Embodied, Situated, and Extended Cognition
Traditionally, philosophers and psychologists hold that the mind receives some information about its environment through the senses, and uses this information to reason and make decisions, which may lead to action.
Some philosophers and psychologists, whose views are often grouped together under the headings “embodied cognition,” “situated cognition,” and “extended cognition,” have criticized this conception of the relation between cognition and the cognizer’s environment. Although the views denoted by these headings differ in some respects, for simplicity I will use the expression “extended cognition” in what follows, noting the differences between these views when appropriate. It has often been noted that this new movement combines several distinct positions without clearly marking their differences (e.g., Wilson 2002, Rupert 2004).[122] In this section, I briefly distinguish four threads, before discussing in some detail the idea that mental states and psychological processes are not located in the brain.A first thread is methodological. Proponents of extended cognition contend that a proper understanding of psychological processes involves examining the environment in which cognition takes place (e.g., Hutchins 1995) - a position often referred to by the label “situated cognition.” This methodological claim is sometimes justified on the grounds that psychological processes are designed (by evolution or by learning) for specific (physical and social) environments. To illustrate, according to Gerd Gigerenzer and Ulrich Hoffrage (1995), the processes underlying probabilistic reasoning are designed to manipulate representations of natural frequencies, rather than probabilities, consistent with the fact that for most of human history, probabilistic information was only available in the form of natural frequencies.
A second thread highlights the importance of agency in understanding cognition (e.g., Noe 2004, Gallagher 2005).
This emphasis is supposed to stand in contrast to cognitive psychology’s traditional focus on situations that involve no or little action (e.g., chess playing, remembering words on a list, etc.).A third, more radical thread takes issue with the idea that cognition involves manipulating representations (e.g., Brooks 1991, Thelen and Smith 1994). Anti-representationalists typically focus on some phenomena that proponents of representation-based approaches to cognition explain (or would explain) by means of representations and representation-based processes. They then explain these phenomena without positing any process that manipulates representations. On this basis, they draw the following induction: if postulating representations is not needed to explain these phenomena, behavior and cognition at large can be explained without representations (for discussion, see, e.g., Vera and Simon 1993, Clark and Toribio 1994).
A fourth strand of argument, often referred to by the labels “extended cognition” and “extended mind,” focuses on the location of mental states and psychological processes. Philosophers and psychologists have often identified token mental states with brain states and psychological processes with neural processes. In sharp contrast, Mark Rowlands writes that “[c]ognitive processes are not located exclusively inside the skin of cognizing organisms” (1999, 22); while Andy Clark and David Chalmers argue: “[W]e will argue that beliefs can be constituted partly by features of the environment, when those features play the right sort of role in driving cognitive processes. If so, the mind extends into the world” (1998, 12). According to this view, at least some token mental states are external to the body or involve extra-corporeal objects as proper parts, while cognition involves the manipulation of these entities.
Two well known examples might usefully illustrate this view. Consider first how we perform a complex arithmetical operation by hand, such as the multiplication of 37 by 23 (Clark and Chalmers 1998, Adams and Aizawa 2001, Noe 2004, Adams and Aizawa 2008).
We write down one numeral below the other. Focusing on the rightmost digital of each numeral (“7” and “3”), we multiply the numbers they express. We write down “1” on a third line and write “2” as a carry-over. We then multiply 3 (the number expressed by the leftmost digital of the first numeral) by 3 (for the rightmost digital of the second numeral) and add the carry-over. We write down the numeral “11” left of the numeral “2” (and so on). To perform this multiplication, we create and manipulate objects (viz., numerals) that are external to our body in a rule-governed manner. According to proponents of extended cognition, the numerals are part of our mental states and their rule-governed manipulation counts as psychological processing. As Alva Noe puts it, “[i]f the pencil and paper are necessary for the calculation, why not view them as part of the necessary substrate for the calculating activity?” (2004, 220).Consider a second example. Clark and Chalmers (1998) propose that in some situations, a notebook can literally be part of someone's memory. They compare a normal woman, Inga, who relies on her memory to determine the address of the Museum of Modern Art, with an Alzheimer patient, Otto, who relies on his constantly available notebook to determine the address of the museum. Clark and Chalmers contend that in spite of the differences between Inga's and Otto's cases, both Otto and Inga believe that the Museum of Modern Art is located on 53rd street:
To provide substantial resistance, an opponent has to show that Otto's and Inga's cases differ in some important and relevant respect. But in what deep respect are the cases different? To make the case solely on the grounds that information is in the head in one case but not in the other would be to beg the question. If this difference is relevant to a difference in belief, it is surely not primitively relevant. To justify the different treatment, we must find some more basic difference between the two.
(Clark and Chalmers 1998, 6)This last example is useful to bring to the fore the central argument for the view that mental states and psychological processing extend beyond the skin: there is no significant difference between some states that involve extra-corporeal entities and some brain states, or between the manipulation of extra-corporeal entities, such as consulting a notebook, and the manipulation of mental representations, such as consulting one's memory. If there is really no significant difference between them, then some states that involve extra-corporeal objects are genuine mental states and some processes that involve manipulating these objects are genuine psychological processes.
Unsurprisingly, this fourth thread has caused a fair amount of discussion among philosophers. Most critics grant that if there were no significant differences between states of the brain and states involving extra-corporeal objects as proper parts, or between processes involving only brain states and processes involving extra-corporeal objects, then not all mental states and psychological processes would be in the head, but they deny that the antecedent of this conditional is satisfied. Particularly, endorsing the computational representational theory of mind (§1), Fred Adams and Ken Aizawa (2001) have argued that mental states are representations that are endowed with an original content and that psychological processes are computational processes defined over these representations. For them, these two properties are “the mark of the cognitive.” Because the extra-corporeal objects that are manipulated (for instance, the addresses in Otto's notebook) do not have any original content, they do not count as mental states and their manipulations do not count as psychological processes.
There are two main worries with Adams and Aizawa's argument. First, it rests on controversial (though widespread) necessary conditions for something to be a mental state and for something to be a psychological process, ones that might be rejected by proponents of extended cognition.
Second, accepting Adams and Aizawa's necessary conditions, proponents of extended cognition might reply that a state counts as mental provided that some of its parts have an original content, and that a process counts as psychological provided that some steps in this process involve states with original content (or states with some parts having an original content).Robert Rupert's (2004) main argument against extended cognition does not fall prey to these worries, because he does not assume a specific mark of the cognitive. Rather, focusing on memory, he highlights the differences between the properties of memory retrieval on the one hand and the use of extra-corporeal objects to store information on the other. A large number of generalizations have been found about how people store information in memory (e.g., interference effects[123]) and how they retrieve information from memory (e.g., recency effects[124]). He correctly notes that few of these generalizations apply to the gathering of information from physical mnemonic aids, such as notebooks. Furthermore, any generalization that could apply to information retrieval from both memory and mnemonic aids would probably be about a much larger class of systems, which would include, but not be identical to, the class of cognitive systems. Rupert concludes that treating states and processes within the brain and states and processes involving extra-bodily objects as physical parts is not a promising strategy for cognitive science.
4.