Clark, one of the distinctive features of conscious awareness is the capacity to disengage from enacted specifics. Driving a car, for example, requires precise enactive sensorimotor coupling for beginners and experts alike, but for the practiced driver the details drop out of awareness. Noë’s view identifies consciousness with all the activities of the extended mind and thus implies that skilled enactors remain aware of everything engaged by their performance.

Clark also enlists clever experiments and reports of brain deficits that suggest a dissociation of awareness from sensorimotor knowledge. Familiar optical illusions make objects that are in fact identical look as though they differ in size, but when one reaches to grasp them, one’s fingers open to the same (correct) extent regardless of context. The eye-to-hand loop is not fooled by appearances, however things may seem to the conscious mind. Similarly, some brain lesions can impair the ability to describe a scene while sparing the capacity for fluent sensorimotor interaction, whereas other lesions have the opposite result. Thus there seem to be two partially distinct systems, one mediating fluent behavior and the other generating the model of the world available to consciousness.

From Noë’s point of view, to supersize the mind while leaving consciousness inside the head seems arbitrary, if not fainthearted. Any such distinction regards one fantastically complex information processing system as conscious, while declaring another, equally complex system, not. What’s the difference?

An expansive, totalized theory of consciousness like Noë’s solves the problem by dissolving it:

The problem of consciousness, then, is none other than the problem of life. What we need to understand is how life emerges in the natural world.

Descartes framed the modern mind with a sentence beginning “I am . . . ,” launching centuries of debate about how to complete the thought. But until now, that first-person grammatical form has been the eye of the storm, a nexus of subjectivity to which a world appears. In a truly post-Cartesian world of looping, evanescent, chattering, clattering networks, the problem of consciousness may simply disappear. The sum may turn out to be less than the whole of its parts.

Dan Lloyd is the Thomas C. Brownell Professor of Philosophy at Trinity College, Connecticut. He is the author of Radiant Cool: A Novel Theory of Consciousness (The MIT Press, 2004) and is currently working on a book-length philosophical dialogue titled Ghosts in the Machine.