Tuesday, December 28, 2010

Menary and Sterelny are for HEMC?

Indeed, Sterelny argues that our cognitive capacities are deeply dependent upon the cognitive niche such that the cognitive capacities of the agent are transformed by interactions with the niche. It would be closer to the mark to talk of the niche ‘extending’ into the agent, in the sense that the agent acquires new  representations and new capacities by being ‘coupled’ to or scaffolded by the niche. For Sterelny, there is no cognitive agent without embedding in the cognitive niche.
This is consistent with HEMC:
Cognitive processes depend very heavily, in hitherto unexpected ways, on organismally external props and devices and on the structure of the external environment in which cognition takes place (Rupert, 2004, p. 393).
And, of course, interacting with the environment causes changes in the mind and brain.  That's what learning is all about.  Moreover, we talk about the impacts of environmental interaction in the development of visual capacities in Chapter 9 of The Bounds of Cognition.  It's common ground that agents are transformed by their interactions with their niche.

33 comments:

  1. Given that apparently everyone,including you, thinks that we are non-trivially coupled to the environment and that this is such a compulsory fact that we couldn't be what we are without it, why can't we abandon the idea that it's useful to talk as if cognition could possibly be the kind of thing that's bounded? These recent posts on how you aren't arguing what people say you are makes me more and more convinced that the entire concept of 'cognition' being 'bounded' makes no sense at all, even from your perspective. Coupling, constitution - these all seem, based on what people are apparently saying, to be entirely the wrong questions and no one seems to have noticed.

    ReplyDelete
  2. Well, lots of the EC folks want more than just causal coupling to the environment; they want cognitive processes to be constituted, in part, by bodily and environmental processes. This is the heart of the debate as Clark takes it up in his Supersizing the Mind .

    Maybe Menary does not want to take this up. His recent paper does not really seem as though he does, but his 2006 paper begins by (apparently) defending Clark against the A&A, 2001, paper.

    The coupling-constitution distinction may well not make any sense if you don't think that cognitive processes are a distinct kind from non-cognitive processes. But, the EC people think that they are distinct kinds.

    Here is why, I think, Menary and others can't help themselves from pointing out causal coupling as a path to EC: Causal couplings are ubiquitous (by everyone's account), so if they were good evidence for EC, then EC would be ubiquitous.

    The problem, of course, is that causal coupling does now warrent EC.

    To me, the issues make sense, only they are views and the arguments are wrong. This is in contrast to EP, where I am often baffled by what is being claimed.

    ReplyDelete
  3. Sellars and Davidson argue that knowledge acquisition (the latter specifically re learning a language) is a social practice involving the would-be acquirer, a community (for teaching, justifying, endorsing, et al), and "the world". Questions:

    1. Is knowledge acquisition included in "cognition" as that word is used in this forum?

    2. If so, is the process described above an example of what the term "EC" is intended to encompass?

    If not, why?

    3. If so, what benefit might an EC advocate consider accrues as a consequence of viewing the process as EC?

    If not, why?

    I've just finished Andy Clark's "Natural Born Cyborgs", and despite finding parts of it quite informative I don't see that significant benefits accrue to viewing - for example - Otto's notebook as a cognitive extension of him rather than as merely a tool. In fact, since a critical part of "knowledge" per Sellars and Davidson is justification by a relevant community (asking for and giving reasons), it appears that entries in the notebook may not meet their criteria for "knowledge".

    ReplyDelete
  4. I get that your argument is that coupling doesn't give you constitution. But your last few posts have gotten me thinking that the entire framing is off.

    Take the parity principle - if you see something performing a function that if, it was in the head, you'd call it cognitive, you should call it cognitive. This is just as incorrect as the idea that cognition is a set of processes which couple to the environment and should be treated as such. Both these claims seem to miss the point to me, namely that it just doesn't make sense to ever try and identify cognition with only one part of the brain-body-world system that implements "cognitive behaviour".

    Your recent posts have been talking about how A&A, like everyone else, are happy to accept that the brain has to be coupled to the brain and world; that this coupling is non-trivial, deep and meaningful, compulsory, etc. No one wants to say these pieces aren't needed, but no one seems to want to admit that it's actually the system that's cognitive (does anyone? Tony, I guess - anyone else?). The arguments all seem to be about which parts of the system can be cognitive.

    You're not going to agree with any of this :) It's on my mind as I work through Tony's book again and figure out what's right and what's wrong about that approach. He's right, at one level, about the importance of non-linear couplings. I think he's wrong in some of the specific fights he picks, though.

    I think this is why you don't get Gibson, either. The ecological approach is never quite framed in these terms, but it's an attempt to argue that cognition is the behaviour of an organism who perceives and acts via information and affordances; there aren't cognitive bits coupled to non-cognitive bits, that's just the wrong place to look. This is the root of my disapproval of fMRI too - even when you do it right it just can't tell you anything useful.

    ReplyDelete
  5. Both these claims seem to miss the point to me, namely that it just doesn't make sense to ever try and identify cognition with only one part of the brain-body-world system that implements "cognitive behaviour".

    Yes, I think this is really close to the core of the problem. The EC folks, at times at least, seem to have this kind of operational approach to cognition, it's whatever enables you to accomplish a task. Once you have that picture, then most of the rest of EC falls out easily. By contrast, A&A, like good cognitivists, think that cognitive processes are mechanistically specified, not operationally specified.

    ReplyDelete
  6. Hi, Charles,

    1. Well, I think that whether "knowledge acquisition" so construed is cognition is part of what is up for grabs. To me, there seem to be social-pragmatist strains of EC, where this would count as cognition. But, to me, "knowledge acquisition" as described above sounds like behavior, part of which is regulated/controlled/shaped by cognitive processes. I don't think the EC folks are all that explicit about this difference; this is my take on what is going on in the background though the EC types seem ready to me to reject any distinctions I make.

    Take another case that I think of. Is visual search a cognitive process or a behavior? I think it's a behavior that involves cognitive processing, but I think that lots of EC folks will take it to be cognitive processing.

    Getting all this on the table is part of what A&A have been pushing for is asking for a "mark of the cognitive". Let's have an account of what they think the cognition is in extended cognition.

    2. I think there are at least some EC types who would be willing to count the sorts of cases you describe as extended cognition. This is sometimes called socially distributed cognition, so not all of the EC examples will be of this Sellarsian type.

    3. I'm not sure what the EC folks think they gain by treating Otto's use of his notebook as a cognitive process. It depends on what is loaded into that account. If one takes the operationalist line, that whatever enables Otto's to get about, is cognitive, then you get the EC conclusion pretty easily, but that does not tell you much about the "important" work that Otto's brain does. If one takes the mechanistic line, then one would have a surprising conclusion that the brainy processing and the brainy + notebooky processing are sort of alike. But, if these are just "coarse" similarities, that have yet to be articulated, it is not clear to me what one gains by this.

    ReplyDelete
  7. Sellars and Davidson argue that knowledge acquisition (the latter specifically re learning a language) is a social practice involving the would-be acquirer, a community (for teaching, justifying, endorsing, et al), and "the world"

    You know, if this is what one means by knowledge acquisition, then clearly that is extended, right? So, what we have to get clear on, is what we are going to be talking about when talking about extended cognition.

    But, I don't think that the EC folks have been all that forthcoming on this. The usual strategy is to come up with some "thin" theory of what cognitive processing is, e.g. it's information processing. Well, since information is processed in lots of places, one gets EC. But, if one takes the social-pragmatist line (or some other behavioral approach to what cognition is), one gets EC by other means.

    ReplyDelete
  8. "if [the Sellars/Davidson view] is what one means by knowledge acquisition, then clearly that is extended, right?"

    Quite so, which motivated my queries. But note that the key components of the S/D process - the would be knower and what Davidson calls a "community of minds" - are presumably "cognitive" for any reasonable definition of that word. And this suggests a possible distinction between Inga+Inga's memory and Otto+his notebook: the former arguably is cognitive, the latter arguably isn't.

    A key component of Sellars' view of "knowing" is justification: being able to assert a proposition and to justify that assertion to a relevant community (to offer reasons for considering the proposition "true"). Now, I assume most people have some sense of the justification process in normal conversation: if challenged to justify an assertion of a certain type, one will often refer to a source that will be considered authoritative within a relevant community: "The Bible says so", "I just looked it up in an encyclopedia", "I read it at the BofC blog", etc. Even in Otto's case, he might remember to include source cites in appropriate notebook entries so that if he remembers to consult his notebook but is challenged, he can respond with a cite.

    But suppose a source is challenged, ie, justification for the assertion that the source is credible is demanded. Presumably, Inga can respond with reasons for considering the source credible; and again, Otto could as well - in principle, assuming he remembers enough about the source in question. But note the number of things that it is necessary for Otto - the presumed Alzheimer sufferer - to remember in order to participate in this process.

    Of course, one can reject Sellars' view and assume that Otto "knows" propositions in his notebook despite not being able to justify them. But that suggests that mere access to a proposition would be considered sufficient to claim to "know" it - which in turn suggests that one "knows" all propositions accessible in any credible source - a very generous interpretation of what it means to "know" something.

    "to me, 'knowledge acquisition' as described above sounds like behavior"

    I have the impression - as usual, not necessarily correct - that behaviorism (at least of the "classical" variety) is out of favor these days, so I just want to confirm that this comment is not meant to be dismissive, at least if it is based on my one sentence summary of a large part of Sellars' EPM essay. His argument addresses (often implicitly since he tries to keep the presentation general) sensation and perception, and proposition formation, endorsement, and justification. It's hard for me to imagine a reasonable definition of "cognition" which wouldn't apply to a process that involves all of these.

    "Is visual search a cognitive process or a behavior?"

    Having spent a fair amount of time trying to learn about the vision process, I probably have an opinion. But I'm not familiar with the term "visual search". Definition?

    "A&A have been pushing for ... a 'mark of the cognitive'."

    Based on the posts and comments here, I had noticed that this "mark" seems to be AWOL. But then I've found lots of other terms in Phil of Mind that appear to be either essentially undefined or multiply defined - which is especially hard on a tyro. ("Perception" is another term that I've found extremely elusive, despite - or perhaps because of - the numerous, and often long, relevant SEP entries.)

    ReplyDelete
  9. But I'm not familiar with the term "visual search". Definition?
    Visual search is simply the process of using vision to scan an array or scene looking for something in particular. In the lab this tends to be letters or something, although Mary Hayhoe's done a lot with more naturalistic VR settings and I just saw Ian Gilchrist presenting some data on a more useful task.

    ("Perception" is another term that I've found extremely elusive, despite - or perhaps because of - the numerous, and often long, relevant SEP entries.)
    This is because it's different things to different people - like so many things in psychology.

    ReplyDelete
  10. I agree with both of Andrew's last comments!

    I think behaviorism is on the outs, but there do seem to me to be traces of it in Gibson.

    But, actually, I think that everyone agrees that there is behavior. What is up for grabs is whether or not there are additional processes--cognitive processes--that play a role in influencing behavior. The cognitivists think yes, the behaviorists and their kin no.

    Well, a mark of the cognitive seems (to a first approximation) to be necessary to get any traction on the issue of EC. But, yes, like so many things there is dispute about what it is, and even whether we need one. But, I don't let that bother me, even if Menary might.

    More and more, however, we may not need a mark of the cognitive so much as just some greater explication of what one is talking about when one says that cognition extends. If one means behavior, then that surely extends.

    ReplyDelete
  11. I think behaviorism is on the outs, but there do seem to me to be traces of it in Gibson.
    Everyone always says this like it's a bad thing.

    ReplyDelete
  12. "Visual search is simply the process of using vision to scan an array or scene looking for something in particular."

    OK, thanks Andrew.

    I distinguish between "reliable differential responsive disposition" (Ryle's term, I think), which does not involve recognition, only differentiation: within one's mental picture of the contents of the FOV, the ability to respond differently in some way to the image on the left and to the image on the right. But if one assumes (as I do) that recognition involves knowledge - that the image on the left is of what we have learned to call a 'triangle', the one on the right is of what we have learned to call a 'square'"), and further assumes that any knowledge component is a necessary and sufficient condition for a process to be labeled "cognitive", then VS as Andrew describes it ("looking for something in particular", eg, a "triangle") seems to me obviously cognitive. What part of that (hopefully) logical sequence might someone who claims that VS is noncognitive dispute? Perhaps the "sufficient" part; but If so, what are the other requirements for sufficiency?

    "[The question] is whether or not there are ... cognitive processes--that play a role in influencing behavior. The cognitivists think yes, the behaviorists and their kin no."

    In order to distinguish his approach from classical behaviorism, Sellars calls it (arguably unfortunately) "methodological behaviorism". But I think that term has to do with the latter part of EPM, which involves modeling mental processes and testing those models against evidence comprising reports by "thinkers" or "feelers". But my comment had only to do with Sellars' concept of the process of knowledge acquisition, which critically depends on intersubjective agreement - we can't "know" anything without participating with others in the giving and asking for reasons that justify our beliefs (a position similar - equivalent? - to Wittgenstein's Private Language Argument, and also to Davidson's concept of "triangulation" in learning a language). I don't see a sense in which that process is "behaviorist" except the presumably uninteresting sense that any overt human activity involves behavior.

    "we may not need a mark of the cognitive so much as just some greater explication of what one is talking about when one says that cognition extends"

    That's certainly what I've gleaned from reading the posts and comments here. Additionally, a proposed benefit to viewing the situation as EC rather than as merely an expanded use of tools would be nice. Which on further reflection suggests to me that I should recant my position that knowledge acquisition ala Sellars is EC. In the absence of that "greater explication" of EC, such a claim doesn't appear to be meaningful.

    ReplyDelete
  13. What part of that (hopefully) logical sequence might someone who claims that VS is noncognitive dispute?

    Well, I take it that VS is a cognitive process in the sense that it is a behavior that has a subprocess that is cognitive. Or, another way of putting it is that VS is a cognitive process in the sense that it implicates cognitive mechanisms. (I think that your differentiating between recognition and differentiation gets at this. Recognition would be the cognitive subprocess.)

    Think about the process of baking a cake. Mixing floor and eggs are part of the process of baking a cake in a perfectly natural sense, but only a subprocess of baking a cake is genuinely baking a case in an oven.

    No, what you describe as knowledge acquisition isn't behaviorism, at least as I understand it. Still, knowledge acquisition is probably a cognitive behavior and I see little reason to think that that behavior is not extended. Behaviors typically are extended, I think.

    But, extended behavior is not the same thing as extended cognition. That is where all that is going.

    ReplyDelete
  14. I've just read over some of the "great TSRM shark debate" and wonder if the RDRD-cognition distinction might be relevant. My inclination is to say that the shark has an RDRD to root around in the vicinity of the apparent source of a detected electric field and ingest certain things it may find there, but that having the concepts "edible food present (absent)" are cognitive states unavailable to any creature not capable of language (at least in my knowledge-based concept of cognition). The shark stops looking simply because it can't keep trying any one spot indefinitely, not because it changes its (non-existent) cognitive state.

    Or maybe I'm missing the point of the debate.

    ReplyDelete
  15. Yes, there was a "great TSRM shark debate".

    First of all, it is unclear to me that TSRM believe in cognitive states or perceptual states. They only talk about states of affairs of the shark-niche system, which might well be just behavior.

    The shark stops looking simply because it can't keep trying any one spot indefinitely, not because it changes its (non-existent) cognitive state. Ok, the Sellars-Davidson line is that sharks don't think, since they don't have the requisite community interactions. Ok. So, if the shark is not cognizing or perceiving, then what is going on what that shark anyway?

    Davidson or some Davidsonian has probably discussed this, but when I hear the garage door opener and when my dog hears the garage door opener, I would say that I know and my dog knows that my wife is home. But, Davidsonians apparently won't say this. So, what is going on with the dog?

    ReplyDelete
  16. Part of the shark story is an attempt to recognise that the shark is engaged in behaviour that has meaning to it; the electrical field specifies something with value and meaning. The shark isn't just a Skinnerian-behaviourism automaton, rooting round because it has no choice: rather, it's engaged in functionally relevant meaningful behaviour grounded in perceptual information. Watch anything by David Attenborough or listen to biologists discuss animals, and it's always talked about in these terms but with little or no reference to representations, models or linguistic content of any kind (until you get to chimps, who suddenly have all these mental models and plans, apparently).

    No one wants to talk about shark cognition in linguistic, representational terms. But its behaviour is rich and complex and not merely stimulus-response. TSRM and the ecological approach want to have a story to account for this that then equally applies to us, to get away from all the complexity psychologists like to think goes on in us.

    I've missed this somewhere: RDRD?

    ReplyDelete
  17. "reliable differential responsive disposition" Couple of comments back ...

    ReplyDelete
  18. Oh, ok - sailed right by that.

    But if one assumes (as I do) that recognition involves knowledge - that the image on the left is of what we have learned to call a 'triangle', the one on the right is of what we have learned to call a 'square'"), and further assumes that any knowledge component is a necessary and sufficient condition for a process to be labeled "cognitive", then VS as Andrew describes it ("looking for something in particular", eg, a "triangle") seems to me obviously cognitive. What part of that (hopefully) logical sequence might someone who claims that VS is noncognitive dispute? Perhaps the "sufficient" part; but If so, what are the other requirements for sufficiency?
    A big part of ecological psychology is how you can have knowledge without representations or any other traditional 'cognitive' apparatus. You do this via perceptual information, ecological laws, etc etc. So a 'non-cognitive' reply is to simply deny that knowledge entails cognition. The other option is to redefine cognition as extended, or embodied, or nonrepresentational, and to say 'sure visual search is a cognitive process, but not what you mean by cognition'.

    Gibson recognised that knowledge is still part of the game; we don't just differentiate things reliably, we engage with them meaningfully (and so do sharks etc). But calling knowledge 'cognitive' doesn't actually help all that much, there's no clear mechanism implied.

    None of this is trying to convince Ken, just lay out that there are ways to answer Charles' question :)

    ReplyDelete
  19. "if the shark is not cognizing or perceiving, then what is going on what that shark anyway?"

    Andrew may not like my response to this question, but hopefully I'll partially redeem myself in a comment to follow which addresses issues he raised.

    I assume what's going on with the shark is neither cognition nor perception: the former because I assume cognition requires knowledge, which in turn requires language, the latter because I assume "perception" to be cognitive in that it includes recognizing what is sensed as being a "certain sort of thing". From that perspective, the shark merely senses the presence of something that may be edible and tries to eat whatever appears to be the source of the field.

    Somewhat along these lines, I recall seeing a film clip showing the defensive mechanism of some fish that pulls a porcupine-like maneuver by deploying spines when eaten by a predator. This precludes the predator's digesting the prey and consequently causes the predator to eject it. Presumably, were the predator capable of recognizing the prey as a "certain sort of fish" that always proves to be inedible, it wouldn't make the mistake of trying to eat it.

    Another distinction I make is between cognitive learning of propositional knowledge (learning that) and non-cognitive learning of skills (learning how). I have no idea, of course, what causes a shark to follow an electric field to possible food, but I assume it's either something innate or a skill developed by mimicry or something similar.

    ReplyDelete
  20. "when I hear the garage door opener and when my dog hears the garage door opener, I would say that I know and my dog knows that my wife is home. But, Davidsonians apparently won't say this. So, what is going on with the dog?"

    To merge Sellars' and Davidson's positions was probably a mistake on my part. I don't know what if any position Davidson has on learning in general (as opposed to learning a language), so I'll stick to trying to apply my understanding of Sellars' position.

    He distinguishes between knowledge acquired inferentially and knowledge acquired non-inferentially. Presumably, no one has any serious issues with how inferential knowledge is acquired - it's by inference! (Doh.) His dispute with sense-data theory, Russell's "by acquaintance" idea, et al, was about the acquisition of the non-inferential knowledge necessary to start inferential chains. And non-inferential knowledge is what he argued was either acquired or endorsed via "community interactions". So, my (probably not very satisfying) answers would be as follows.

    You have learned either directly from members of your various "communities" - parents, friends, teachers, colleagues - or indirectly through credible information sources - encyclopedias, educational videos, etc - about electric motors, gears, drive chains, etc, and that mechanisms employing such components make certain sorts of noise. You have also learned about remotely controlled garage doors and have acquired the skill of recognizing the specific sound of your garage door being opened. Finally, having programmed the remotes yourself, you know that there should only be two that will activate your garage door opener, one of which your wife has, the other you have. Recall that Sellars considers a component of knowing to be the ability to give reasons for what one says (believes). Then the following (belief, reason) sequence applies:

    You believe that you recognize the sound of your garage door opening because of your acquired skill.
    You believe that the garage door opener has been activated because it normally operates only when activated by a remote control.
    You believe that your wife is home because no one else is home to manually activate the garage door opener, so only she could be activating it, using her remote.

    Your dog presumably has learned at a minimum to associate the particular sound made by a garage door opener with the imminent arrival of someone, perhaps even of a member of your household. Whether s/he has acquired the ability to infer from your presence that the new arrival will be some specific other member of your household, I doubt but of course don't know. But I am certain that s/he doesn't have the propositional knowledge "your wife is home" since no dog can.

    ReplyDelete
  21. "But its behaviour is rich and complex and not merely stimulus-response. TSRM and the ecological approach want to have a story to account for this that then equally applies to us, to get away from all the complexity psychologists like to think goes on in us."

    I don't consider a stimulus-response model incompatible with rich and complex behavior, so I'm perfectly comfortable marrying the two even in humans. My suspicion is that human behavior is essentially RDRD with cognition (however defined) as an added feature that just makes the responses a lot more complex, interesting, and harder to analyze.

    And notwithstanding my use of "mental imagery" in a previous comment, I have no problem with the concept of non-representational knowledge. In trying to attack the problem of explaining visual imagery, I've tried to think about how much we could do in principle if we didn't have it and what additional capabilities accrue to having it. It appears that we can do a lot without it - identify the presence and relative motion of objects, recognize shapes, patterns, use weapons, play sports, et al - although perhaps we can do some of those things better with it. However, one thing it seems we can't do in the absence of visual imagery is to describe the contents of our visual FOV to others, an ability which obviously also requires language. So, it appears that while visual imagery isn't necessary for some activities, the combination of visual imagery and language is a potent enhancement.

    I've been reading Sabrina's posts on Gibson's book, and your comments on his approach will motivate me to pick up the pace a bit.

    ReplyDelete
  22. Speaking of mental imagery: Sabrina posted this about a man with no imagery. It seems less important than people think. (Feel free to comment over there, we'll see it.)

    My problem with this take on cognition is that it all assumes that the requisite information has gotten 'into the system' so it can form the basis of an inference, or become associated with an action, etc. The process of 'getting into the system' is perception, and the ecological analysis suggests a) traditional sense-data/sensation based theories are insufficient to allow what is required, but that b) there is information available sufficient to do the job, but it's different from what normally gets talked about. Once you get into that information, you start having potential explanations for complex, flexible behaviour that avoids the limits of SR psychology but doesn't require problematic linguistic cognitive apparatus.

    ReplyDelete
  23. I just reread Gary Williams' recent post on Tyler Burge's NYT article and realize that much of the content of my last three comments could be replaced by the single statement "What GW says", thereby effecting a dramatic improvement in erudition and coherence. I'm not so sure about his comments on language and representation, but to the extent I understand the rest of his post, I completely agree.

    In particular, the part about the primate and the strawberry is essentially what I was trying to convey in my comment about the shark.

    ReplyDelete
  24. Andrew -

    I've been wishing for an article exactly like that, but had little hope of finding it. Many, many thanks (and to Sabrina as well). I'm going to comment on it here because my response relates directly to my comment on mental imagery above.

    Because it better suited my purpose in that comment, I presented the relationship between mental imagery and describing a "scene" as the need for the former in doing the latter. But what I actually hypothesize is that a verbal description is formed first and the associated mental imagery is an illusion attendant thereto. Two reasons: first, my take-away from Libet is that the capability we call "consciousness" is in general not an a priori capability that aids in representation, decision making, executing our "free will", etc, but rather an a posteriori capability to report the consequences of essentially the highly complex human version of RDRD to which I alluded earlier. And a verbal description of what we "see" in our FOV is often part of that reporting. But reporting not to the problematic homunculus in our internal Cartesian Theater but rather to other members of our various linguistic communities.

    Which leads to the second reason. Since the intersubjectivity our linguistic capabilty enables seems to be a critical component of many of our special capabilities, it seems possible - perhaps even likely - that enhancing our capability for intersubjectivity by creating verbal descriptions of the contents of our FOV takes importance-priority over whatever benefits accrue to creating the associated mental imagery, and therefore may take causal-priority as well.

    Consistent with this view, my candidate explanation for the difference in the characteristics of the observed processing delays for the those in the control group - who visualized the rotations - and MX who couldn't do that is that the time required for the rotational processing is about the same in each case but that the additional unnecessary formation of mental images by the former as they "stepped through" the rotation added delay in proportion to the number of steps required. Since no detailed results were presented in Zimmer's article, I of course don't know whether that explanation is consistent with the study results.

    It's not clear to me whether this view is closer to or further from your EP view. But note that I use "S-R" in a general sense, not as shorthand for "S-R psychology", about which I know nothing (I don't even know what it is). Also, I have no vested interest in any specific approach. I've been thinking in terms of Sellars' epistemological approach only because it's the only one with which I'm reasonably familiar, but I'm open to being "reeducated". Since Gary Williams' views resonate and I gather he is a Gibson fan, I may well be especially susceptible.

    ReplyDelete
  25. the shark merely senses the presence of something that may be edible and tries to eat whatever appears to be the source of the field.

    But, why say that the shark senses the edible, rather than that the shark senses the electrical field or chemical gradient, then infers the edible? (TSRM say that two are identical and it causes the sorts of problems I described earlier.)

    You might find that sensing smells and sensing food are dissociable, as by single electrode recordings that reveal differences in patterns of activation. Or, (if you weren't Andrew) you might use fMRI (on a shark?) to disentangle them.

    ReplyDelete
  26. Whether s/he has acquired the ability to infer from your presence that the new arrival will be some specific other member of your household, I doubt but of course don't know. But I am certain that s/he doesn't have the propositional knowledge "your wife is home" since no dog can. Actually, I suspect that the dog, Motley, can inferwho it is, since she reacts more enthusiastically to my wife's opening the garage door than to mine.

    ReplyDelete
  27. "why say that the shark senses the edible, rather than that the shark senses the electrical field or chemical gradient, then infers the edible?"

    Despite sincere attempts, I too inevitably fall into the sloppy language trap against which I rail. So, I'll see your modification and raise you one:

    The shark senses the electrical field [or whatever], and if in certain biological states (that we might describe as "being hungry") reacts - either innately or because of parental training - by following the field to its apparent source and attempting to consume whatever it finds there.

    Not withstanding failed attempts, my objective is to make the description of the shark's behavior as close to the description of a thermostat's "behavior" as possible. (I suspect everything is in essence just a thermostat, although some are terribly complex.)

    "I suspect that ... Motley can infer who it is"

    Well, "infer" seems tricky. I might rephrase as follows:

    Motley can distinguish between scenarios which are consistent with my wife's arrival but not with mine, and scenarios which are consistent with my arrival but not with hers.

    This is testable. Have a third party monitor Motley's reactions to:

    1. You are in the house, in Motley's presence, and your wife arrives.
    2. Those roles are reversed.
    3. Neither is in the house and you arrive.
    4. Neither is in the house and your wife arrives.

    It might well be that Motley would respond differently to scenarios 1 and 2, but I would be stunned were Motley found to be capable of distinguishing 3 and 4.

    ReplyDelete
  28. The shark senses the electrical field [or whatever], and if in certain biological states (that we might describe as "being hungry") reacts - either innately or because of parental training - by following the field to its apparent source and attempting to consume whatever it finds there.

    Ok. But, what is the connection (from the shark's perspective) between sensing an electric field then engaging in a food search? Isn't there going to be an inference from the sensing of the electric field to the belief that there is something edible? This is the thing that TSRM were trying to block by saying that "detecting the field" and "taking there to be something edible" were in fact the same. (Although I argued that this made plenty of trouble for them.)

    Well, I think that in truth Motley does not do much when I arrive home first, but does a lot when my wife arrives. When I come home she lays on the floor hardly concerned, when my wife comes home she's jumping on the door to the laundry room (which separates the garage from the kitchen). I think Motley also uses time of day to figure out who is coming even before seeing who walks through the door. So, Motley can't tell just by the sound of the garage who is coming, but by using time of day she can figure out which of us it is likely to be.

    ReplyDelete
  29. "Isn't there going to be an inference from the sensing of the electric field to the belief that there is something edible"

    I really, really want to avoid attributing activities like inference and belief to sharks. It may turn out that they do one or both of those things, but I want to see how far we can go without assuming that they do. In fact, I want to see how far we can go without assuming that we do them, so I'm not about to give in on sharks until evidence demands it.

    The thermostat analogy is intended to avoid attributing inference - thermostats don't "infer" from the decline in temperature below a threshold that it's time to activate the heater, they just mindlessly do the latter in response to occurrences of the former. In thinking about sharks, I recall the trained seals at Sea World - they presumably weren't inferring anything from the sounding of the bell, just mindlessly responding as they had been trained to do.

    Similarly, I hypothesized that the shark tries to eat whatever appears to be causing the field, hoping thereby to avoid attributing belief and concepts like "edible" to it.

    "TSRM were ... saying that "detecting the field" and "taking there to be something edible" were in fact the same."

    Since I am inclined to interpret "taking there to be" as a belief and "something edible" as a concept, I don't even accept that the shark can do the latter, never mind that it is in some sense "the same" as the former. So, I can't imagine why anyone would say this.

    ReplyDelete
  30. I agree with you that the identification TSRM want to make is pretty far fetched, but it is clear that they want to identify "detecting a field" and "taking there to be something edible" as a means to avoid having to say that the shark infers the latter from the former. They are pretty explicit about that.

    I don't know that the seals at Sea World are responding mindlessly. There are clear behavioral analogies between them and the thermostat, but if you want to know whether the seals are thinking, you would have to try to get beyond mere behavior and look at the mechanisms. We can agree on that, right?

    ReplyDelete
  31. "you would have to try to get beyond mere behavior and look at the mechanisms. We can agree on that, right?"

    Of course. I was admittedly over simplifying. Even a basic thermostat obviously isn't really all that straightforward when you go to the micro level. There is variability due to metal fatigue, hysteresis, direct light, corrosion, et al. But whether or not to worry about those effects depends on context. For every day use in a home system, such effects are deep in the noise, and it would be foolish to worry about them. In satellites, they typically are critically important.

    And absent a context, it isn't immediately obvious that whether a shark does something that could arguably be called "inference" and whether Otto's notebook should be considered part of his "cognitive" system are important issues. Is there a context in which they are?

    ReplyDelete
  32. The mechanism matters: the shark is doing inference implies bits of that shark are up to different tasks than if it is perceiving the affordance 'edible by the shark' via detection of information. So, for example if you want to figure out if the brain is representing or not, you will be interested in which mechanism it's implementing.

    ReplyDelete
  33. Regarding the Otto case, what I think is at stake is what theory applies to it. So, I think that there is going to be a rich scientific theory of the processes that go on in Inga's head, but only superficial points of analogy between that and what Otto does with his notebook.

    Now, this is exactly where Clark and I differ. Clark thinks the differences are superficial, where I think that the similarities are superficial. That makes for a nice symmetry between Clark's view and mine, but what it leaves out, to my mind, is that Clark is downplaying a lot of psychological research that has taken place over the last century or so. He's saying in a thinly veiled way that that is not that important.

    So, what is at stake, to my mind, is what sorts of things are covered by what scientific theories. And the less informative your theory, the more it is likely to cover. So, if you just say that the mind is any old dynamical system, then it won't be a surprise at all to find that that extends.

    ReplyDelete