Persons and the Extended-Mind Thesis

An extract from Lynne Rudder Baker’s paper:

Cognitive scientists have become increasingly enamored of the idea of extended minds. The extended-mind thesis (EM) is the claim that mentality need not be situated just in the brain or even within the boundaries of the skin. EM is the modal claim that it is possible that the mind is not bound by skull or skin. EM is quite radical: A mind is a collection of processes

that easily extends to tools, programs, other minds, and language. Cognitive states may have all sorts of components—neural, bodily, environmental. The heart of the extended-mind thesis is that we biological creatures can “couple” with nonbiological entities or features of our environment and thereby expand the entities that we are. Some versions do away with enduring agents altogether; “extended selves” (Clark and Chalmers 1998, 18) are relatively transitory couplings of biological organisms and external resources. There is a huge and complex literature on the idea of an extended mind, both pro and con. I focus here on some of Andy Clark’s work, especially the article he wrote with David Chalmers in 1998, “The Extended Mind.”

Here is my plan for the article. First, I show how EM can be seen as an extension of traditional views of mind. Then, after voicing a few qualms about EM, I reject it in favor of a more modest hypothesis that recognizes enduring subjects of experience and agents with integrated bodies. Nonetheless, my modest hypothesis allows subpersonal states to have nonbiological parts that play essential roles in cognitive processing. I present empirical warrant for this hypothesis and show how it leaves room for science and religion to coexist.


One way to understand EM is to start with a traditional picture of mental states and then see how EM revises it. Here is one traditional picture: Many mental states have content—states of desire are satisfied or not, intentions are fulfilled or not, beliefs are true or false. Typically, contents are given by the that-clauses that follow psychological and linguistic verbs such as thinks, believes, desires, intends, says. Thoughts and other contentful states are said to have two kinds of properties: properties determined by the content and properties of the vehicles that carry content. (The distinction brings to mind Descartes’ distinction between representative, or objective, reality and formal reality.)

What makes a thought the very thought that it is is its content. That is, states that have content are individuated by their contents. The thought that snow is white differs from the thought that grass is green in virtue of the difference between snow’s being white and grass’s being green. The contents of thoughts (and other mental states)—that snow is white or that grass is green—are carried by vehicles, traditionally thought of as neural states. Neural states are internal states, “in the head.” Call this view vehicleinternalism.

Even if, as traditionally supposed, vehicles are internal to the thinker, the contents of thoughts may be determined by phenomena outside the thinker (or so many think). The view that the contents of our thoughts— and, hence, the identity of which thoughts we can have—are determined by features of the environment is called content-externalism. To take a well-worn example, Pam, who lives on Earth where there is H2O (water), may have the thought that water is wet. Now suppose that there is another world in which there is an abundant liquid that looks like water but is not water because it has a different chemical composition. Suppose also that people in that waterless world drink, brush their teeth with, and swim in the water look-alike. The inhabitants speak a language similar to English, but when they utter what sounds like “water” in English, they are not speaking of water but of the other stuff, the water look-alike. In that world, where there is no water (no H2O), a molecular duplicate of Pam—call her Cam—could not have the thought that water is wet. The duplicate’s thought can be reported in English as the thought that twater (the stuff in the other world) is wet, but it cannot be reported as the thought that water is wet. Cam’s thoughts that correspond to Pam’s water-thoughts are twater-thoughts. Cam cannot have any water-thoughts. Because Pam and Cam are molecular duplicates, their brain states are of identical types. But if content-externalism is true, their thoughts are not of identical types.

Although content-externalism is not altogether uncontroversial, it is well-entrenched enough to say that a version of the traditional view combines vehicle-internalism and content-externalism. We may see EM as an extension of the externalism of contents to an externalism of vehicles. With the combination of vehicle-internalism and content-externalism in the background, EM treats vehicles in a way analogous to the way that the (externalist) traditional view treats content. EM is a kind of extreme externalism in that not only the determinants of content but also the vehicles may be located outside the organism. Clark, an early proponent of EM, characterizes EM as “the view that the material vehicles of cognition can be spread out across brain, body and certain aspects of the physical environment itself ” (2005, 1). EM in effect extends content-externalism to vehicle-externalism (Hurley 1998). Until recently, vehicles were thought to be only brain states (vehicle-internalism). According to vehicle-externalism, however, not only is the content determinable by features of the environment, but the vehicle also may be spread out into the environment. Vehicle-externalism supposes that cognitive processes may have vehicles that include aspects of the environment.

For example, beliefs are normally embedded in memory, but they need not be. Consider Otto, who is impaired in such a way that he cannot form new memories. He writes down what he wants to remember in a notebook that he always carries. Suppose that Otto is on Fifth Avenue in New York City and is looking for the Museum of Modern Art (MoMA). He knows that he cannot simply search his memory for the location of MoMA, so he automatically reaches for his trusty notebook and looks up the address: 53rd Street. The information in the notebook—just like the information stored in brain-based memory—“is reliably there when needed, available to consciousness and available to guide action, in just the way that we expect a belief to be” (Clark and Chalmers 1998, 13). Viewed from the lens of EM, the skin is seen as an artificial boundary.

In one of the most important early articles on EM, Clark and Chalmers state that “when it comes to belief, there is nothing sacred about skull and skin. What makes some information count as a belief is the role it plays, and there is no reason why the relevant role can be played only from inside the body.” For some of Otto’s mental states—his extended beliefs—Otto and his notebook are coupled; they form a cognitive system, all components of which are causally active. The “relevant parts of the world are in the loop, not dangling at the other end of a long causal chain” (Clark and Chalmers 1998, 9). Hence, extended cognition is sometimes called “active externalism” (p. 8).

As Clark puts it later, “taken as a single, integrated system, Otto-andthe- notebook exhibit enough of the central features and dynamics of a normal agent having (amongst others) the dispositional belief that MOMA is on 53rd Street to warrant treating him as such.” He asks rhetorically, “If an inner mechanism with this functionality [passive aspects of memory] would intuitively count as cognitive, then (skin-based prejudices aside) why not an external one?” (Clark 2005, 7) The point of EM is that neither the organic brain nor the skin sets a boundary on the vehicles of cognition. Features of the environment may or may not be components of the vehicle.

In general, tools extend cognition. A tool, “even when temporarily in use, is rapidly assimilated into the brain’s body maps and is treated (temporarily) just like a somewhat less sensitive part of the body.” For example, the receptive visual field of a macaque using a rake for as little as thirty seconds becomes elongated as if the rake were part of the arm (Clark 2005, 8). Use of a tool, even temporarily, changes neural maps. Neural plasticity “makes it possible for new equipment to be factored deep into both our cognitive and physical problem-solving routines” (p. 9). So, we become physical and cognitive hybrids—part biological and part artifactual.

Not only is there physically extended cognition, there is socially extended cognition as well. As many have observed, their spouses are their external memory devices. My husband serves as part of a vehicle for many of my memories. For such memories (as well as in other ways), a proponent of EM may say that my husband and I are coupled. Coupling between agents is effected by language, among other things. Language “is not a mirror of our inner states but a complement to them. It serves as a tool whose role is to extend cognition in ways that on-board devices cannot” (Clark and Chalmers 1998, 18).

Clark emphasizes that hybridization (Otto-and-his-notebook) is quite normal. We routinely use “transparent technologies” such as pencils for calculating sums. We are just shifting combinations of biological and nonbiological elements.