Some more thoughts on Behaviourism

Ken writes:

I’ve been trying for some time to write a post on Behavourism because it’s something that holds a perennial fascination for me. I have a number of B F Skinner’s books on the go About Behaviourism, Science and Human Behaviour and Verbal Behavior. This last is the book Noam Chomsky made his name reviewing. It’s very heavy going, but it is brimming with interesting ideas (although a proper behaviourist would never put it in those terms). Anyway, I twisted my ankle this weekend and it can’t really bear my weight, so Alice has cleared her day at work to look after the kids so I can sit with my leg up and blog.

Most people have only a very vague and possibly caricatured notion of what Behaviourism is. I remember a joke about behaviourism doing the rounds when I was at university. What did one behaviourist say to the other after sex? “That was good for you. How was it for me?” Implying among other things, that behaviourism has no account of first-person experience or introspective self-knowledge. In fact, Behaviourism doesn’t deny the reality of feelings and experiences although it does challenge the traditional account of how we should understand them and their place in psychology.

Behaviourism seeks to understand an organism’s behaviour in terms of a history of differential reinforcement or punishment that shaped that behaviour. Behaviour assumes its specific form because in the past certain forms of behaviour were rewarded and certain other forms were punished. For example, a tennis player’s serve takes the form it does because tossing the ball so high and swinging so hard cleared the net and landed in the square in the past, whereas tossing it slightly differently and swinging slightly differently either didn’t clear the net or didn’t land in the square. Those movements of the body that resulted in a legal serve were selected by their good consequences for the player and reproduced in subsequent behaviour. It is the environment that determines whether the consequences are good for the player. If the rules of tennis were different, and a different trajectory of the ball were highly prized, a different combination of movements would be selected. This is a different sort of explanation from the one that goes ‘because that’s the way the player wants to hit it’, but, the behaviourist argues, it’s ultimately more satisfying.

In the usual run of things, people explain something like the way a tennis player serves the ball in terms of the beliefs and thoughts and attitudes of the player. An explanation that appeals to the player’s intentions is a regress stopper. It gives the explanatory process a stopping point. You can but you don’t have to ask why the player chose to play the shot that way, because human agents are taken to be capable of spontaneous action (that is, genuinely initiated action, action without a previous cause).

If I understand him correctly, Skinner denies that human agents do truly behave spontaneously. He thinks that all behaviour has a determinate cause (or allowing for irreduciably probabilistic causal connections in atomic domains, all behaviour is made highly probable by some antecendent event). Human actions are never the result of spontaneous (i.e. uncaused) decisions by human agents. So behaviourism is deterministic. Determinism is a hard doctrine to counternance, because it implies that human actions are never free. They are the ultimate consequences of chains of causal interactions beginning aeons and aeons ago in the distant past. If each link in the chain is causally sufficient to bring about the next, then our actions were already going to happen before we were even born.

I happen to think determinism is more plausible than the alternative, which holds that at some point there was a gap in the causal order that required the intervention of a non-physical causal agent to overcome. All my actions, whatever their wider meaning might be, are at one level the movements of the parts of my body (nerves and muscles and that sort of thing). It is incredible to me that these physical body parts become activated without some other physical cause.

Behaviourism implies determinism but not vice versa. You don’t have to be a behaviourist if you’re already prepared to be a determinist. Most philosophers and cognitive scientists would be determinists but not behaviourists. They think the common sense psychological explanation of human action in terms of beliefs and desires and that sort of thing is actually compatible with determinism because the mind is somehow just another piece of the still not very well understood physical causal story. Behaviourism is radical in seeking to reject the explanation of behaviour in terms of these familiar categories (of belief, desire and so on).

I think the strongest reason to take this sort of explanation seriously comes from the parallel between behaviourism and the theory of evolution. Both theories see current forms of phenomena as the result of historical processes of selection. Behaviour takes its current form because in the past behaviour it resembles was reinforced and dissimilar behaviour was not. The organism takes its current form because in the past organisms with that form survived while those without it did not. The mechanisms of shaping and selecting phenomena are different, but complementary. Natural selection works by a sort of winnowing out of unsuccessful forms, whereas behavioural selection increases the probability of reinforced forms. A certain trait has a survival value so subsequent generations of organism that inherit that trait are created and over time the trait spreads through the population as more and more of the subsequent generations have it. In behaviour modification, actions that were reinforced are simply repeated which happens at the expense of untried forms that otherwise might have been emitted. Behaviour modification relies on the survival value of being disposed to find certain things rewarding (the attention of other people, smiles, physical affection, certain tastes and so on). These dispositions ensure that certain actions are rewarding so they are repeated when similar circumstances recur.

Evolutionary explanation and the explanation of behaviour in terms of a history of reinforcement both do away with explanation in terms of the decisions of an agent. Evolutionary theory has rid us of the need to see the variety of species of animals as the result of the creative decisions of God, and behaviourism would rid us of the little god in the human mind for explaining human behaviour. (This reminds me of another joke about behaviourism. The American philosopher Sydney Morgenbesser, on having behaviourism explained to him, apparently said, ‘So what you’re saying is: Don’t anthropomorphize people!’ That’s astute. It’s absolutely spot on, but it’s not a criticism of behaviourism. Behaviourism is a rival to traditional belief-desire psychology.

It can sometimes seem as if the behaviourist explanation presupposes the kind of thoughts and feelings and inner goings-on that the behaviourist officially disavows. For what is rewarding but positive feelings of satisfaction? Chocolate is rewarding because the boy likes the taste. Praise is rewarding because the boy wants to feel loved and valued. And so on. Skinner’s answer to this criticism is that the feelings, while genuine, aren’t needed to explain the mechanisms of the reinforcement of behaviour. He appeals instead to the survival value of dispositions to find certain things rewarding. Organisms with these dispositions were able to learn to respond to the environment in effective ways, and this conveyed a selective advantage (“About Behaviourism”, p.52). It is a corruption of the behaviourist explanation of things to say the agent is motivated to behave in such and such a way because they like the feeling the reward brings. That is smuggling a homunculus into the explanation that doesn’t need to be there.

Its easy to see the principles of classical and operant conditioning at work in utterances Frank makes. For example, sometimes he says Good Boy to himself (he actually says ‘Bo Boy’, but that’s his version of Good Boy). A behaviouristic explanation of this might go something like this. He does things that we reinforce by saying ‘Good Boy’. He is disposed to find our warm smiles and attention rewarding and purely by association is conditioned to find ‘good boy’ itself rewarding. So it is slightly rewarding when he says it to himself. So he does. So there’s an explanation of how the behaviour came about and continues, but if we scale this sort of thing up to adults, we seem to lose hold of a distinction between learning and proficient stages. I mean, you can see how someone could acquire behaviours by a training process of reinforcement, but when does it end. Is the behaviour of a competent tennis player still under the control of the reinforcing environment? Does the professional still play the shots as they do for reward and reinforcement? I don’t know what the behaviourist would say about that, but I suspect that the answer would be that there is no genuine distinction between learning and proficient behaviour. Learning never ends. Proficient behaviour is highly effective at getting its reward. So effective it doesn’t feel like learning anymore, but it is still operating according the the same principles. Or maybe the distinction needs to be recast in different terms. Perhaps proficient behaviour is rewarded on a more intermittant schedule and perhaps it is rewarded by other sorts of reinforcers than the ones it was trained up with.

In spite of the unsettling consequences, I think behaviourism must be true of human behaviour. For on the one hand, it is easy to see how it applies to some aspects of human behaviour, and on the other hand it’s hard to imagine circumstances that violate the rules of behaviour actually happening. If you get rewarded for doing something, you’re more likely to do it again in the future. Things that don’t pay, people tend to stop doing. But now what about the exceptions? Well, are there any? If I do something that is rewarding, e.g. well remunerative, what would interfere to make me unlikely to do it again in the future? The contingencies of reinforcement might change to make it no longer pay, or alternatively, I could find some incompatible actions more rewarding (for example, a woman who is a successful lawyer might leave her job to be a full-time mother because she finds it more rewarding). In general, it is hard to conceive of a case where someone finds something rewarding but doesn’t do it (unless they find an incompatible action even more rewarding. Actions are ‘incompatible’ if and only if you cannot perform both). So what would make behaviourist principles stop working?

23 thoughts on “Some more thoughts on Behaviourism

  1. mairij

    Does Behaviourism account for creativity – for entirely novel forms of action or modes of expression. I understand that one of the elements of a good rugby team are that at least some of the players are capable of novel, unexpected behaviour. But novelty also involves risk – risk of failure, rejection or punishment.

    1. ken

      I think so. Creative action isn’t action that is uncaused, it’s simply novel. It’s an effect type that hasn’t occurred before. But that can be accounted for by the change in circumstances. Behaviour happens in a context. If the context is different, then it is to be expected that behavioural dispositions developed in one context will play out differently in the new one.
      Behaviourists talk about generalisations of behaviour. If eating red fruit has been rewarded in the past, this also raises the probability of eating other kinds of non-green fruit. If striking with an open hand has been rewarded in the past, kicking and striking with a closed fist have also been made more probable. In other words, reward and punishment also operate on a host of tacit and unexpressed behaviours which were alternatives to the actually expressed behaviour. Thus there is scope for changes in behaviour when the context changes and something that was rewarded in the past ceases to work.

  2. Interesting read. I wonder how self-destructive behaviour can be explained. I suppose substance abuse gives immediate satisfaction but over the long-term is not a beneficial route to take. Maybe many humans find it hard to see the long-term benefits of certain behaviour and can only focus on short-term. Maybe that differentiates between people who achieve goals and ‘success’ and those who don’t? You’ve got me thinking now.

    1. ken

      I think there’s a lot behaviourists can say about self-destructive behaviour. Your guess is right that it has to do with the immediately rewarding consequences. But Skinner says other things too. For instance, he uses gambling addictions to illustrate the effects of different schedules of reinforcement. It turns out that you get much higher rates of response when the subject is not rewarded every time they perform the action but only every n-th time, or on average every n-th time. They did experiments with animals where they would gradually require the animals to respond more and more times for the same reward. Subjects who have been habituated to working hard for their reward in this way persist in the behaviour far longer after rewards have stopped then those that haven’t.

  3. gazza

    Do you think it’s true of verbal behavior — that is, speech? (For example, is my production of these words in this comment to be explained by my history of reinforcement?)

    1. ken

      It’s hard to say, isn’t it. If I recall correctly, Skinner was apparently stimulated to work on verbal behaviour because AN Whitehead asked him what was the cause of his saying “No black scorpion is falling upon this table” at a Harvard dinner one time. The book ‘Verbal Behaviour’ is just a speculation regarding how the principles of behaviour might apply in the case of verbal behaviour (i.e. it’s not the result of direct empirical studies as such, but the extrapolation from behaviourist principles). But I think Chomsky’s review of “Verbal Behaviour” doesn’t provide any cogent criticisms of the approach. There’s an excellent rebuttal by Kenneth MacCorquodale.
      The thing to say about your sentence is that it is the effect of multiple causes. It has a grammatical form that was selected by the contingencies of reinforcement. It has vocabulary that was selected by the contingencies of reinforcement. It has a pragmatic relevance that was selected by contingencies of reinforcement. And so on. But it’s usually not possible to say what those causes were because they are lost in the past.

  4. gazza

    I didn’t know there was ever a reply to Chomsky’s review. Might give that a read. Still, though, it seems to me that the way you’re setting things up, Behaviorism can become true only at the cost of becoming trivial. Take your last paragraph, where you consider whether there could be a case where S finds action A to be the most rewarding at the time (of the available actions), but still doesn’t do A. You imply that there couldn’t be such a case. Now, what is your definition of ‘most rewarding’ here? Perhaps it is ‘most reinforced in the past’. But in turn, that depends on what you mean by ‘most reinforced’. Now on the face of it, there are plenty of cases where a person does, in fact, do something that is not the most reinforced action available to them — most notably, where the action is entirely novel. Now, you could try to save your claim by refining the meaning of ‘most reinforced’ so that, say, the purportedly novel action is composed of parts each of which has been reinforced in the past. But (a) I’m very unconvinced that this will always be true, and (b) even if you manage it, I’m worried that the result will basically just be a collapse into the thesis of determinism — which is not the thesis that the behaviorist is actually trying to defend.

    1. ken

      I’m not sure I follow your worry that behaviourism will collapse into determinism. It certainly is deterministic, but there’s more to the view than that. But the difference from behaviourism doesn’t have to make itself felt here at this point, right? If the behaviourist explanation of something coincides with what a bog standard (non-behaviourist) determinist would say, that doesn’t show behaviourism itself collapses into determinism.

      I touched on novel actions in my reply to mairij above. Reward reinforces not only the actions that brought the reward about but also an indeterminate range of alternatives that might have brought about the reward if the actual behaviour hadn’t done so. For example, if a subject was rewarded by turning on a light switch, and in the past they have always done this right handed but in the current circumstances are prevented from using the right hand by an arm sling, they may use their left hand, or right elbow or perhaps something else as the reward also raised the probability of these alternatives. This is how behaviourists are able to train animals to perform very specific behaviours. They start by rewarding anything that is generally like what they want the animal to do (maybe just standing in one half of the cage) and then get more and more particular in their requirements. The animal finds that what it was doing no longer works and they have to try other actions until something is successful. This tacit reinforcement of penumbral potential actions is a different route from the compositional one you suggest for dealing with novelty, which goes some way to meet your point a), although for verbal behaviour the compositional solution looks promising.

      1. gazza

        Take the light switch case. (I assume, btw, you meant to say ‘rewarded FOR turning on a light switch’.) You suggest that when S is rewarded, the rewarded behavior is not just the act of flipping the switch with the right hand, but the more general act of flipping the switch. OK. How do we know what act, exactly, has been rewarded? Is it the act of flipping the switch up? Or just flipping it from one position to the other, regardless of initial position? Or is it the act of making the light come on? If we supply S with another way of making the light come on — say, a button — has S in fact been reinforced for doing that already, by virtue of the reward received for flipping the switch? You might say, “No, because S doesn’t yet know that the button will turn on the light.” Indeed. But then how does S know that the switch will turn on the light if flipped down as well as up? Or if flipped with its left hand as well as with its right?

        The general point, which I’m sure you’re familiar with, is that unless we consider what’s going on inside the creature’s mind, we have no way of telling what the creature has been reinforced for.

        Of course, none of this denies the point that a creature’s behavior can be shaped in just the way you describe. Indeed it can. Methodologically speaking, one can shape a creature’s behavior quite effectively using such methods. But that doesn’t establish the sort of behaviorist claim you’re trying for. It only shows that the creature has certain internal states which allow it to have some sort of grip on what sorts of actions are and are not related to, or similar to, ones that it has previously carried out. Thus, it might figure that flipping the switch up is similar to flipping it down, and try that; but it might see insufficient similarity between the button and the switch, so that the button would not become a behavioral target. Of course, a _different_ creature might see that similarity and try the button. But the point is, we’ve gotten entirely away from reinforcement history here. We’re talking about cognitive differences.

  5. ken

    I think this is a really interesting disagreement. I’m certain Skinner would have nothing to do with the suggestion that to know what is being reinforced you have to look inside the subject’s mind and determine how they understand/identify their action. That’s obviously a big big no no.

    I think the position would rather be that it is in effect only after the fact that we can identify what was reinforced. In fact, it’s only after that action and subsequent actions so that the individual instance can be seen as a regularity. Reinforcement is essentially a historical process. Speaking on an individual occasion you’d have to say what is rewarded is EVERYTHING the creature did before getting the reward (including the context in which it happened). So that is using the right hand, flicking up, touching the switch, and so on and so on. It’s only after the behaviour is repeated in different circumstances that the theorist can say it was this aspect of it that was responsible for the reward and not that.

    It occurs to me that Fodor’s recent criticisms of natural selection are very similar to what you’re saying about individuating what is reinforced. His argument turns on it not being possible to look inside Nature’s mind to see what it’s rewarding. If I’m right that would be another connection between the two theories.

    1. gazza

      “Speaking on an individual occasion you’d have to say what is rewarded is EVERYTHING the creature did before getting the reward (including the context in which it happened).”

      But that can’t be right — because NOT every action the creature was doing at that time (just before the reward) is an action that will increase in frequency.

      I haven’t read Fodor & Piattelli-Palmarini’s book, but apparently they themselves make the analogy between behaviorism and natural selection totally explicit. They think they’re basically the same view dressed up in different terms. And so, just as behaviorism is false, so is Darwinism. A nice review which discusses this point is Godfrey-Smith’s: As he points out, the analogy is not perfect. Behaviorism may be wrong even if Darwinism is right.

      I think the key difference might be as follows. G-S discusses the example of an organism with a coat color U which has poor camouflage and thus gets eaten, while one with coat color T does not. We can trace the causal path from the former organism’s coat color to its ending up as dinner, via U’s effect on the predator’s eyes, visual system, etc. Of course, we may not be able to pick out this particular causal path as the important one until a long time after the fact, once we see that organisms with coat color T increase in the population. Now compare the selection process for behavior. Since, as you noted, there are very many behaviors that the organism was doing just prior to the reward, we need some way of saying which one is actually reinforced. Again, we may not be able to tell until a long time after the fact, when we see what behaviors increase in frequency. But the fact will remain that the causal path connecting the reward to the behavior will be mediated by cognitive processes inside the organism. We will see this because we will see that different organisms in the same situation (and thus the same reinforcements) will develop in behaviorally different ways.

      So I think the fact that in both cases the selection process can only be seen in retrospect isn’t important. What’s important is that in the case of behavior, we see that the selection process involves factors that are independent of the creature’s reinforcement history. Thus behaviorism fails because it tries to leave out a significant part of the causal story about how behavior happens. By contrast, evolutionary theory involves no similarly restrictive claim about how traits come to increase in a population. When we discover (say) that creatures with coat color T survived while those with U became dinner, we simply say that that very set of events constitutes selection for T and against U.

      1. ken

        Skinner would agree that the full causal story of the etiology of behaviour will involve processes that occur within the skin (the brain and musculature etc). But your calling these ‘cognitive processes’ is tendentious because it identifies two ways to understand these processes. They can be understood as physical processes or as mental processes. You can’t insist that the causal story about behaviour has to make use of mental processes. You can’t assume that the processes are simply the beliefs and desires (etc) of traditional psychology. The concession that part of the causal story concerns brain events doesn’t undermine behaviourism. The connection between the environment A and physical bodily processes B is lawful. So is the connection between bodily processes B and subsequent behaviour C. So there must be lawful connections between A and C the environment and subsequent behaviour. Behaviourism focusses on lawful AC connections. At most missing knowledge of B means AC laws will have to be probabilistic instead of exceptionless. I don’t know how far our understanding of brain processes has come since Skinner’s day, but the only way it could undermine it would be to discover neural states that were identical with mental states. Has that been done yet?

        p.s. the fact that two animals in the same situation can behave differently is due to different reinforcement histories. Yes there will be internal differences too, but that shows up in reinforcement history. You can teach a dog to do tricks but you can’t teach a cat. Cats just don’t have the dispositions to seek to please their masters that dogs have.

  6. gazza

    “The connection between the environment A and physical bodily processes B is lawful. So is the connection between bodily processes B and subsequent behaviour C. So there must be lawful connections between A and C the environment and subsequent behaviour. Behaviourism focusses on lawful AC connections. At most missing knowledge of B means AC laws will have to be probabilistic instead of exceptionless.”

    But what if the BC laws are _less_ probabilistic (that is, have fewer exceptions) than the AC laws? That is, what if we get better predictions of an organism’s behavior by basing our predictions on its present internal states than on its past reinforcement history? I think, of course, that cognitive science has shown that this is indeed the case for complex organisms.

    Perhaps you’ll simply reiterate that the internal states are themselves merely determined by reinforcement history. My response is to demand the evidence. That internal states are determined by reinforcement history has to be proven by experimentation. And I take it that the attempt to prove that claim failed, especially concerning language.

    This is where my earlier comment about a collapse into determinism comes in. In your argument you say “environment” instead of “reinforcement history”. Do you mean the latter? If you don’t, then you’re just saying that all behavior is determined by _something_ outside of us. I won’t disagree with that, but that’s not behaviorism as I understand it. Behaviorism makes the specific, empirical claim that behavior is determined by reinforcement history — right?

    I think you might now say that there _is_ evidence for the connection between reinforcement history and current internal states, but only in retrospect: that is, only after the fact can we tell what behaviors the organism was reinforced for. So when the cognitive scientist says what internal states of the organism caused its current behavior, the behaviorist steps in and says that those internal states must (therefore) have been the ones that the organism was reinforced for having in the past; so the history of reinforcement predicts the internal states; QED. But this is purely ad hoc. It trivializes the theory.

    1. ken

      But what if the BC laws are _less_ probabilistic (that is, have fewer exceptions) than the AC laws? That is, what if we get better predictions of an organism’s behavior by basing our predictions on its present internal states than on its past reinforcement history? I think, of course, that cognitive science has shown that this is indeed the case for complex organisms.

      Yes, that’s a fair point, to the extent that we can examine the internal states directly (and not simply infer their presence from the behaviour they’re supposed to cause). But this doesn’t substantiate traditional belief desire psychology unless those states can be identifed as traditional beliefs and desires (and other traditional mental states). It might be the case that the Churchlands are right and direct knowledge of the neurophysiology of behaviour will undermine the traditional psychological concepts. Behaviourism and eliminative materialism could work in tandem to account for the traditional phenomena of psychology. I think there’d still be a role for behaviourism as a practical technology.

      One reason I don’t think beliefs will ever be identified with brain states is the strong rounds for externalism about content. There are good reasons for thinking that you couldn’t have a thought like ‘the cat is on the sofa’ in a world without our actual linguistic practices. Also the holistic and dispositional nature seems to present a problem for traditional psychology. It’s simply indeterminate how many beliefs a person can properly be said to have at any one moment because you can’t simply take what a person would sincerely avowal as delineating the extent of their beliefs.

      Assuming for a moment that belief desire psychology has a part to play in understanding people, I think the best way to construe what beliefs etc are is not as physical mental states at all but as social normative statuses. I think that’s the biggest contribution Robert Brandom has made to contemporary philosophy.

  7. Re: Creativity – YES. In a famous experiment, Karen Pryor began rewarding a dolphin for unique behaviors – the dolphin was only given a tweet and a fish if it offered something new. At first it got rewarded for simple things, such as a flipper wave or a leap. But as time wore on the dolphin had to get more and more creative, eventually becoming more and more elaborate and unusual in its behaviors – swimming backwards, creating complex routines and so on.

    That being said, Skinner and Watson were both extremists. We now know that there are certain genetic programmed responses. For example, they have found that a squirrel raised in an all-cement environment and fed nothing but scientifically bland food will, at first opportunity, grab a nut, run away with it, and attempt to bury it.

    Of course, our individual personalities are also driven by WHAT rewards us. Attention can be rewarding to an extroverted and confidant personality, while it can be punishing to a shy and retiring one, for example. Thus the same consequence can result in wildly different behavioral responses between two individuals.

    1. kenanddot

      Thanks for the comment. I meant to reply sooner. That’s fascinating about rewarding creativity. If Skinner is an extremist, it would be interesting to know more about approaches that try to take the middle ground. I would have thought Skinner could incorporate instinctual responses and the variation in what people find rewarding the same way; simply by saying that you have to take some dispositions as primitive. Doesn’t he take it as a brute fact for his purposes that we find sweet tastes more rewarding than bitter ones, for example, despite the fact that there’s obviously got to be an evolutionary explanation? I mean is he really competing with physical, neurological theories?

      1. In a way. Watson and Skinner were both what I would call radical behaviorists, and they tried to extrapolate the results of their research to higher level behaviors. Much of their extrapolations have since been effectively disproven or at least called into question.

        Watson rejected the idea of internal mechanisms as functions of behaviour, such as thoughts and feelings. Obviously, he knew that they existed, because he himself could feel his own, but since they were not measurable, he did not believe that they could or should be studied. He dismissed “private events” as “epiphenomena” and tended to view all members of a species as tabula rasas- he famously made a claim that if you gave him a random child, and the power to control all aspects of the child’s environment, he could raise a doctor or a thief, as he chose.

        We now know that this is not particularly true. Often identical twins separated at birth and raised in different families show remarkable similarity to each other despite wildly different environments. Clearly genetics have more of a sway on behavior than Watson anticipated.

        Skinner was slightly different. He did not reject the study of internal events, but he considered them to be behaviors in and of themselves, and therefore not valid explanations FOR behavior. While he would have acknowledged that there are different kinds of reward, he argued that rewards are based on physiological drives, such as the drive for food. He also argued that all learning is based on one of these drives.

        He also believed that highly complex behaviors, including thoughts and feelings, were primarily related to environment, and tried to extrapolate to processes such as language. Noam Chomsky took Skinner’s arguments apart in several extremely thorough papers, which are highly lauded.

        Some points he made include:

        “If a drive is postulated on the basis of the fact that learning takes place, the claim that reinforcement is necessary for learning will again become as empty as it is in the Skinnerian framework. There is an extensive literature on the question of whether there can be learning without drive reduction (latent learning). The “classical” experiment of Blodgett indicated that rats who had explored a maze without reward showed a marked drop in number of errors (as compared to a control group which had not explored the maze) upon introduction of a food reward, indicating that the rat had learned the structure of the maze without reduction of the hunger drive.[…] Results of this kind can be handled by reinforcement theorists only if they are willing to set up curiosity, exploration, and manipulation drives, or to speculate somehow about acquired drives24 for which there is no evidence outside of the fact that learning takes place in these cases. ”

        “Imprinting is the most striking evidence for the innate disposition of the animal to learn in a certain direction and to react appropriately to patterns and objects of certain restricted types, often only long after the original learning has taken place. It is, consequently, unrewarded learning, though the resulting patterns of behavior may be refined through reinforcement. Acquisition of the typical songs of song birds is, in some cases, a type of imprinting. Thorpe reports studies that show “that some characteristics of the normal song have been learned in the earliest youth, before the bird itself is able to produce any kind of full song.”

        “Even a very young child who has not yet acquired a minimal repertoire from which to form new utterances may imitate a word quite well on an early try, with no attempt on the part of his parents to teach it to him. It is also perfectly obvious that, at a later stage, a child will be able to construct and understand utterances which are quite new, and are, at the same time, acceptable sentences in his language. Every time an adult reads a newspaper, he undoubtedly comes upon countless new sentences which are not at all similar, in a simple, physical sense, to any that he has heard before, and which he will recognize as sentences and understand; he will also be able to detect slight distortions or misprints. Talk of “stimulus generalization” in such a case simply perpetuates the mystery under a new title. These abilities indicate that there must be fundamental processes at work quite independently of “feedback” from the environment. ”—-.htm

        I’m not entirely a Chomsky fan, because his arguments are often taken to mean that humans have an innate language mechanism which other species cannot learn (the chimpanzee Nim Chimpsky was originally named by researchers as a way of thumbing their nose at Chomsky – sadly, it ended up actually proving Chomsky’s point – despite much anecdotal evidence that Nim did acquire and use language in his home environment, the carefully constructed lab instruction actually resulted in a Clever Hans syndrom, thus Nim’s capacity for language was scientifically “disproven” in laboratory – a frustrating result for the researcher, who knew perfectly well that Nim understood language on a casual basis).

        We now know that behaviorism explains much, but not all, behavior!

  8. kenanddot

    The thing is, I think Chomsky’s success with his 1959 “Review of ‘Verbal Behavior'” owes more to sociological reasons than merit. As applied to language, it’s hard to understand and unpalatably reductive for most people. It is explicitly committed to determinism. Chomsky flattered people that human language and thought is unique and exceptional and provided a space for human spontaneity in the mind. Chomsky also said it was possible to study the mind in a scientific way without being scientistic. So basically, I’m saying Chomsky told people what they wanted to hear, so it’s no wonder they agreed with him.

    But I would like to learn more about what behaviourists say about phenomena like imprinting or a fish’s instinct to swim back to the river where it was spawned to lay eggs. It’s hard to believe they would simply deny the existence of unlearned behaviours like that. It seems to me they should say simply that when a behaviour is learned, it is learned through the stimulus response framework of operant conditioning. Behaviourism is much more pertinent to human behaviour because the existence of innate behaviours is much less clear. And isn’t that what Chomsky was trying to argue, that humans have a language instinct?

    1. Yes and know.

      Chomsky argued that humans have an inborn language building mechanism and to a certain extent he was right – my psycholinguistics course spent a lot of time discussing the universal grammar used by toddlers, and the newborn’s ability to perceive subtle differences in verbal sounds (which is pruned out later on as we age) as well as the Wernike area of the brain, which is much less developed in other creatures. Clearly, humans DO have an innate capacity for language, particularly figurative language, the way that dogs have an innate capacity for hunting.

      But, of course, it is simply a CAPACITY for language. It is still a learned behavior. A child not exposed to language early on, such as Genie, loses that capacity and will always have difficulty with complex sentence structures as well as figurative speech.

      BUT, Skinner’s attempt to explain language using behaviourism was a poor one at best. Subsequent studies (including the studies of Genie) show that while operant conditioning can be used to help the acquisition of speech, particularly in autistic children who require more extrinsic motivation than other children, it does not explain all the aspects of acquisition, such as the critical period of language acquisition.

      The mistake that Skinner and Watson (and Chomsky!) made was the mistake that many theorists make – the belief that their theory is all encompassing. Just as Newtonian physics do not apply on the subatomic level, behaviorist theories also break down when attempts are made to apply them outside of their natural sphere. Skinner once said that we don’t need to study the brain, since operant conditioning explains everything.

      Most psychologists and even behaviorists these days would agree that this is nonsense, and that many important things have been discovered and will be discovered through study of the brain.

      Study of the brain has revealed that we do have a section of the brain devoted to language acquisition. But our environment, schedules of reinforcement, and many other factors also play a part.

      The truth always seems to lie in the middle.

      As a dog trainer, and as someone whose Psychology degree is made up of every conditioning and behaviour course offered at her university, I admit myself to be a fan of behaviorism. I feel that behaviorism IS an elegant theory which reduces the need for many others. For example, dominance theory is virtually unnecessary, since all behaviors caused by dominance theory can also be explained via classical and operant conditioning – and so Occam’s razor tells me that dominance theory is crap.

      Also, I love that we are still finding new ways to apply operant conditioning, such as through cognitive behavior therapy. I was delighted to learn ways that I could “train” myself out of my own anxiety, and to be shown how I was subconsciously rewarding my on anxious behavior.

      But Skinner attempted to use behaviorism to explain many behaviors which are more innate than he was willing to admit at the time.

      The publications of the Brelands, who had tried to use operant conditioning on circus animals and frequently found that basic instincts overpowered conditioned behaviors, proved that instinct is a powerful force, and that much more behavior is driven by basic genetics than previously was thought.

      Since then, ethologists have discovered fixed action patterns, such as the squirrel one I described. The greylag goose is another commonly cited example of an animal exhibiting complex behavior which is inborn and not easily altered through operant conditioning. These did not refute the claims of Skinner, but added a new dimension to the way we perceive behavior. We also know that personality traits such as fearfulness can be inherited from parents, and separated twin studies have shown some very interesting results which suggest that many things are much more genetically based than Watson or Skinner would ever have believed.

      Skinnerian conditioning is elegant, but it is not the be all and end all.

      Modern behaviorists now take a more holistic view, believing that behavior and learning are complex interactions between genetics, brain structure, temperament (once believed by Watson to be entirely based on environmental conditioning, but now known to be an inborn tendency which can sometimes be overcome by conditioning) and environment.

      Skinner and Watson were right on the money in many ways, and I believe that many people would be better served to apply some basic operant conditioning to their lives. Many of the dog owners I have worked with have actually been rewarding negative behaviors in their dog, and it takes a surprising amount of work to help them to see that and correct it!

      Nevertheless, I do feel that Skinner was overstretching himself with Verbal Behavior. Or, perhaps I should say that his explanation of Verbal Behavior has since been found to be incomplete, and that some of Chomsky’s arguments certainly have merit. Modern researchers of linguistics combine a cognitivist and a behavioral approach when dealing with language acquisition, but there is still a certain amount of in-fighting.

      As an amusing side note, just as much arguing goes on within the ranks of behaviorists – for example, there is the subject of maximization vs meliorization – Maximization proponents won’t speak to Meliorization proponents at conferences.

      1. kenanddot

        Did Skinner over stretch himself with Verbal Behaviour? I would love to know. It is such a densely written book and the ideas are so strange and unfamiliar that I find it really hard going. But I do think it is really the centrepiece of his theory, because so much psychology is, you know, that inner speech. I don’t believe a behavioural theory that didn’t address the inner speech/ thinking of agents could possibly be adequate.

        I think there are good reasons from the perspective of the philosophy of language to attempt something like Skinner’s reductionist account of verbal behaviour. It’s filling an important niche as an account of language that doesn’t stipulate meanings and can deal with indeterminacy of reference.

      2. Yes, I believe he overstretched himself. Skinner was further ahead than Watson, who believed that thoughts were meaningless and irrelevant, while Skinner views thoughts as behaviors. It is this view which lead to cognitive behavior therapy, which has been extremely useful in treating many mental health disorders. So he was on the right track.

        But his attempt to explain how we acquire language is full of holes. Chomsky pointed out many of them, and research conducted since has backed up some (though not all) of Chomsky’s points (as I’ve said, I’m not a Chomsky fan, but the man was right about a couple of things).

        Some examples of language acquisition which cannot be explained by Skinner’s model:

        The acquisition of grammar: Skinner proposes that we acquire correct speech through positive and negative reinforcement. We swear, we get corrected by our parents, we learn no to swear. We say “I runned away”, we get corrected, we learn to say “I ran”. Unfortunately, research has since shown that parents rarely correct their children in day-to-day interactions, but that children acquire the grammar correctly after a while anyway. In fact, children often start out saying “I ran” and then suddenly backslide into overgeneralization, adding “ed” to any past tense verb. This marks the point where the child has left echoic speech (merely imitating what his parents have said) and is developing a concept of basic grammar. It takes even more time for him to pick up on the exceptions to the rule and develop correct speech again. The behavioral approach doesn’t explain this very well. There are many newer models of language acquisition which are based on behavioral principles which do manage to explain this by incorporating concepts from cognitive science.

        Generative language: Chomsky’s famous example was “colourless green ideas sleep furiously”, showing that we can recognize a novel and nonsensical sentence as grammatically correct. His point was that we are capable of creating AND understanding sentences which have never been said before (in linguistics class I watched a great video with George Carlin coming up with novel sentences) such as: “Put the zebra in my back pocket” or “I’m running late for my Skype session with God”. Both of those sentences are intelligible (if unusual) and according to Lord Google, have never been written before. Behaviorism has a great deal of difficulty explaining both how I can create a novel combination of words which still has meaning, particularly out of the blue and with no particular stimulus to prompt it, and how you can comprehend and react to this completely novel sentence.

        On the other hand, Chomsky was wrong to suggest that we are born knowing language. Funnily enough, children like Genie have somewhat disproved both theories – Chomsky because in a deprived environment, Genie was unable to achieve normal speech, and Skinner because no amount of careful teaching and training was able to give Genie normal grammar.

        I think that when considering Skinner we must remember that Psychology has progressed mightily since his day. More modern theories such as Social Interactionism and Relational Frame Theory are probably closer to the truth, incorporating aspects of both Skinner and Chomsky. Those will be tossed aside some day as well, no doubt, for even better theories…

  9. kenanddot

    I’m afraid I don’t find Chomsky’s arguments at all convincing. There is a very helpful critical response to Chomsky’s case for linguistic nativism by Geoffrey Sampson (The Language Instinct Debate, Continuum Press, 2005, first published as ‘Educating Eve’, 1997). It covers a lot of ground and one of the important things it does is investigate Chomskian claims about how people speak, but quantitatively, through engagement with linguistic corpora rather than remaining at the level of linguistic intuition.

    I believe he addresses the point about parental feedback as well, but doesn’t the objection rely on a limited picture of what would be available as feedback to the child learning language? I think it’s quite likely that feedback from parents and others could be a lot more subtle than ‘No we say it this way’. It could be as subtle as just pausing slightly enough to register the mistake in the subtle movements of their faces.
    The child is learning lots of things at the same. The general rules, like -ed for the past tense, could be learned as clitics, and the exceptions learned as echoics. They might put the two together and get nonwords like ‘ranned’ but that goes extinct when they notice no one else says it. Isn’t the mixed up speech exactly what would be expected if the child’s responses, or different aspects of a response, are in the process of coming under the control of multiple stimuli at the same time?

    As for the creativity of language, haven’t you already addressed that in essence with that piece on creative behaviour. Behaviour can be genuinely creative and novel too. There’s no fixed limits on human behaviour. The meaninglessness or otherwise of the resultant sentence is a function of the context of utterance. It will be meaningless if it doesn’t cohere properly with the context, but it will make sense in context. People have written stories in which Chomsky’s sentence makes sense.

    I think the reason it becomes harder to learn languages as you get older is that you have too many other things competing with your attention and that sometimes the languages can be antagonistic. A lot depends on your motivation (to drop the behaviourist language for a moment). Adults, say immigrants or refugees, have trouble learning the language of their new countries, just like they have trouble fitting in generally, because they already have ways of doing things that work very well for them and they can’t bring themselves to wholly give them up in the way they would need to to pass completely in the new country. I’m not saying for a moment that they should give up the old ways! But hanging on to them stops them properly assimilating. So, I think there’s a sufficient external reason to explain why second language acquisition is difficult. We don’t need a developmental cut off switch. I would be happy to say the brain is less pastic and adaptable at old age too. I would just take that as the physiological consequences of habitual behaviour (just like the behaviour of working out at the gym has physiological consequences). I don’t think neurological changes per se jeopardise an external ‘nurture’-type explanation.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s