A blog about problems in the field of psychology and attempts to fix them.

Tuesday, September 27, 2011

Beyond the Brain: Embodied Minds

Continuing coverage of Beyond the Brain, by Loise Barrett. The book presents much of its material in a way that will push the average reader's comfort zone, but not smash their cherished ideas head on. This is a reasonable stylistic choice and, overall, a great strength. The next two posts will talk about the book's presentation of embodied cognition, hopefully giving fair balance to the places where the book did a solid job, and where it could have hit a little harder.


"Embodied" has become a bit of a buzz word recently. Most people who use the term seem well intentioned, but seem to have little clue about the full implications of their thoughts. Those who use the term lightly seem merely to mean that they do not believe in a non-corporeal soul or mind-independent-of-the-body. What they do not seem to understand (and what I mean when I say they use the term 'lightly') are the radical implications of such statements for reshaping psychology.

That is, they want to keep using the frameworks they inherited from people with firmly dualistic ideas: Keep the old terms, keep the old problems, keep the old methods and interpretations. It is Orwellian double-think, pure and simple. These people manage to hold at the same time the following contradictory thoughts: 1) It is silly to think about the mind as separate from the body. 2) The mind is best understood and investigated using terms that continue the logic of metaphysical dualism.

Those who use the term 'embodied' in a heavy-weight manner understand that full acceptance of the term requires that much standard thinking about psychology be revised. One tension amongst those who embrace embodied control of action is whether there is room to talk about the mind at all in an embodied system. That is, does a commitment to embodiment require the denial of all things mental, or can we talk sensibly about an embodied mind?  Barrett does not explicitly discuss this controversy, and she splits the difference between the two positions... which is clearly the correct decision given the tenor of her book.

Embodied Not Minds

In many parts of the book Barrett seems to be using the notion of embodiment to replace the notion of minds. A few examples of this include:

Discussion of Grey Walter's modified "turtle" robots. The robots had a very small number of behavioral options, but performed several interesting (not directly programed) behaviors, such as moving objects in their environment into piles. (The story about how they feed themselves is the coolest part, but you can get that if you read the book.) The robots had lights on their heads that could turn on or off, and they could respond to lights, including of course the lights of other robots. Just for kicks, the experimenters decided to put a mirror in enclosure. Details aside, the robot seemed to dance in front of the mirror, and Walter noted that such behavior "on a purely empirical basis, if it were observed in an animal, might be accepted as evidence of some degree of self-awareness." (p. 46) One moral of the story, according to Barrett, is that behavioral complexity should not be mistaken for mind-stuff:
Grey Walter's robots highlight perfectly how we cannot simply translate behavioral complexity into an assessment of an animal's cognitive complexity. (p. 47)
complex behavior can be produced, not only by a very simple mechanism, but also by a mechanism that bears absolutely no relation to the behavioral outcome produced when that mechanism operates in the real world. (p. 49)
 The the latter quote is taken slightly out of context, it applies here as well. The point is that nothing inside the turtle will resemble a self-recognition mechanism, a pushing things into piles mechanism, or a get yourself more power mechanism, yet all these behaviors, or reasonable facsimiles thereof, are enacted by the turtles.

There is also the excellent example of how crickets "choose" mates. Note that common wisdom would have you believe that there is a cacophony of woodland sounds through some sort of cognitive signal vs. noise "information processing" system, and then the cricket would have to weigh the available options, again through some sort of cognitive, information-processing, decision system. Quite to the contrary. The cricket has eardrums on its legs, and a series of interneurons that have a slow decay from activation back to resting threshold. The male song is carefully attuned to this decay in a way that steers the females towards them. The female's auditory and nervous system is designed to respond by orienting towards whichever eardrum first affected by a male's chirp. One moral is that:
females are not "picking out" and then tracking a male's song against a background of other kinds of songs and noise; rather, they simply don't perceive the background. (p. 52)
That is, the physical form of the cricket does much of the work we might intuitively suspect required involvement of a complex nervous system. And from this:
The interaction of the cricket's body, its "brain", and the environment may allow females to solve all the apparently complex problems of mate finding by... [repeatedly following the rule] turn to the side that fires first. (p. 53)
This almost seems to simple to be true. But Barrett smoothly takes us back into the world robotics, where we meet Webb's cricket robot, which shows that it really can be that simple. Implication:
Orienting toward individual sound pulses isn't what one would expect if females were analyzing the entire pattern of song and then choosing the best one; instead choice emerges from the kind of simple auditory steering process Webb used in her robot. In other words, these results again reinforce the idea that the "steering" mechanism and the "picking-out" mechanism are one and the same thing. It is the rhythm and temporal pattern of the male song that simultaneously "steers" the female to its source, and "discriminates" the male's chirps from the background.
Note the scare quotes. They imply, or at least so I read it, that we should be suspicious of the claim that the cricket is "picking-out" anything, or "discriminating" anything. Instead, like the first robot's seeming self-recognition, the cricket is doing things that might mislead us to ascribe such mental processes.

This comes up again in one of the big morals from the chapter dedicated to the still ridiculously-fascinating example of the Portia spider's predatory behavior. Given a complex enough situation, the spider will scan its surroundings for some time before moving off to capture its prey, and the spider avoids potential routes with several types of problems. Moral:
Far from demonstrating that spiders plan a detour route in an insightful way, these detailed and careful experiments reveal that... It is this set of simple mechanismsnot insightthat allows them to reach the prey... (p. 70).

Barrett will have have a few passages that suggest these earlier examples were intended to be about cognition proper. She asserts, for example, that: "cognition need not be—either by definition or by logical inference—a purely computational process." (p. 129) However, in context, these and several other examples seem to suggest that the result research that uncovers more and more "embodiment" is that we ascribe to animals less and less "mind".

Embodied Minds

However, Barrett does not banish all talk of minds, and is not (when the book is taken as a whole) making the crude behaviorists claim that mental phenomena do not exist. Barrett notes positively, for example, Gibson's argument that "psychological phenomena are not things that happen 'inside' the animals, but are found in the relations between animals and their environments." (p. 95) I'll cover Barrett's discussion of ecological psychology later, but for now it is enough to note that she clearly believes there is a thing called "perception" that animals do, and that it can be properly described as "mental", even under a firm commitment to embodiment.

Mental terms become pervasive in the discussions of extended cognition that book concludes with. Quoting Clark and Chambers Barrett tells us that:
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 accepting as part of the cognitive process, then that part of the world is (for that time) part of the cognitive process. (p. 198)
This takes the notion of embodiment and extends it dramatically. Gibson's message was something like: If you use your toes to help peak up to a high shelf, then your toes are part of your perceptual system. Clark's message can be phrased with sane seeming examples: When you make a grocery list, the physical list is part of your memory. Or it can be phrased in ways that seem more bizarre: If the roll of the cobblestones determines the path you take when walking down the street, then the roll of the cobblestones is a part of your choosing where to go.

Most aggressively, perhaps, Barrett tells us that "acting simply is a form of thinking." (p. 143) This stuck out to me because I am typically tempted to phrase it the other way around, namely that "thinking" is simply a form of "acting", though of course an equivalence relationship can be expressed either way. 

In any case, in these sections it is clear Barrett is arguing that evidence for embodiment or externalization does not require us to abandon talk of minds and mental processes.

Brief Digression - Embodiment vs. PSYCHOLOGY!

One of the nice things about Barrett's book is that it is clearly about psychology. Chemero's "Radical Embodied Cognitive Science" definitely hits the readers with the more extreme embodied-cognition view at full force, making no pretension that it will be an easy introduction to these issues. Most chapters in his book have several examples illustrating the possibilities of embodied behavioral control. In almost all chapters, he provided a final example that I would refer to as the "Little Albert" example. Recall that when Watson performed this experiment much was already known about the power of Pavlovian conditioning, but many psychologists still didn't care about it, nor any other animal work. That was because it was not obvious how a drooling dog was relevant to the problems of 'PSYCHOLOGY!' (imagine a 1950's radio voice saying "SCIENCE!", with a cymbal crash and slowly moving jazz hands). Watson new it was crucial to the advancement of behaviorism that he changed that attitude, and so he decided to create and cure a phobia, because 'phobias' are blatantly 'PSYCHOLOGY!', and curing phobias is something that would be undeniably useful to 'PSYCHOLOGY!'. Similarly, Chemero would give several examples of animal behavior or mechanical control that solidly proved his points, but that were not clearly psychology, and then would try to end on an example that would convince the reader Chemero was employed in the correct department and that he wasn't crazy to put the word "cognitive" in his book's title. This strategy worked well, in my opinion, and most professions (the target audience) will agree. However, my students really needed me to provide the Little Albert metaphor before they understood what was going on. Because Barrett's book adopts a less extreme presentational style, and talks more fluidly about mental processes, I never found myself wondering "How is this going to connect to psychology?" This is true even in the sections in which embodiment seemed anti-mental.

Next Time

How Beyond the Brain could have pushed the embodiment argument a bit further, i.e. why Barrett needs Holt.


  1. My copy just arrived and I'm going to start it tonight. Based on this, I'm going to like what I read :)

    Barrett's use of 'embodiment' to replace 'mind' or 'computation' seems bang on; she sounds like she actually has the interesting bit about embodiment the right way round, unlike some people I could mention.

    I have mixed feelings about her trying to reclaim, rather than abandon, words like 'mind'. While I firmly believe 'representation' has to go because it's more trouble than it's worth, I haven't quite got there with 'mind' yet. It may be that while we are in this transitional period, it's worth keeping a few terms and simply redefining them. It works with 'mind' and 'cognition' because their definitions are so vague (especially cognition).

    Anyway, I like what I hear about this so far!

  2. Eric -

    A very timely post for me given my current pursuits, specifically some Davidson essays collected under the title "Sub-, Inter-, Ob-jectivity" (SIO). SIO has been as helpful as anything I've encountered in trying to get a grip on "mind". But I too keep wondering if the use of that term couldn't be eliminated with no loss, perhaps a gain, in net clarity. Or at least used with more discretion in venues other than the popular press.

    Which leads to a question: has anyone ever written on the behavior we refer to as "mental" from a purely action-oriented POV? Not ala strict behaviorism (which I rightly or wrongly understand to have assumed that "all is revealed" by behavior) but trying to describe everything in terms of the physiology of action? I get the impression that Gibson went some way in that direction but don't know how far.

    Love the cricket and spider examples, since over-ascription of "cognitive" abilities is becoming a pet peeve. However, "discriminating" - in the cricket context - seems OK since it's a standard term in signal detection. Barrett apparently wants to avoid the idea of extracting the chirp from background noise, but that seems the essence of the "attuned" leg-borne eardrums. In fact, in radar there is something (not coincidentally) called a "chirp filter" that is matched to a chirp-like radar signal.

    Re grocery lists. The general concept of extended memory seems pretty tricky. One making a grocery list presumably "knows" what's on it in the Sellars sense of being able not only to produce the list to help remember what to get, but also to "enter the space of asking for and giving reasons" (ie, justify each entry if asked).

    But as I understand it, the infamous Otto can't do the latter, so he doesn't "know" what's on it; he just reads it or hands it to a clerk. Ie, he only has access to the information on the list. And if access alone is sufficient for extended cognition (ie, no "knowledge" ala Sellars required), it seems that anyone with a handheld device that provides access to the Internet has almost everything currently known to humankind in "extended memory" - a seemingly unreasonable conclusion.

    With that understanding of extended cognition, using the "roll of the cobblestones" (not sure what that means) seems to qualify if the walker can justify it, eg, give reasons why it works. Although as usual, it's unclear to me what treating that as extended cognition buys us.

    Davidson claims that thought requires language but that not all thoughts need be expressible in language; in which case, at least some thoughts get expressed in language, an action. Whether or not every action expresses a thought seems an open question. (I'm assuming "action" implies intent, which eliminates as obvious counterexamples unintentional bodily movements). So, I'd guess that saying "some thoughts result in action" is as strong as is justifiable. No?

  3. Charles,
    Stay tuned for the action-oriented POV! It might not be exactly what your looking for, but that is the direction the next post is going in.

    As for the cricket, recall that in an earlier post I suggested that psychologists should not be allowed to use the word "discrimination" without also saying the behavior that discriminated (http://fixingpsychology.blogspot.com/2011/08/myth-of-knowledge.html). I am fine with the claim that "the female cricket's orienting behavior" discriminates the sound of a male cricket from other sounds. However, we should be careful not to confuse that behavioral achievement with any implications of intellectual knowledge or understanding - which is what most people hear if you simply say "the female cricket discriminates the male's signal from the other noises."

    The extended cognition stuff is a bit odd. Like you, I find certain aspects appealing, but think it can easily be taken too far. What makes it seem most reasonable, to me, is that I often find that I have trouble remembering things after writing them down. It is as if I am (physiologically) assuming that the list will serve as memory, and therefore (physiologically) change to a state where I am less able to remember without the list.

    As for the roll of the cobblestones, I was trying to point out that unlike modern roads and sidewalks, stone roads were a very bumpy surface. If one stepped on the edge of a stone, and their foot rolled outward, it would be natural for the whole body to drift outward. However, as true as that is, it would be very odd for an observer to claim that the curve of the stone's surface was part of our "decision" to alter course.

    As for the last point, I would say that distinguishing thoughts from actions should be suspect (and that I think Barrett and I are in agreement on that). The claim that "some thoughts result in action" is, by my account justifiable because indeed: Some actions do result in other actions. However, it seems Davidson (and most readers) would want some sort of dichotomy between thought and action, which makes me... nervous.

  4. And if access alone is sufficient for extended cognition (ie, no "knowledge" ala Sellars required), it seems that anyone with a handheld device that provides access to the Internet has almost everything currently known to humankind in "extended memory" - a seemingly unreasonable conclusion.
    Access is more than having the device. Google's key insight into the web was that content wasn't the issue, but finding content was. Microsoft, etc, went after ways to make it easy to add content, Google went after search, and almost everything they do is centred on search technology. This is why they are successful, because they recognised the real issue.

    Given my iPhone, an internet connection and a little time, I do, indeed, know everything on Wikipedia :)

  5. Eric -

    I'm actually with you and Barrett on thoughts and actions. Saying that it is an "open question" was my way of hedging. Also, don't infer Davidson's position based on what I say. He seems a bit vague on the issue in the essays I'm reading (perhaps he was hedging as well). They date to the '80s, so his later position may have been different or at least more explicit. FWIW, Rorty seems to be an "action" man, and he credited Davidson for much of his relevant insight. So, it's quite possible Davidson was one as well.

    Andrew -

    Given our respective backgrounds in relevant areas, there are few issues on which I'd be foolish enough to stand my ground in opposition to you. This is one.

    I agree with Eric's early post on the care with which "know" and its cognates should be used. I try always to make it clear that I'm using "know" in the Sellars sense. Of course, that doesn't mean it's the preferred sense or that I'm applying it correctly, and I'm always open to being convinced that it isn't or that I'm not (a quite likely possibility). But contrary statements that neither correct my understanding of Sellars nor offer an alternative interpretation of "know" seem to me empty.

    Even using my interpretation of the Sellars sense of "know", I see uncertainty with respect to online-accessible (or readily retrievable, if you prefer) material. If I claim to "know" that B. F. Skinner was born in 1904 and show someone the relevant entry, few would challenge my claim because most would assume that such "raw facts" are not in dispute. But if someone claims to "know" that Grant was a chronic drunkard and shows me that statement in the relevant wiki entry, I will challenge their claim. (I have no idea whether the entry says that or not.) I would judge presence in a wiki entry to be insufficient justification for the implicit claim that one side of a highly controversial issue is "true".

    Of course, if cognition doesn't require knowing (in some sense), all this is irrelevant. But in that case, I'd be even more inclined to ask "why such a fuss?" We're all familiar with the proper use of stored information whatever the medium, convenience of access, and effectiveness of retrieval. What pressing issue is resolved by considering some of that material to be a component of "embodied minds"?

  6. Note: in my examples, both "entries" were intended to be from wiki. My point wasn't to question the general credibility of wiki but to emphasize that credibility does become an issue for controversial questions. Ie, more is required to "know" than retrievability.

  7. Super quick note: I was surprised to learn, not to long ago, that cognition used to be synonymous with knowing. That is, cognition did not 'require' knowing, it was 'knowing.'

    Nowadays, philosophers have warped questions of 'knowledge' to cover a weirdly redistricted range of topics. Thus, the overlap is unclear.

  8. Truth is a separate issue; my brain has plenty of incorrect things lurking in it too. I'm just suggesting that your argument against Clark's 'reliable access' criterion for extended cognition isn't as clear as you suggested; having the iPhone is necessary, but not sufficient, you still need the ability to retrieve the information.

  9. "Truth is a separate issue"

    Agreed, but I didn't address it. I said that implicit in a claim to "know P" is the claim "P is true". As you suggest, whether it actually is "true" (in whatever sense one prefers) depends on unstated factors.

    I prefer the (quite unpopular) version that says "truth" is the result of discussion within an appropriate community, with which Sellars' "asking for and giving reasons" fits nicely. (Presumably the genesis of Rorty's infamous "truth is what your peers let you get away with saying".)

    It appears that our initial disconnect was simply "failure to communicate" - I intended "access" to include "retrievability". It's unclear to me what the former would mean without the latter. But if the distinction is a matter of ease or immediacy, I see a two edged sword. It seems clear to me that ease of acquisition and "knowledge" aren't necessarily compatible.