Re: The Dance of the Copy Theorist
Larry Kaye
Allow me to point out the mentalistic concepts that you are using:
I can get a pigeon to accurately *call* a shirt "crumpled"
*finding* small fossils in the dirt
*describe* our seeing
*correctly* answer *questions*
your *misconception*
they have *conceptualized*
you *appear* to have *come to the conclusion*
does a pretty good job of *interpreting*
need be *observable* (especially by an outside *observer*)
Let's take the last one. Can you translate the idea of observation into purely behavioral terms? (Note that it's possible to have a blind person who could face something with their eyes open and move their eyes in the same way as someone scanning the object.)
Try to translate any of these notions into purely behavioral descriptions and you'll begin to see Chomsky's point.
Best,
Larry
P.S. From what I've seen of pidgeon training, I doubt very much that you could get a pidgeon to distinguish shirts from slacks or jackets, etc. let alone crumpled from uncrumpled shirts! (As a mentalist, I'd say that pidgeons' innate conceptual limitations prevent them from aquiring the concept of a shirt and probably also the concept of being crumpled.)
-----Original Message-----
From: Glen M. Sizemore [mailto:gmsiz...@COX.NET]
Sent: Sunday, May 08, 2005 1:15 PM
To: PSYC...@LISTSERV.UH.EDU
Subject: Re: The Dance of the Copy Theorist
Larry: This Skinnerian view is a non-starter for many, many reasons. Here are a few
main ones:
Many of the qualities we perceive are not physical qualities or kinds. for
instance, I can see a crumpled shirt. But there's no hope of providing a
reduction of that property to purely physcial properties. (Even though, no
doubt, these properties "supervene" on physical properties--every crumpled
shirt is made up of some complex set of physical attributes or other.)
Glen: I'm not sure it is impossible (nor am I sure what this has to do with Skinnerian behaviorism - this is a recurring theme), but as Herrnstein and Loveland pointed out in their studies of discrimination of natural categories, finding out what is controlling the behavior of individual subjects in the real world or in controlled experiments involving discrimination of natural objects is likely to be so difficult as to be nearly impossible and is not likely to tell us much. [My view is that it IS likely impossible - if there are enough of the features present in an example, that were present when behavior was reinforced,we get "accurate" responding. The next example might contain only a few of the features, but if other features are present, you will get accurate responding.] You somehow think that this is somehow antithetical to some principle espoused by behaviorists, but I don't recognize the principle that you seem to be pointing to (see below) as one that is, in fact, !
espoused by behaviorism. We know, for example, that it is exposure to contingencies of reinforcement that is responsible for performance in experiments where non-humans are trained to respond "appropriately" to natural categories. I can get a pigeon to accurately call a shirt "crumpled" and so can you - but you would first have to read some of the literature generated by the experimental analysis of behavior (now usually called behavior analysis). The issue of "pinning down" the exact dimensions controlling behavior has nothing to do with the fact that the behavior that we do see is profoundly influenced by contingencies of reinforcement.
Larry:
also, by the way, likely that the so-called secondary qualities are not
physical/external--e.g., if color is not a property of the external
environment, then it can't be that I see color in virtue of being reinforced to
respond selectively to the colors. To get some idea of the problem: if you
claim that I have been reinforced to respond selectively to 520nm light waves
and label them "green" then you will have no explanation of why I also classify
the shadows of red and white light as green (since the reflected wavelengths
are not 520nms. But all the same, color shadows look vividly colored.) So the
first p!
roblem is that many of the qualities we see are not physical qualities of the
world, so an explanation in terms of reinforcement of selected reponses to
physically characterized inputs won't work.
Glen: I can gain control over responding by manipulating wavelength and arranging the proper contingencies. This does not mean that "wavelength" is the only variable that will control the response "red" or "green" or some equivalent response in a non-human animal. If I present a black valentine for a very brief period of time, observers will probably report that it is red. This does not mean, however, that the wavelength we associate with "red," and the contingencies of which such stimuli are a part, are not important variables.
Larry: It's useful to consider the next two points together: Second, it seems obvious
that we respond to only a small amount of the vast number of things and
qualities that we actually see.
Glen: Right. This is because if we want to be sure to gain control by a particular dimension of a stimulus, we must arrange explicit contingencies [BTW, can you tell me how we would do this?]. In a classic experiment in the EAB, responding is reinforced in the presence of a circle on a red background, and extinguished in the presence of a square on a green background. Some animals will respond to a circle on a black background (rather than to a plain red stimulus), whereas others will respond to a plain red stimulus. We could, of course, make it so that the animals would only respond to color or to shape, but the necessary, specific contingencies were not arranged in the experiment described above.
Larry: Third, it is flattly false that in order to see
something novel a person must be reinforced in some way. I might, for instance,
pass a person on the street, take a good look at them and not respond in any
way.
Glen: The "looking" IS a response, and its characteristics can be plausibly attributed to contingencies - PAST contingencies.
Larry: And if I meet them again, I may well recognize that it's the same person.
But there's been no reinforcement of any sort, nor have I responded in any
way.
Glen: Where, in time, do you think the reinforcement is supposed to be? No reinforcement need be observable (especially by an outside observer) when you looked at the person. The characteristics of your "looking" have been produced by PAST reinforcement. A person whose behavior has been reinforced by, say, finding small fossils in the dirt, will likely do so better than someone who sees just as well, but has never found a small fossil in the dirt. This is because the behavior of looking has been shaped by reinforcement, not because the first person "sees better."
Larry: It is equally possible to, say, spot a new kind of animal--a kind of bird I've
never seen before--and not respond in any way nor be reinforced in any way yet
see and recognize it again a while later.
Glen:
We learn to respond in subtle ways (at time "t=0") that somehow "makes it so that we are later able to answer questions [and correct answers are reinforced] about what we have seen." At first we are trained to describe our seeing while we are engaging in it, but with a gradual increase in delays we "come to note" (i.e. learn to respond in certain ways) because doing so has been reinforced (albeit at some delay) later when we correctly answer questions. But, again, your misconception here concerns the temporal locus of reinforcement; the behavior that you acquire is due to a PAST exposure to contingencies. Additionally, the behavior acquired is extremely subtle and complicated. Although cognitive psychologists might "study" such behavior, they have conceptualized it poorly AND they are not interested in its ontogeny - but the latter is simply an outcome of the former.
Larry: Think of a nice vacation in scenic vistas--how much of the scenery do we
actually respond to? Andwhat's the reinforcement--seeing the beauty of the
scene? (=disguised mentalism--see below)
Glen: Things seen can indeed be reinforcers - but what you are saying here has very little to do with behaviorism. Somehow you appear to have come to the conclusion that everything physically present when a response is reinforced should automatically be equally potent as a controlling variable.
Larry: Fourth, it's just not true that children learn to see in vitue of
"reinforcement", at least as far as, say, selectively being given food pellets
(or M&Ms, etc.) for appropriate responses. (What a horrible upbringing that
would be!) Parents typically begin rewarding and punishing children (for good
and bad behvior--which often has little to do with the kind of learning you're
thinking about) long after they can see/perceive, and for that matter
communicate a bit.
Glen: Even in non-human animals a variety of events function as reinforcers, and in humans it can no doubt be shown that certain sorts of novel visual and auditory stimuli function as reinforcers, as well as a variety of subtle (and often not-so-subtle) social stimuli. With reference to learning to see and hear etc. the early reinforcers are, I suggest, aspects of the sensory stimulation themselves. These very quickly become embedded in the larger set of contingencies that reinforce behavior that operates on the environment in a more dramatic fashion.
Larry: Fifth and finally, the underlying problem with the Skinnerian approach is that
there's no non-circular way of specifying what counts as a reinforcer.
Glen: Of course there is, and the answer given here basically covers all of the issues you raise below (even though I will respond more specifically to some of them). Reinforcement is simply the name give to a set of observations. These are:
1.) There must be a contingency between response and consequence.
2.) There must be an increase in the rate of the response.
3.) The rate increase must be due to the contingency (this last one requires some other observations and is often taken for granted - but I don't wish to get unnecessarily complicated here).
Larry: As we
all know, virtually anything can be rewarding depending on what your desires
are like.
Glen: There are a bewildering variety of reinforcers, as I have already suggested, but saying that this is because of our "desires" is what is circular, since "our desires" are inferred from our behavior.
Larry: Pain can be rewarding for a masochist or maybe for an athlete in
intense training, and an M&M is not rewarding for those who don't like
chocolate or who know that it will make them sick.
Glen: The above observations are not at odds with the behaviorism that I am familiar with (except for the circularity implied by "masochist," "don't like chocalate," as should be clear by now.
Larry: Skinner ultimately resorted to saying that a reinforcer is something that
increases the frequency of emission of responses, but as I say that's a
vacuously circular specification in relation to Skinner's main thesis:
responses increase *because* they're being reinforced. (Similar problems occur
for defining what a "response" is--if you're standing and looking at a
painting, is that "looking behavior"? (=disguised mentalism) Is it behavior at
all?) Now, if one adds mentalistic psychology: the response is increasing
because the reinforcer is satisfying desires of the subject, then everying's
fine; as Chomsky put it in his famous review of _Verbal Behavior_ what's beng
touted as behavioristic psychology is really just common sense belief desire
psychology with a new jargon that makes it appear like it's behaviorism.
Glen: I'll just respond to the portions that I have not already covered. Our behavior when we stand in front of a painting is frequently quite subtle (but the "eye fixation point" literature directly measures some of it), and it is certainly a philosophical position to say that some behavior may be you appear to have come to the conclusion (the issue of private events) or may not be observable by anyone! But this is not disguised mentalism because specifying the unobservable event as behavior constrains its nature, and suggests features of an animal's ontogenic history that are important in order to observe the sorts of things from which the private behavior is inferred. Notice, however, that I am not saying that "having desires" is private behavior. There is nothing circular about the definition of reinforcement (as I have shown), and nothing that requires asserting the existence of unobservable entities (like "desires") that are simply inferred from behavior and then emp!
loyed to explain it. Reinforcement is not a theory, but its extrapolation to complex behavior is philosophical, but there is no circularity there. I say, for example, that the way we behave when we look at something is reinforced behavior, but this is not circular in any sense. All it means is that the behavior exists at its current frequency because it has produced certain consequences in the past. It is not demonstrated - it is interpretation. So what? There are two issues: can the behavioral processes demonstrated in the laboratory lead to the cogent, internally-consistent interpretation of human behavior, and second, can that interpretation be shown to be correct (or at least likely) by experimental means. Despite what Chomsky says (nothing in his review is relevant to what Skinner wrote in Verbal Behavior, just as what you have written above has not much to do with what I said, or what Skinner said) I think that behaviorism does a pretty good job of interpreting comple!
x behavior in terms of the behavioral processes demonstrated in the la
boratory. "Proving" this is a different story (and some of the difficulties are primarily ethical - we could settle the dispute, for example, about how much exposure to language and other people, is necessary for acquisition of language and social behavior, but much of the easily imaginable research could not ethically be carried out in humans).
With the utmost respect,
Glen
<snip the tail>
>
> From: Larry Kaye <Larry...@UMB.EDU>
> Date: 2005/05/07 Sat PM 05:07:02 EDT
> To: PSYC...@LISTSERV.UH.EDU
> Subject: Re: The Dance of the Copy Theorist
>
> This Skinnerian view is a non-starter for many, many reasons. Here are a few main ones:
Steven Lehar
Peaty >
I think Stephen Lehar's cartoons illustrating epistemology and explanations
of the analogue nature of much of the processing are on the money. ...
However I disagree strongly with some of his assumptions in that I
think that far less is actually being modelled at any given moment than
Stephen says is necessary.
< Peaty
All I am saying is that how much of the world is actually being modeled at
a given moment is exactly how much of the world is being experienced at
that moment. I do not claim that experience is reified at full resolution
in all directions...
http://cns-alumni.bu.edu/~slehar/focal/focal.html
because that is not our experience. Our experience is more fragmented, like
successive glances around the world...
http://cns-alumni.bu.edu/~slehar/focal/focal1.html
with visual resolution dropping off dramatically in peripheral vision. If
that is how it is experienced (not to the naive observer, but to the
informed observer) then that is how it is represented in the brain also.
Steve
Glen M. Sizemore
Allow me to point out the mentalistic concepts that you are using:
I can get a pigeon to accurately *call* a shirt "crumpled"
Glen: Translation: I can arrange contingencies such that a pigeon will peck one key if a crumpled shirt is present and another if some other object or photo is presented (two key version). Or: I can arranges contingencies such that a pigeon will peck a key at a high rate if a crumpled shirt is present and at a low rate if some other object or photo is present. Even without the translation, however, I fail to see what is mentalistic about “calling” a shirt crumpled. In ordinary language, this means simply that a person will often emit the vocal response “crumpled” (usually as part of a larger sample of vocal behavior) in the presence of a particular category of objects (as judged by other members of the culture). What is mentalistic about this?
Larry: *finding* small fossils in the dirt
Glen: translation: engaging in behavior that is reinforced by the acquisition of a small fossil. I could say a lot more but it seems unnecessary. Indeed, like the previous example, it is not clear to me that there is anything mentalistic about it, even without the translation. If we (educated speakers of English) were to observe people engaging in certain sorts of behavior (i.e., orienting their eyes toward a small patch of dirt while sifting the dirt, picking up a small fossil and putting it in a container etc.), we would say “they’re finding small fossils.” Where is the mentalism?
Larry: *describe* our seeing
Glen: translation: emit verbal behavior (vocal or otherwise) that is under stimulus control of the behavior called “seeing” or, more likely, specific perceptual responses (i.e., “I see a dog”).
Larry: *correctly* answer *questions*
Glen: translation: emits specific verbal (vocal or otherwise) responses that are reinforced in particular contexts, in general, by other members of the culture.
Larry: your *misconception*
they have *conceptualized*
you *appear* to have *come to the conclusion*
does a pretty good job of *interpreting*
Glen: Sorry, I don’t have to avoid terms in a non-technical context. For example, I see nothing wrong with “bear in mind that…” in ordinary usage.
Larry: need be *observable* (especially by an outside *observer*)
Let's take the last one. Can you translate the idea of observation into purely
behavioral terms? (Note that it's possible to have a blind person who could
face something with their eyes open and move their eyes in the same way as
someone scanning the object.)
Glen: translation: to show stimulus control (or, really, any antecedent stimulus function). If we want to know if a thin wire is observable to a pigeon, we would reinforce responding in the presence of the wire, and withhold reinforcers in its absence. If discriminated responding emerges, we can conclude that the wire is observable by the pigeon.
Larry: Try to translate any of these notions into purely behavioral descriptions and
you'll begin to see Chomsky's point.
Glen: I’ve been at this a long time, and I have read both Chomsky and Skinner (as well as MacQuorquadale’s response to Chomsky’s review). How about you?
Larry: Best,
Larry
P.S. From what I've seen of pidgeon training, I doubt very much that you could
get a pidgeon to distinguish shirts from slacks or jackets, etc. let alone
crumpled from uncrumpled shirts! (As a mentalist, I'd say that pidgeons' innate
conceptual limitations prevent them from aquiring the concept of a shirt and
probably also the concept of being crumpled.)
Glen:
Watanabe, S., Sakamoto, J., & Wakita, M. (1995). Pigeons' discrimination of paintings by Monet and Picasso. Journal of the Experimental Analysis of Behavior, 63, 165-174.
Pigeons successfully learned to discriminate color slides of paintings by Monet and Picasso. Following this training, they discriminated novel paintings by Monet and Picasso that had never been presented during the discrimination training. Furthermore, they showed generalization from Monet's to Cezanne's and Renoir's paintings or from Picasso's to Braque's and Matisse's paintings. These results suggest that pigeons' behavior can be controlled by complex visual stimuli in ways that suggest categorization. Upside-down images of Monet's paintings disrupted the discrimination, whereas inverted images of Picasso's did not. This result may indicate that the pigeons' behavior was controlled by objects depicted in impressionists' paintings but was not controlled by objects in cubists' paintings.
Key words: stimulus control, concept, pattern discrimination, vision, key peck, pigeon
With affection,
Glen
<snip the tail>
>
> From: Larry Kaye <Larry...@UMB.EDU>
> Date: 2005/05/09 Mon AM 01:14:29 EDT
> To: PSYC...@LISTSERV.UH.EDU
> Subject: Re: The Dance of the Copy Theorist
>
Larry Kaye
None of your analyses are successful, although I can see (no pun intended) that I'm not going to convince you. But I'll take one last shot anyway at critiquing a few of your attempted translations:
Just because a pigeon can distinguish a single picture of a crumpled shirt form another picture does mean that the bird has the concept of a crumpled shirt. If it could distinguish *any* picture of a shirt from pictures of similar things (other items of clothing) then we might say it has the concept of something like a shirt-appearance, although it still wouldn't know the social role of shirts-see below-and if it could distinguish any picture of a crumpled thing from pictures of non-crumpled things then it would the makings of the concept of being *crumpled*. You seem to suggest that training on a single picture would demonstrate this ability, but, clearly it doesn't-I believe in your terminology what I'm skeptical about is the ability of the response to generalize-you can't just assume that that will happen.
On to your translations:
Larry: *describe* our seeing
Glen: translation: emit verbal behavior (vocal or otherwise) that is under stimulus control of the behavior called "seeing" or, more likely, specific perceptual responses (i.e., "I see a dog").
Suppose that every time I see someone looking at me I say "hello". It doesn't follow that "hello" is a description of them looking at me. Moreover, I can have a description of an item that is not in any sense under the stimulus control of the item. "Aristotle was the teacher of Alexander" is a perfectly fine description of Aristotle yet it is not in any sense under the control of the presence of Aristotle. (And even if he'd return from the dead this description wouldn't be under such control-I doubt that I'd recognize him.)
So your translation of "describe" fails. As I say, it's useful to try and see why these sorts of analyses are way off the mark.
Let's take the last one. Can you translate the idea of observation into purely
behavioral terms? (Note that it's possible to have a blind person who could
face something with their eyes open and move their eyes in the same way as
someone scanning the object.)
Glen: translation: to *show* stimulus control (or, really, any antecedent stimulus function). (my emphasis)
You ignore the fact that someone can observe something that they never respond to. Again, the analysis is unsuccessful.
As for:
Watanabe, S., Sakamoto, J., & Wakita, M. (1995). Pigeons' discrimination of paintings by Monet and Picasso. Journal of the Experimental Analysis of Behavior, 63, 165-174.
Pigeons successfully learned to discriminate color slides of paintings by Monet and Picasso. Following this training, they discriminated novel paintings by Monet and Picasso that had never been presented during the discrimination training. Furthermore, they showed generalization from Monet's to Cezanne's and Renoir's paintings or from Picasso's to Braque's and Matisse's paintings. These results suggest that pigeons' behavior can be controlled by complex visual stimuli in ways that suggest categorization. Upside-down images of Monet's paintings disrupted the discrimination, whereas inverted images of Picasso's did not. This result may indicate that the pigeons' behavior was controlled by objects depicted in impressionists' paintings but was not controlled by objects in cubists' paintings.
Key words: stimulus control, concept, pattern discrimination, vision, key peck, pigeon
An impressive study-suggests that pigeons can distinguish complex patterns (and, maybe shows that these painters' styles are a bit more superficial than some might think) but those are points that any mentalist can accept. But this doesn't touch on the concerns about the concept of a shirt noted above-what I'm suggesting is that being a shirt cannot be reduced to any collection of stimulus patterns. Consider a T-shirt, which is a perfectly fine shirt (now) but which was a somewhat radical departure in its day, stimulus pattern-wise, from traditional shirt-appearances. In other words, why call a t-shirt a shirt? Not because I it looks like traditional men's shirts-it doesn't, but because of its similar social role.
If you're really interested in the limits of the analyses of concepts and meaning in terms of stimulus situations, you might look at Chapter 2 of Quine's Word and Object. The failure of analyses like yours doesn't lead Quine to abandon behaviorism but rather makes him skeptical about the notions of meaning and of concepts. [It is, however, a notorious difficult chapter to understand.] (Just to be clear, Quine's skepticism about meaning is not a view I agree with, but I'm just trying to make a sympathetic suggestion.)
Best,
Larry
P.S. I've been at this a long time too. I studied Verbal Behavior with a student of Skinner's (as an undergrad in the mid 70's) and accepted the outlook for about a year. I then came to accept Quine's more sophisticated behaviorism for a short while. But Chomsky's and Fodor's critiques of behaviorism lead me to reject any form of behaviorism in favor of cognitive psychology. I later studied with both Fodor and Chomsky at MIT, but by then Skinnerianism was a dead issue. My examples probably reflect Fodor's inspiration-the crumpled shirt example is his, but from a different context.
-----Original Message-----
From: Glen M. Sizemore [mailto:gmsiz...@COX.NET]
Larry: *describe* our seeing
Larry: *correctly* answer *questions*
Glen: Sorry, I don't have to avoid terms in a non-technical context. For example, I see nothing wrong with "bear in mind that..." in ordinary usage.