COGS Seminar, University of Sussex

University of Sussex, Falmer, Brighton BN1 9QN

Tuesday May 10, 4pm Pevensey I, Room 1A7


Perception, Action, and the Extended Mind

Professor Mike Wheeler
School of Arts and Humanities: Philosophy
University of Stirling\


Abstract

According to the extended cognition hypothesis (henceforth ExC), there are actual (in-this-world) cases in which thinking and thoughts (more precisely, the material vehicles that realize thinking and thoughts) are spatially distributed over brain, body and world, in such a way that the external (beyond-the-skin) factors concerned are rightly accorded cognitive status (where 'cognitive status' signals whatever status it is that we ordinarily grant the brain in orthodox, non-extended cognitive theory). David Chalmers, one of the original architects of ExC, has recently articulated an objection to the view which turns on the claim that the idea of cognitive extension is in conflict with an intuitive thought that we ought to preserve. Chalmers puts that intuitive thought like this: 'It is natural to hold that perception is the interface where the world affects the mind, and that action is the interface where the mind affects the world. If so, it is tempting to hold that what precedes perception and what follows action is not truly mental.' Chalmers proceeds to offer a defence of ExC against the worry. In my talk I'll (i) set the scene with some comments about how one ought to understand ExC (comments that involve some criticisms of Andy Clark's version of the view), (ii) explain Chalmers' objection and his response to it, (iii) argue that Chalmers' response fails, and (iv) suggest that we should solve the problem by ditching the intuitive thought. This final move will enable me to address a challenge that, up until now, has arguably not been met successfully by advocates of ExC, that is, to say what consequences the view has for empirical work in cognitive science and psychology.


Mike Wheeler is Professor of Philosophy at the University of Stirling. Prior to joining Stirling Philosophy in 2004, he held teaching and research posts at the Universities of Dundee, Oxford, and Stirling (a previous appointment). His doctoral work was carried out in the School of Cognitive and Computing Sciences at the University of Sussex. His primary research interests are in philosophy of science (especially cognitive science, psychology, biology, artificial intelligence and artificial life) and philosophy of mind. He also works on Heidegger and is interested in exploring ideas at the interface between the analytic and the contemporary European traditions in philosophy. His book, Reconstructing the Cognitive World: the Next Step, was published by MIT Press in 2005.


Messages to the list are archived at http://listserv.liv.ac.uk/archives/philos-l.html.
Prolonged discussions should be moved to chora: enrol via
http://listserv.liv.ac.uk/archives/chora.html.
Other philosophical resources on the Web can be found at http://www.liv.ac.uk/pal.

WWWW (What Wheeler Wouldn't Write)

2 Online intelligence is generated through complex causal interactions in an extended brain-body-environment system Recent work in, for example, neuroscience, robotics, developmental psychology, and philosophy suggests that on-line intelligent action is grounded not in the activity of neural states and processes alone, but rather in complex causal interactions involving not only neural factors, but also additional factors located in the nonneural body and the environment. Given the predominant role that the brain is traditionally thought to play here, one might say that evolution, in the interests of adaptive efficiency, has been discovered to outsource a certain amount of cognitive intelligence to the nonneural body and the environment. In chapters 8 and 9 we shall explicate this externalistic restructuring of the cognitive world-with its attendant (typically mild, but sometimes radical) downsizing of the contribution of the brain in terms of what Andy Clark and I have called nontrivial causal spread (Wheeler and Clark 1999). (Wheeler, 2005, p. 12)

Now, this is something that Wheeler did write back in 2005, but it is, I speculate, what Wheeler wouldn't write now.  I think he has gotten to be more careful about loose phrases such as "intelligence is generated through".  That's ambiguous between a thesis about ontogenetic development, on the one, hand (which A&A think is true) and a thesis about, say, the supervenience base of intelligence (which A&A think is not true).  "Grounded" might also be ambiguous, but I would read it as a kind of supervenience claim.  The appeal to evolution, however, suggests that it is not an ontogenetic thesis that is up for grabs, but a phylogenetic thesis.  All these are different claims and I think that Wheeler is probably now on to this. I think he may believe all three theses; I only believe two.  But, this is just a marking out he lay of the land ...

Wheeler 2005 on representation and computation 2

The representational theory of mind and the computational theory of cognitive processing are empirical hypotheses. However, they are empirical hypotheses whose truth has been pretty much assumed by just about everyone in cognitive science.  (Wheeler, 2005, p. 8). 

Now, I suppose that if your aim is to undermine opposing views, it's easiest just to say that they are simply assuming something.  But, when you have major, explicitly acknowledged empirical hypotheses--such as that cognition involves rule and representations--then it seems unlikely that such hypotheses will be mere assumptions.  So, if you want to understand the major tenets of an opposing view, you should probably dig around and find out why they hold them.

So, I don't know why Gibsonians are so interested in the direct perception of affordances, but I figure there must be some experimental result or something that drives this.  This is not just something they assume.  And, I assume that there is some reason that Maturana and Varela (and it seems Evan Thompson following them) think that life and mind are very intimately related.  I have no idea what that is, but I'm not going to go out on a limb and say they are just assuming that there is a connection.

Now, of course, finding out why some group holds a view takes a lot of time.  I've been rooting around trying to find out what drives EP.  I don't think reading things like Gibson, 1979, or TSRM are really doing it for me.  I think I need to go back to some of the earlier experimental work.  I've read some Maturana, Varela, and most of Mind in Life, but I still don't get it.

Wheeler 2005 on representation and computation 1

The representational theory of mind and the computational theory of cognitive processing are empirical hypotheses. However, they are empirical hypotheses whose truth has been pretty much assumed by just about everyone in cognitive science.  (Wheeler, 2005, p. 8).

I like Mike's writing because it is clearly wrong, rather than obscurely wrong like so much of the EC lit.  (Isn't there a principle of deontic logic according to which if you're going to be wrong, you should be clearly wrong?)  Mike is probably right that the representational theory of mind and the computational theory of cognitive processing are empirical hypotheses whose truth has long been pretty much assumed by just about everyone in cognitive science.  But, I think it's because folks have largely been satisfied that the empirical case for that has been made, so that it's time to move on to other dimensions of these views.  It's, to me, just like the fact that so much of biology has pretty much assumed that evolution is true.  That has been established, so it is time to move on to other dimensions.

Remember that representationalism emerged in opposition to Skinner's behaviorism which was often anti-representational and which was orthodoxy circa 1955 (right?).  In that context, representationalism didn't just emerge by assumption; there must have been something somewhere that gave at least some people some reason to think that there must be some representations somewhere somehow.  Just so, evolution emerged in opposition to creationism  which was orthodoxy circa 1855 (right?).  In that context, evolution didn't just emerge by assumption; there must have been something somewhere that gave at least some people some reason to think that there must be some evolution somewhere somehow.

For Sutton, et al. HEC vs. HEMC is not at Issue (Mostly)

Placing different cases within such a multidimensional framework is a more fruitful empirical project than continuing to debate whether cognition or memory is ‘really’ extended or ‘merely’ embedded.  (Sutton, et al.)

So, here, HEC versus HEMC seems to be dismissed or downplayed.

Assessing these two distinct lines of thought, Clark saw the complementarity between heterogeneous inner and outer resources as grounding ‘the more interesting and plausible argument’:

The argument for the extended mind thus turns primarily on the way disparate inner and outer components may co-operate so as to yield integrated larger systems capable of supporting various (often quite advanced) forms of adaptive success. The external factors and operations, in this model, are most unlikely to be computationally identical to the ones supported directly in the wetware ... (Clark 1998, p.99)

(Sutton, et al.)

But, here Sutton, et al., seem to be citing Clark, with approval, as holding the view that complementarity leads to HEC.  Maybe Sutton, et al., don't care to fight over HEC versus HEMC, but I am assuming that Clark did and was supporting HEC.

Yet, as we’ve noted, some common objections to parity- or functionalism-based extended cognition do not apply to complementarity-based extended cognition: in turn, the latter view may face different challenges of its own. Complementarity therefore deserves fuller and independent exploration if we want to evaluate the overall case for extended cognition. One tack for such constructive exploration involves detailed application of complementarity considerations to the key domain of memory, and this is the driving aim of the research program we describe in the second half of this paper. First we need to examine responses to complementarity.   (Sutton, et al.)

Now, in the first passage, Sutton, et al., suggest that they are not going to engage the HEC versus HEMC issue, but then in the second they cite with approval Clark's apparent efforts to argue for HEC by way of complementarity.  Then in the third, they suggest that this is a complementarity argument for EC, which I'm guessing is HEC.

It doesn't matter that much to me which road Sutton, et al. want to take, i.e. whether they want to engage HEC versus HEMC or not, but what I care about is that a) complementarity does not seem to support HEC and b) complementarity does not rule out intracranial cognition.  And, in an exchange over at Gary's blog, John seems to agree with me on this. 

And, I think John and I also agree that the study of the complementary relations between brain and body and environment are ok.

So, I'm not sure that I have that much in the way of substantive disagreements with Sutton, et al.

Menary(?) and Wheeler(?) Buy the Distinction between Causation and Constitution

Wheeler (chapter 11) argues that the extended mind is a kind of extended functionalism. Wheeler points out that the extended mind is not simply a weak claim about the causal dependence of some cognition on external factors (cf. Adams and Aizawa's coupling-constitution fallacy). It is a stronger claim involving the constitution of cognition, at least in part, by external factors. Therefore, the extended mind is not simply an embodied-embedded thesis that treats external props and tools as causally relevant features of the environment. It is a thesis that takes the bodily manipulation of external vehicles as constitutive of cognitive processes. (Menary, 2010, p. 21).

So, this pretty clearly buys into the causation/constitution distinction in order to make out the hypothesis of extended cognition.  So, as I noted earlier, pace Hurley, it seems to me that it is the job of the advocates of EC, not the critics, to explicate this distinction.

EC and Multiple Realization

Wheeler here presents an account of the relationship between EC and MR that I think was already articulated in Sprevak's Journal of Philosophy paper "Extended Cognition and Functionalism".  (I haven't read Sprevak's article in probably two or three years.)

In order to fly, EM needs to embrace a key feature supported by functionalist theorizing namely multiple realizability. A little philosophical history will help here. Functionalism (in its non-extended form) freed physicalist philosophy of mind from a kind of neural chauvinism. If our mental states were constituted by their functional roles, and the material contribution of our brains was merely implementationa I in character, then robots, Martians, Klingons and gaseous creatures from the outer limits of the universe could all join us in having mental states, just so long as the physical stuff out of which they were made could implement the right functional profiles. Stretching the word' skin' to include boundaries made of tin and gas, traditional functionalism bequeathed to the mind what we might call within-the-skin multiple realizability. And within-the-skin multiple realizability requires within-the-skin implementational materiality. But now extended functionalism merely plays out the same logic beyond the skin. If the specific materiality of the substrate doesn't matter to cognition, outside of the fact that it must be able to support the required functional profile, then what, in principle, is there to stop things-beyond-the-skin counting as proper parts of a cognitive architecture? Nothing, that's what. And this beyond-the-skin species of multiple realizability, which is just another way of characterizing the core philosophical commitment of EM, requires beyondthe-skin implementational materiality. If we look at things this way, the really radical and revolutionary movement was functionalism, not EM. EM simply makes manifest one of the implications of functionalism.  (Wheeler, 2010, 33).

Now, I think one has to be careful here.  One has to draw a distinction between a modal EC claim and a non-modal EC claim (as Wheeler now does in his Extended X manuscript).  The modal claim, essentially, is that it is (logically? nomologically? metaphysically?) possible for cognition to extend.  The non-modal claim, essentially, is that cognition (actually) extends.  Functionalism makes the modal claim at least pretty plausible, even though one can raise objections about the nomological and metaphysical cases.  But, functionalism, it seems to me, does not warrant the claim that cognition is (actually) extended.  Functionalism allows for multiple realization; it does not lead to (guarantee?) cases of multiple realization.  And, even if functionalism did entail actual multiple realization, it would not thereby entail actual multiple realization outside of brain.  Maybe it would only entail MR in brains.

As an aside, Carl Gillett and I have argued in "Levels, Individual Variation and Massive Multiple Realization in Neurobiology", for example, that many cognitive processes are actually multiply realized (or more technically, multiply implemented) in the brain at many different levels, but I don't see that cognitive processes are realized very often, if at all, out side the brain.

All of this is pretty rough--not what I would put in a journal article, but I think the lay of the land is clear enough for a blog post.

Are Wheeler and Clark Closet Cognitivists?

It sound like that here:

So what sort of overarching theory of the cognitive is favoured by EM theorists? As Clark (2008, 44) notes, the fact is that '[a]rguments in favour of [EM] appeal mainly, if not exclusively, to the computational role played by certain kinds of non-neural events and processes in online problem-solving'. In other words, EM theorists overwhelmingly conceive of cognition as a matter of information processing. Their distinctive observation is that, given this view of what cognition is, extra-neural factors - including the stuff of material culture - may, in some cases anyway realize the target phenomenon just as readily as neural tissue. For example, taking it that memory is, at least in part, a matter of the selective storage and context-sensitive retrieval of information, the EM theorist with a cognitive-archaeological bent might contend that information that is poised appropriately for context-sensitive retrieval may be stored in a Mycenaean Linear B tablet just a readily as in a Mycenaean brain.

Extended functionalism and implementational materiality 

Of course, not any old kind of information-processing profile will do here. To say that it would would be to fall prey to Rupert's worry about explanatory inefficacy. No, genuine cognition will be found only in a (perhaps rather small) subset of information-processing systems. (Wheeler, 2010, p. 32).

 To me, computational information processing of a restricted sort sounds a lot like the manipulation of representations, or a description of cognitivism sans the optional non-derived content condition.  This seems to me to reinforce the idea that cognitivism per se does not beg the question against EC, even though many advocates of EC and enactivism reject cognitivism.  Wheeler later drops the computational part for an extended "generic" functionalism, but that is still consistent with a thin sort of cognitivism that Adams and Aizawa have been urging since "The Bounds of Cognition" in 2001.

Wheeler on "Generic Memory"

So, what of this "coarse grained", supposedly locationally-neutral theory of memory?  Wheeler writes,

Rupert sees this sort of response coming, and so develops his memory-oriented critique further by arguing that any attempt to fix a generic kind that would subsume internal and extended systems would need to be so devoid of detail (in order to subsume all the different profiles) that it would fail to earn its explanatory keep. But this seems wrong. Indeed, it's important to note that we would surely not intuitively withdraw the epithet 'memory' from an internally located system which did not exhibit the generation effect, but which continued to achieve (something like) the selective storage and context-sensitive retrieval of information, so why should we withdraw that epithet from an extended system with a similar profile? But if that's right, why think that exhibiting the generation effect is a defining dimension of memory, rather than an accidental feature? This gives us some reason to think that there must be a generic account of what memory is that covers both cases, and that has explanatory bite. (Wheeler, 2010, p. 31).

Again, it seems to me that there is a lot going on.  Adams and Aizawa have challenged the "generic mental categories move" from as early as our first "Bounds of Cognition" paper.  We note in Bounds how the generic category of memory has decreased in significance in psychology, being replaced by more specific types of memory, such as procedural and declarative memory.  This category is losing its "explanatory bite" even though the term hangs on.  Wheeler is, I think, right, to claim that "we would surely not intuitively withdraw the epithet 'memory' from an internally located system which did not exhibit the generation effect".  But, we also don't withdraw the epithet 'memory' from the hardware in a standard personal computer, because it does not exhibit the generation effect.  But, the generic category of memory that indifferently includes both the computer hardware and the brain's wetware does not seem to be a single scientifically interesting kind.

To return to the case mentioned by Justin Fisher, one might think that there is a category of flying that consists of things the move through the atmosphere.  And, surely we want to say that insects, birds, jet planes, helicopters, and rockets fly.  We would not want to withhold the "flies" epithet from them.  Nevertheless, the loose category of things that move through the atmosphere is of pretty limited interest to the entomologist studying insect flight, the ornithologist studying bird flight, the aeronautical engineer, and the rocket scientist.

The Benchmarks of Cognitive Parity

So, here is where things get more serious.

what are the benchmarks by which parity of causal contribution is to be judged? Here is the wrong way to answer this question. First we fix the benchmarks for what it is to count as a proper part of a cognitive system by identifying all the details of the causal contribution made by (say) the brain. Then we look to see if any external elements meet those benchmarks. Why is this the wrong way to go? Because it opens the door to the following style of anti-EM argument: we identify some features of, say, internal memory that are not shared by external memory, and we conclude that since the parity principle is not satisfied, EM is false. (Wheeler, 2010, pp. 30-31).

There actually seems to me to be a lot going on here.

First, no one (Rupert, Adams, or Aizawa, in particular) want to fix the benchmarks by identifying all the details of the causal contribution made by the brain.  At most, Rupert, Adams, and Aizawa want to fix the benchmarks by identifying all the cognitive details of the causal contribution made by the brain.

Second, it seems to me perfectly fine to ask, "If one were to fix the benchmarks by reference to the cognitive details of the brain, then would we find these same cognitive details realized in brain-body-world combinations?"  The answer might be "no", but that does not mean that there is anything wrong with the question.

Third, bear in mind that, on the assumption that these fine causal details are multiply realizable, then it would not beg the question against EC.  If the details are multiply realizable, then it is in principle possible they are realized outside the brain.  Fineness of detail per se does not cheat the advocate of EC.

Fourth, and here is where Wheeler ultimately gets the conclusion he wants.  He can say that, even though Rupert, et al., are right that the fine causal contribution is not extended (what a preposterous idea!), there is this other kind of causal contribution -- call it "coarse causal contribution"-- that is extended.

Yet, it is misleading to suggest that Rupert, et al. are asking "the wrong question".  They are not begging the question.  They are not asking an absurd question.  The only thing is that it does not lead to the conclusion that Wheeler wants, namely, that there is some extended cognition.

So, I disagree with Wheeler's subsequent analysis:

To be clear, the Rupert-style argument under consideration is not suspect in virtue of being anti-EM, rather it is suspect because it begs the question against EM by assuming that II what counts as cognitive should be fixed by the fine grained profile of the inner. Such question-begging can be avoided, and Rupert's criticism resisted, if we adopt the following alternative strategy for saying what the benchmarks are by which parity of causal contribution is to be judged. First we give an account of what it is to be a proper part of a cognitive system that is essentially independent of where a candidate element happens to be located with respect to the internal-external boundary (however that boundary is to be determined). Then we look to see where cognition falls - in the brain, in the non-neural body, in the environment, or, as the EM theorist predicts will sometimes be the case, in a system that extends across all of these aspects of the world. (ibid., p. 31).

But, Rupert, et al., are not begging the question on fine grained details.  They are just asking a question to which the consensus answer now seems to be, "That kind of cognition does not extend".  Moreover, by "extracting" one's benchmarks for cognitive parity from the brain exemplar does not prejudice the location of other realizations of that exemplar.

Parity of Causal Contribution?

So, on t0 some of the substance of Wheeler's "Minds, Things, and Materiality".

In broad terms the parity principle states that if there is equality with respect to governing behaviour, between the causal contribution of certain internal elements and the causal contribution of certain external elements, then there is no good reason to count the internal elements concerned as proper parts of the cognitive system while denying that status to the external elements concerned. Parity of causal contribution mandates parity of cognitive status.  (Wheeler, 2010, p. 30).

Now, as I read this, it is somewhat ambiguous.  "Equality with respect to governing behavior" can mean

a) Equal insofar as two things (processes?) both govern behavior.
b) Equal insofar as two things (processes?) both govern behavior in the same way.

Wheeler, et al., can advance whatever theory of equality they wish, but I think Wheeler intends the second reading.  So, then, the issue becomes what is meant by "in the same way".  That's what the very current debates regarding "coarseness of grain" are all (mostly?) about.  More on that tomorrow....

Enactivism vs EC

As I begin to work through Mike Wheeler's "Minds, Things, and Materiality," I'm struck by how much I have underestimated the extent to which the advocates of EC are working through the differences between EC and enactivism.  (One of Wheeler's ideas is to draw attention to the EC and enactivist themes in a paper by Malafouris.)

For much of the last ten years, the embodied, embedded, enactive, extended (EEEE) crowd has presented something of a unified front.  I knew that Noe (and Rowlands) think that consciousness extends, where Clark does not.  And, I have had this (unpublished) paper indicating rifts along representationalist lines.  And, I have been thinking about this Complementary EC/Revolutionary EC distinction for a few months.  But, it is now clear that some of the major players in the EC movement are looking at the differences between EC and enactivism.  There is the special issue of Topoi, edited by Julian Kiverstein and Andy Clark.  There is this paper by Wheeler that is now out.  There is some discussion along these lines in Thompson's Mind in Life.  There are probably other examples as well.

So, I figure there is going to be a rash of papers now articulating the differences.  This, I think, is a good thing.  I find that there is a tendency among some folks in the EEEE crowd to be somewhat dismissive of my criticisms from the perspective of orthodoxy.  But, differences among the heterodox (maybe) cannot be so easily dismissed.