Clark Review of Adams and Aizawa Bounds of Cognition and Rupert Cognitive Systems and the Extended Mind

The published version is now available here.

Clark @ University of Tokyo

Schedule of talks here.

Video of Andy at Renssellaer

Link here.

Here's a Nice Bit of Irony

From Shaun Gallagher's forthcoming "The Overextended Mind":

If, as we confront some task, a part of the world functions as a
process which, were it to go on 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. (Clark and Chalmers 1998, p. 8)

On a strict interpretation this principle appears to measure cognition in
terms of the Cartesian gold standard of what goes on in the head. It suggests
that a process outside of the head counts as cognitive only if in principle it
could be accomplished in the head (Gallagher, forthcoming, p. 1).

The irony, of course, is that Clark, and others, have taken to charging A&A with brain-o-centric bias.

Clark (2008, p. 114) rejects this interpretation, insisting that the
parity principle should not be interpreted as requiring any similarity
between inner and outer processes. Wheeler (2006, 3) explains that the
parity principle does not “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 [and then by looking] to see if any
external elements meet those benchmarks.”  (Gallagher, forthcoming, p. 2)

And, it's good to see them on the defensive on this.

Andy Clark at Rensselaer

More info here.

Call for Papers Special Issue of *Philosophical Explorations* on "Extended Cognition and Epistemic Action"

Guest Editors: Andy Clark (University of Edinburgh), Duncan Pritchard (University of Edinburgh), Krist Vaesen (Eindhoven University of Technology)

Submission Deadline: September 15, 2011

Invited Contributors: Fred Adams (University of Delaware) & Ken Aizawa (Centenary College of Louisiana), Ronald Giere (University of Minnesota), Sanford Goldberg (Northwestern University), Richard Menary (University of Wollongong) and Kim Sterelny (Australian National University and Victoria University).

Background and Aim

According to the thesis of extended cognition, cognitive processes do not need to be located inside the skin of the cognizing agent. Humans routinely engage their wider artifactual environment to extend the capacities of their naked brain. They often rely so much on external aids (notebooks, watches, smartphones) that the latter become a proper part of a hybrid (human-artifact) cognitive system.

The thesis of extended cognition has been influential in the philosophy of mind, cognitive science, linguistics, informatics, and ethics, but, surprisingly, not in epistemology. The discipline concerned with one of the most remarkable products of human cognition, viz. knowledge, has largely ignored the suggestion that her main object of study might be produced by processes outside the human skin.

In this special issue of *Philosophical Explorations* we therefore are looking for papers that explore the ramifications of the thesis of extended cognition for contemporary epistemology in general, and for conceptualizations of epistemic action in particular. The special issue will include five invited papers (by Fred Adams & Kenneth Aizawa, Ronald Giere, Sanford Goldberg, Richard Menary and Kim Sterelny), plus two contributions selected from the papers submitted in response to this open call for papers.

We expect contributions discussing the impact of extended cognition on issues as: epistemic agency and responsibility, cognitive ability, ownership of belief, the distribution of epistemic credit, the sources of belief, artifactual testimony, the growth of knowledge, non-propositional knowledge, the evolution and reliability of extended cognitive processes, the varieties of extended epistemic action.

Submission Details

Please send a pdf-version of your paper (max. 8000 words) to Krist Vaesen. Contributions that do not make it to the special issue may be considered for publication in one of the regular issues of *Philosophical Explorations*.

Further Inquiries

Please direct any inquiries about this call for papers to Krist Vaesen.

Thinking in English and Thinking in Pictures

I will now take it that Adams and Aizawa ought to concede that there is no difference in content between my thought that ‘the harbour Bridge looks beautiful in the sunlight this morning’ and my utterance to you that ‘the Harbour Bridge looks beautiful in the sunlight this morning.’ Indeed I might think that very sentence to myself in my head before uttering it to you. If Adams and Aizawa are happy to accept this conclusion, then we really have no disagreement, because we have an example of thinking in natural language (English in this instance). Similarly it seems obvious to me that a Venn diagram that I am imagining now has the same meaning as the Venn diagram that I am drawing on the page now.

Menary seems to me not to respect the difference between having a thought that P and have an English sentence with the content that P in one's head.  Sure, one can have a thought with the content that the harbour Bridge looks beautiful in the sunlight this morning, and one can even "think this in English", but that's not to admit that there is an English sentence in the head that has the content "that the harbour Bridge looks beautiful in the sunlight this morning".

So, I now think that maybe A&A indulged the Clark and Menary discussion of these cases a bit too much.  Spending a lot of time explaining their view may have obscured our view, which I think is pretty simple. 

1) All thought is in mentalese (a system of mental representation that differs from all natural languages) which gets its semantic content by way of satisfying conditions on non-derived representations. 

2) Thinking in English is forming mentalese representations of English sentences; it is not a matter of having tokens of English sentences in the head.

3) Thinking in images is forming mentalese representations of pictures; it is not a matter of having tokens of pictures in the head.

These are just statements of our view.  There is a huge literature on mental imagery and a large literature on thinking in a natural language, so I would think it would take a lot of work on Menary's part to show that we are wrong on any of 1)-3).  And, he is trying to show, or maybe is just assuming, that we are wrong on 1)-3).

Ramsey's take on "The Notebook itself is not cognitive ..."

For example, Clark (along with Menary) criticizes Adams and Aizawa for confusedly thinking that the extended mind position entails that objects (like Otto's notebook) can be intrinsically cognitive, irrespective of whatever role they are playing. As Clark points out, no proponent of EMH thinks that; it is essential that the object be properly conjoined or integrated with a brain. Yet, in fact, this is an uncharitable interpretation of Adams and Aizawa, whose whole argument is designed to show that integration with a brain is not enough.

Go, Bill!

Pardon my self-indulgence as I enjoy having someone agree with me on EC for a change.

Menary on Weak and Strong Cognitivism 3

Weak cognitivism: cognition involves the processing of representations.
Strong cognitivism: all cognition involves the processing of representations with underived content.

... The weak version is consistent with most of the examples of integrated cognition that I present in Dimensions of Mind (Menary 2010). It only conflicts with examples that do not involve the processing of representations. In fact, the claim that Clark and Menary argue against cognitivism is somewhat strange since both of them allow that representations play an important role in cognitive explanations, they just don’t believe that all cognition involves the processing of representations. So their positions are quite consistent with a weak cognitivism, that cognition often involves the processing of representations, but not always.

Now, I don't propose to make much of the issue what to call a view, but the A&A view is that cognition is a matter of specific sorts of manipulations of vehicles bearing non-derived content. So, there is more to cognitivism than just representations.  [Coincidentally, I just mentioned this in my post on Ramsey for this morning.]

Clark often does talk about information processing--which we can take to be the same as manipulation of representations--but Clark does not, to my knowledge, in the EC literature mention any constraints on the types of information processing (representation manipulation) that might count as cognitive.   I could be wrong.

I do have the vague idea that at some point in his career, Andy did claim that symbol manipulation by lookup table would not be cognitive information processing, and I also have the idea that this was related to some exchange with Dennett.  But, I don't know what paper this would be in.  References would be much appreciated as I have been wondering about this literally for years.

Clark in the NYT Again

They can't get enough of the man! 

Andy Clark in the NYT

Here.

HT to Leslie Marsh.

Who Supports Revolutionary EC 9?

“It is our biological nature (as I argue at length in Clark 2003) to be open to many forms of physical and cognitive hybridization. Some of these (I claim) may be so intimate as to effectively extend the thinking agent. All of them are crucial parts of the nested, iterated and ongoing process of cognitive self-re-creation that is the characteristic mark of human intelligence. It is important that we develop an understanding of ourselves (both scientific and philosophical) that is adequate to this open-ended process of physical and cognitive self-creation. To do so means questioning the notions of the mind and person as essentially biological, and recognizing the very large extent to which the commonplace identification of minds and persons with purely biological structures is itself what Locke (1694) termed a ‘forensic matter’: a matter of legal and moral convenience more than metaphysics, and a convenience, moreover, that must become increasingly inconvenient as science and technology progress. (Clark, 2005, pp. 9-10).”

Who Supports Revolutionary EC 8?

So, Sutton, et al. have (probably) never denied that cognitive processes take place in the brain:

“We are entirely happy to treat study of ‘the kinds of processes that take place in the brain’ as scientifically valid, and to accept intracranial cognition: we have never argued otherwise, and nor to our knowledge has Clark.”

But, there is a fairly robust group of 4EA type philosophers who have.  This includes Haugeland, Hurley, Menary, Rowlands, and Thompson.  In particular, Clark has, at least at times, rejected intracranial cognition.  Here is one point.

“I’d encountered the idea that we were all cyborgs once or twice before, but usually in writings on gender or in postmodernist (or post postmodernist) studies of text. What struck me in July 1997 was that this kind of story was the literal and scientific truth. The human mind, if it is to be the physical organ of human reason, simply cannot be seen as bound and restricted by the biological skinbag. In fact, it has never been thus restricted and bound, at least not since the first meaningful words were uttered on some ancestral plain. But this ancient seepage has been gathering momentum with the ad-vent of texts, PCs, co evolving software agents, and user-adaptive home and office devices. The mind is just less and less in the head.”  Clark, Natural Born Cyborgs, p. 4

Now, I know that in Supersizing Clark has been happy to talk about a "cognitive core",  but here is a case where he appears to deny this. It's probably possible to read this as not denying intracranial cognition (it's hard to really pin an unwanted view on a philosopher), but then again it is also plausible to read this as denying intracranial cognition. 

Vaesen and the State of Play in EC

These are thorny issues indeed, [This is not to say that they can’t be handled satisfactorily. Quite convincing [extended cognition] (counter)arguments are e.g., Clark (2007), Menary (2006) and Rowlands (2009).]

Just for the record, there are further replies to these, including some in The Bounds of Cognition,  Aizawa, (2010), and Adams and Aizawa (forthcoming).

 In truth, I don't see that Vaesen needs to have our critique rebutted in order to have his project go forward.  He doesn't endorse a kind of extended cognition that A&A challenge.  So, to put matters as I see them, even though there is essentially no extended scientific cognition out there, this is not a problem for Vaesen.

"The Value of Cognitivism ... " is now out

Here at Phenomenology and the Cognitive Sciences.

Right back at you, Richard!

Notes on Clark's Review 5

I find Andy's discussion of the neural loops and cognitive architecture very interesting and I think it's something I will want to write about more seriously in the future.

The issue that seems to me to merit more attention is the fact that coupling as discussed in this regard is evidently not the kind of coupling Andy is talking about in terms of "trust and glue".  Having corresponded briefly with Andy on this point, we've agreed that it's a point to "co-ponder".

As I said, I think Andy's review is both generous and helpful.  I think it is much more than a mere informational review, but a commentary that will likely move the discussion of EC forward. 

And, of course, it also discusses Rupert's book!

Notes on Clark's Review 4

it is often best to view the coupling considerations as a means not of directly establishing co-constitution so much as establishing ownership. The right kind of dense coupling (e.g. complex reciprocal causal exchange) is surely part of the explanation of why, for example, the right and the left neural hemispheres count – when the coupling takes just the right form – as two components of a larger processing system rather than as two isolated processors.

Here, Clark highlights an important idea.  Coupling considerations are often thought to establish ownership.

I.
Coupling, however, does not tell you the character of what you own.  Mere ownership of a thing does not tell you whether that thing is a mere tool or a cognitive processing unit.  The fact that Otto is coupled to his notebook might tell you that it is his notebook and part of his cognitive system, but it doesn't tell you whether cognitive processing is taking place in that part of the system.

To switch examples, causal coupling between the CPU and the monitor may tell you that the monitor is part of the computing system, but it doesn't tell you whether the monitor computes.  For that, one would need something like a "mark of the computational".

II. 
Even construed as ownership conditions, coupling is not entirely unproblematic.  On the one hand, Clark does not want the coupling conditions to be too lax, since one will then "own everything".  This would be a kind of "ownership bloat" problem on par with the more familiar "cognitive bloat" problem.  On the other, Clark does not want the coupling conditions to be too stringent, since one will then not "own" things that one apparently does "own".  In section 7.3.1 of The Bounds of Cognition, A&A in effect indicate how Clark's conditions imply a lack of ownership of what appears to be one's cognitive apparatus.  For example, Clark's conditions of "trust and glue" seem to imply that a person with blindsight does not own the apparatus that enables him to detect objects in his "blind" field.

Notes on Clark's Review 3

In my last post, I quoted Andy as saying

The issue concerning non-derived representations (Adams and Aizawa also speak here of non-derived contents) is complex (see Clark (2005) (2008) for discussion) but one point to notice is that such contents/representations are indeed present (assuming the idea is coherent, as I now think it is) in the putative overall cognizing system comprising bio-stuff and further resources.

After a number of exchanges on the topic of non-derived content, Andy now takes the idea to be coherent.  Note that Andy's view articulated above is a slight change from

First, though I shall not dwell upon this, it is unclear that there is any such thing as intrinsic content anyway. Second, in so far as the notion is intelligible at all, there is no reason to believe that external, non-biological structures are incapable of supporting such content. And third, even if they were incapable of so doing, this would not actually compromise the case for the extended mind.  The worry about intrinsic content, I conclude, is multiply fatally flawed. (Clark, 2005, p. 1).

Ironically, I must admit that I think there are challenges in providing a philosophically unproblematic account of the derived/non-derived distinction.  They have to do with cases of "derivation" that are similar to those to which Justin Fisher alludes in his review of The Bounds of Cognition. I think there is a derived/non-derived distinction, only as with so many other philosophical issues concerning characterization, there is a problem in getting the characterization exactly right.

Notes on Clark's Review 2

Although I think Andy's review is both generous and useful, I'd like to make a few more detailed comments on its contents.

Andy writes,

The issue concerning non-derived representations (Adams and Aizawa also speak here of non-derived contents) is complex (see Clark (2005) (2008) for discussion) but one point to notice is that such contents/representations are indeed present (assuming the idea is coherent, as I now think it is) in the putative overall cognizing system comprising bio-stuff and further resources. So the real question here concerns the acceptability of derived representations or contents as genuine elements in a cognitive process that quite clearly involves many non-derived ones too.

In first part of this comment, Andy has in mind the fact that when Otto uses his notebook, A&A will maintain that there are representations bearing non-derived content in Otto's brain, even if there are representations bearing only non-derived content in Otto's notebook.  So, requiring that cognitive processes involve non-derived content or non-derived representations won't do anything to show that Otto provides an instance of extended cognitive processing.

I think Andy's point here is well taken.  He draws attention to a weakness in the way A&A have formulated the non-derived content condition.  It is not enough to say that cognitive processing must involve non-derived content; it is not enough to say that non-derived content must be present in a cognitive process.  Something a bit stronger, or more explicitly stronger, appears to be needed.  So, I think the thing to say is that the cognitive vehicles of representation must bear non-derived content. The representational vehicles in Otto's notebook don't satisfy this condition.  Maybe there is some problem with this as well, but there you have it.

I've commented on this point before here.  It's also in print as footnote #10 in my paper, "The Coupling-Constitution Fallacy Revisited" in Cognitive Systems Research.

Notes on Clark's Review 1

Andy's review seems excellent to me.  In a short space it articulates some of the issues that have separated his views from the A&A views.  It tweaks some arguments that have already appeared in the literature.  Finally, it presents arguments that, I think, appear in print for the first time.  Here I am thinking of his discussion the role of causal loops in forming a cognitive architecture.  I saw him give a talk that included this last year at the Center for Interdisciplinary Studies at the University of Bielefeld.  There is in this review a lot that I think is not to be found even in his recent book Supersizing the Mind.