Extended Cognition (Part I)

by Carl Pierer

Andy Clark and David Chalmers present a thesis of extended cognition andClark&Chalmers extended mind in their seminal 1998 paper: “The extended mind”. In it, they attack the idea that cognition and mind should be confined to the boundaries of our skull. Instead, they suggest, that the tools and instruments used in cognitive processes are part of the cognitive process. Clark and Chalmers support this claim by the following consideration. Suppose, Wolfram is seated before a computer screen and asked to play Tetris. He has to decide whether certain shapes that keep appearing on the top of the screen will fit into slots at the bottom of the screen. There are three scenarios:

1) Wolfram rotates the shapes mentally to decide whether they fit or not.

2) Wolfram presses a button to rotate the shape on the screen and then compares the shape to the slot.

3) Wolfram lives in the not-so-far future and has a neural implant which allows him to rotate the shape physically in his head. He can either use the neural implant to decide whether the shape will fit or use the initial method in 1). (See Clark & Chalmers 1998)

Clark and Chalmers think that in 3) it does not matter whether Wolfram uses the implant or the mental rotation, either way deciding whether the shape will fit counts as a cognitive process. To meet obvious objections, it can be supposed that the neural implant works exactly like the computer in 2). But if it is the case that the neural implant in 3) is functionally just like the computer in 2), there is no difference between 3) and 2). Neither is there a difference between 1) and 3). But then, if 1) is like 3) and 3) is like 2), 1) must be like 2). Therefore, since 1) is a cognitive process, 2) is a cognitive process as well and cognition extends.

Now, of course it might be objected that 1) is like 3) if in 3) Wolfram choses the initial method, while 3) is like 2) if Wolfram choses the implant. But this objection misses the point: in 3) either choice of method will be part of the cognitive process; thus there is no difference to the cognitive status of the three scenarios.

On Clark and Chalmers' view, Wolfram and the computer create a “coupled system”:

All the components in the system play an active causal role, and hey jointly govern behaviour in the same sort of way that cognition usually does. If we remove the external component the system's behavioural 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 & Chalmers 1998, p. 8)

Coupled systems are thus ubiquitous: the person using their smartphone to find the nearest bus stop, the pianist playing the piano to test their new piece of music as well as the writer jotting down ideas and modifying them in the process all constitute coupled systems.

The argument for extended heavily relies on an assumption known as the parity principle:

If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process.

This principle derives from the idea that it should not matter how exactly a cognitive or mental process is instantiated for it to count as cognitive or mental. For example, whether the sensation of pain is caused by C-fibres in the human brain firing or by little soap bubbles popping in a Martian brain made of chocolate should not be relevant to the phenomenon of “pain”. Provided the functional inputs and outputs in the Martian and the human are the same, they are both experiencing pain regardless of the physical instantiation.

Fred Adams and Ken Aizawa published a couple of papers attacking the extended cognition thesis. They see themselves as defending common sense against misguided philosophers such as Clark, Chalmers and Dennett: they hold that while extended cognition is a theoretical possibility, as a matter of contingent fact there does not exist any actually extended cognitive process. Furthermore, they do accept the parity principle and thereby that he brain-skull boundary cannot be a theoretically significant boundary. However, they require a further condition. One of their main arguments is that for a process to count as cognitive it ought to bear “the mark of the cognitive”. Going into detail of what this mark of the cognitive might be is beyond the scope of this essay. Suffice to say that Adams and Aizawa distil two criteria that are closely modelled on human cognition.

Enter Mark Sprevak, arguing that these criteria are too closely modelled on human cognition. In his paper on “Extended Cognition and Functionalism”, Sprevak suggests that the idea of extended cognition is derived from two basic, intuitively plausible principles, which all four philosopher mentioned so far seem to accept:

(F1) If an organism counts as sufficiently like us on a coarse-grained global functional comparison, then it is a cognitive agent.

(F2) If a cognitive agent contains a representation-manipulating process that is significant for guiding its action (in appropriate ways) when employed, then that process is one of its cognitive processes. (Sprevak 2009, p. 521)

(F2) is related to the parity principle. (F1) is at the heart of the idea that the actual physical instantiation of a cognitive process does not matter to its status as cognitive. Sprevak argues that Adams and Aizawa set the parameters for (F1) too fine-grained, so much so that they rule out too much: that is, precisely the intuition that the details of the instantiation of a cognitive process should not matter. Conversely, jointly accepting (F1) and (F2) – as Clark and Chalmers seem to do – entails a leakage of cognitive processes into the external world. For example, a case akin to Sprevak's:

Let Amadeus be a highly gifted musical talent. He can compose a 6 part fugue on any given theme on the spot. Now consider Babel, who does not have this musical ability, yet she does have a super-computer at her hands. This super-computer will compose a 6 part fugue if presented with a theme in no time. (Sprevak 2009, p. 518)

Sprevak argues that holding (F1) and (F2) entails that Amadeus' and Babel's cognitive abilities are just the same: Take a Martian with a brain with the same physical setup as Babel's super-computer. Without loss of generality, we can postulate that (F1) holds. Furthermore, it is easy enough to see that (F2) will obtain; the Martian uses their super-computer-like brain to compose the 6 part fugue, their brain is thus “significant for guiding its action”. But now, the only difference between Babel and the Martian is that in the one the super-computer is external while in the other it is internal. This, however, is not sufficient to establish that they differ in their cognitive status – if (F1) and (F2) are accepted.

Now, Sprevak further argues that this style of argument (i.e. invoke some Martian that has the process under discussion as a function of their brain to show that the process is cognitive) works for any process whatsoever, hence any process will count as cognitive. Something, it seems, has gone terribly wrong with the premises of functionalism. Indeed, this would present a very fine counter argument to the functionalist view. There is no room for a mild form of extended cognition: if cognition is allowed to extend no line can be drawn where to stop, without facing the charge of being arbitrarily restrictive and thus discounting an intuitively cognitive Martian.

There is certainly something crude about Sprevak's argument, the Martian intuition seems to be stretched beyond the breaking point: Is a Martian, who counts by inflating fleshy tubes inside her brain, still cognitive in the sense we like to say humans are cognitive? But even if this is granted, the second part of this essay, which will be published here on the 19th of January, will follow Richard Menary in arguing that Adams, Aizawa and Sprevak like to interpret the causal connection in the coupled system in a way that differs from Clark and Chalmers'. It is only on this interpretation – and not on the one more favourable to the purposes of extended cognition – that Sprevak's argument against the explosive consequences of extended cognition works.