Recently, I saw a famous learning theorist -- perhaps one of the two most influential learning theorists in the last 40 or so years; if ΔV = αβ(λ - ΣV) means anything to you, you'll have narrowed it down to the two -- give a talk at the behavioral neuroscience area's weekly colloqium here. The talk, on extinction, was interesting I suppose, but what really caught my attention was the speaker's language. At some point, I had to look around to make sure I wasn't at a Watson talk, circa 1915, because he kept saying things that I'd thought, well, people didn't say anymore. For example, twice during the talk, he used the phrase, "the/that stimulus learned," and at least once used the phrase, "the stimulus trained." I don't know about you, but my natural inclination would be to say, "The rat learned," or "The organism learned," but that's because I'm one of those rare birds who actually thinks things of relevance are going on inside organisms' heads. As if to confirm that he wasn't, the speaker even indicated that he thinks neuroscience is boring, in response to a question from one of the behavioral neuroscientists in attendance.
This was, I believe, my first encounter with a real, living, hard core behaviorist in person, and I must say, it was sort of like taking a time machine and watching a T-Rex devour a stegosaurus corpse. It was surreal, to say the least. But it made me wonder, have any of you encountered any hard core behaviorists of the sort who would attribute learning to the stimulus instead of the organism? Were you as blown away as I was? Any good stories?
I've always wondered how a true behavioralist would interpret the big working memory research, like the original 7+/-2 study. Attributing the learning to the stimulus seems like a way to avoid describing it, but maybe that was their view.
Wow! I heard they exist but did not have the fortune to meet one in person yet. I was starting to wonder if they are mythical creatures, like unicorns.
Wait a minute...you're clearly talking about Bob Rescorla, who I used to work with, and he's no behaviorist! In fact, he championed the so-called "cognitive" view of Pavlovian long ago...and was regularly denounced by the "true" behaviorists of the time. Bob regularly talks about the nature of representations and how they change with experience. He was a graduate of Swarthmore and very much influenced by the Gestalt traditions.
Admittedly, I stopped working on his stuff long ago...maybe he's changed, but I doubt it!
Jim, I am talking about Rescorla, and I knew of some of that stuff he'd done in the past too. But in this talk, it was hard not to think that he was a behaviorist, with the "stimulus learned" and denigration of neuroscience. He couldn't even bring himself to refer to the animals at all for most of the talk -- it was just stimuli, training, and extinction. He was just interested in extinction curves that are, in essence, independent of anything but, well, extinction curves.
Interestingly, the R-W model is used pretty frequently in the causal learning literature, which is distinctly cognitive (associations don't have causes, they just have co-occurrences).
Let me quickly add: what I consider a behaviorist, and what, say, people in the Pavlovian Society consider a behaviorist, may be two different things. Rescorla, and people in the P.S., do talk about representation (the function up there has expectancy built in, even), so they're clearly not old-school Pavlovians, and they're not behaviorists in the Skinner sense either. And I don't just mean they're not Skinnerians -- they're not -- but that they're not people who don't really believe there's stuff in the head. They just don't put much in the head, and they don't care a whole hell of a lot about it either.
The Rescorla-Wagner model puts all the "cognition" purely in the environment. Associations *are* in the environment, and we respond to those associations. The R-W model operationalizes "expected" or "surprising" purely in terms of interaction with the environment, instead of circularly inferring something (or, as the cog-neuro people often find, many parallel somethings) that initiate within the organism.
I have to wonder a bit about your recall--if you can find a transcript, that would be lovely--because, you see, I *am* a behaviorist, and I am perfectly comfortable talking about a rat learning, or a person learning, or any other organism learning. Yes, the organism may be learning (among many many other things) to respond to an association that exists in the environment (Pavlov associated the bell and the food; the dogs responded to that association--to say that the dogs "formed an association" would be purely circular), but it is the organism that is learning.
I find modern neuroscience fascinating, but it is at a different level of explanation than the organismic level. It worries me when dualistic language remains in these explanations--we did not discard the cartesian theatre in order to replace it with a brain-body dualism that is functionally the same. The multiple parallel processing that the neurocog people find is processing of environmental stimuli*, and not evidence of some cartesian homunculus, but I still hear such groaners as "when a neuron decides..." that attempt to introduce agency where it does not belong.
*so as not to have you continue to think you are at a Watson talk, remember that stimuli include both antecedent and consequent stimuli, and that the vast vast vast majority of modern behaviorism is not the S-R behaviorism of long ago (Chomsky never understood that), but a selection by consequences (reinforcers and punishers, and discriminative stimuli as well as eliciting stimuli) and an active organism--a whole organism, not a body-puppet controlled by a brain.
You might be surprised at how many of us behaviorists there actually are, and how well radical behaviorism dovetails with cognitive neuroscience. Much better than with the cognitive psychology that made up the "revolution".
Lastly... although the talk of behaviorists as dinosaurs is, at this point, water off a ducks back, I find it a bit odd that this particular bit of ignorance is not merely accepted, but celebrated. There have been attempts to address this--e.g., Todd & Morris, 1992, "Case Histories in the Great Power of Steady Misrepresentation", and some more recent (but less memorable--at least, I can't recall the authors or titles right now) articles addressing the utility of the Verbal Behavior approach. The vast majority of the cognitive psychologists I have met have a deplorable ignorance of my field; this, of course, would not be a problem at all, except that they use their ignorance as evidence of its demise!
Anon, on the transcripts, I very distinctly heard him say, "The stimulus learned," and he said it twice. Several people also confirm the "stimulus trained" remark as well, and none of us heard him talk of the organism learning (he did talk of the organism with things like, "The organism was placed in the...").
On behaviorism: I know a bit about learning theory, and we have some learning theorists here whose talks I go to now and then, but it's true, I don't know a whole hell of a lot about contemporary behaviorism. I don't really try to hide that ignorance, either, 'cause for the most part, I don't care. But I meant this post, and the dinosaur metaphor, as a joke of course.
Also, you should hear how the P.S. people and learning theorists in behavioral neuroscience talk about cognitive psychology. I can guarantee you we're not as harsh in return.
Oh, no offense taken! Sorry if it came across that way! I describe myself as a dinosaur at times!
Around here, there is quite a lot of good-natured ribbing both ways, but at times over the years some individuals have gotten quite nasty. I always figure that, if the two sides are going to battle it out *using evidence*, then whatever their motivation, in the long run science wins. But I would much rather someone admit their ignorance (or lack of interest) than pretend that their own ignorance accurately reflects the field. We get enough of that from anti-evolutionists!
Rescorla was the chair of psychology at my undergrad institution. We had a weekly journal club that he and Gallistel both regularly attended. They were two of the smartest motherfuckers I think I ever met.
In Latin America - especially in Brazil - we're very much hostages of pre-1950s discussions in Psychology. As such, behaviorists are still pretty much a dominant species in here - the very image of so-called "scientific" approaches to Psych... I once saw a talk by a behaviorist (which, surprisingly, was a student under Daniel Dennett) that condemned neuroscience and favored a return to Skinner's tenets. Weirdo.
Behaviorism, I think, is hardly dead, nor does it deserve to die. When we think of learning (at least, in the Hebbian sense) our language I think is inherently behavioristic. The behaviorist sees an organism whose disorganized or "innate" efforts are gradually shaped by an environment which only supports a certain limited range of successful outcomes. Certain behaviors gain strength, others are weeded out, others are simply refined by ever tightening contingencies of reinforcement. But this seems to be the picture of how synaptic plasticity is thought to cause learning at the level of the brain. Certain stimuli recruit certain constellations of neurons and, under the control of the external consequences, some constellations of synapses gain strength while others weaken. The result is the gradual tuning of previously general purpose circuits to a more task oriented organization.
Forgive me for glossing over such an enormous (and still growing) field. But my point is that when we look to explain learning, we invariably have to appeal to the effects of the environment on neural circuits in one way or another. The logic of Skinnerian behaviorism is analogous to the logic of Darwinian evolution, in that the action of the environment on essentially random variations is the determiner of ultimate form. So much more could be said about this, but I have not the space here to enter into it. But this logic does save us the trouble of ascribing creativity to the marvels of either the human brain or the natural world; each can produce marvelous forms through selection.
The "cognitive" in cognitive science may simply be a shorthand for the effects of an especially complicated environment. Studies of animal cognition still use what could basically be described as operant conditioning. And while some are obviously capable of learning tremendously complicated tasks, and some are obviously capable of behaving in amazingly appropriate ways with hardly any trial or error, the basic datum is still behavior and nothing else. The animal learning language is not taught "concepts", it is simply placed in an environment which best supports the types of behaviors to which "concepts" are frequently attributed as a cause. Of course in the lab, we can point to the conditions arranged in the environment as a more useful surrogate to mentalism*, but that is not the trend in today's research.
Anyhow, nobody thinks the organism is empty. But it's important to recognize that what we study as the biology of cognition may just as easily be called the biology of behavior, and perhaps more appropriately so.
*I wrote this and realized that yeah, we do sound pretty dated. Oh well.
Are behaviorists all that rare? All economists are more or less hardcore behaviorists, though with neuroeconomics that's changing probably.
Physio, oh yeah, Rescorla's a seriously bright guy.
Could "the stimulus learned" have meant "the stimulus was learned [by the animal]"?
Even without context I do believe you...learning theorists haven't changed their terminology very much in the past 50 years.
I did some work with the Rescorla Wagner model and it is very "cognitive". Its just that learning theorists when talking of the mind, don't make a distinction between "complex associations" and "representations". At the level of mind they deal with, I have trouble making this distinction myself. Also, aren't connectionist models and computer simulations of competing learning models inherently cognitive?
The "animal" is seldom mentioned at these talks because those do this kind of work assume that all intelligent systems are based on the kinds of simple learning mechanisms. The particular organism is irrelevant (for them).
I was not at the talk, but Rescorla *IS* an "external" cognitive-behaviorist - so I would agree with Anon. I had the pleasure to meet Rescorla on several occasions when I was working at Rutgers University Center for Cognitive Science and I would agree that he and Randy Gallistel, who was the director of RUCCS, are wicked smart!
I would love to mention that Gallistel occupies a weird intellectual position between straight up behaviorism and hard-core cognitive science. Conversations with Gallistel are clear in how these two fields mesh for him, but talking to his students can require consider patience, because, well, frankly, the two positions are fundamentally oppositional.
Of course the real problem - and I know that this comment I am about to make is opening a discussion that our wonderful blogger didn't intend to open, but whatever - is that Behaviorists, like John Watson, deny the very existence of consciousness. But then again, information processing types, who moved from the pigeon/rat metaphor to the computer metaphor are not much better. ï¿½In fact, the [conscious] experience itself may well be epiphenomenal with regard to mental activity. The experiences may play no part in how thinking actually gets done - just as the flashing lights on the outside of a computer play no part in the processing going on insideï¿½ (Kosslyn, 1983). Dennet, Pylyshyn, Fodor, Hofstadter etc., etc., etc. all agree with Kosslyn.
I have a similar reaction to people who think that narratives, themes, symbols, conflicts, defenses and compromises are irrelevant to an understanding of the mind.