Monday, March 26, 2007

Why Be Anti-Behaviorist

Behaviorism is unpopular. It is dismissed by cognitive scientists developing intricate internal information processing models. It is neglected by cognitive ethologists and ecological psychologists convinced that its methods are irrelevant to studying how animals and persons behave in their natural and social environment. It is rejected by neuroscientists sure that direct study of the brain is the only way to understand the causes of behavior.

Remnants of behaviorism survive in both behavior therapy and laboratory-based animal learning theory. In the metaphysics of mind, too, behavioristic themes survive in the approach to mind known as functionalism. Functionalism defines states of mind as states that play particular causal-functional roles in animals or systems in which they occur. Paul Churchland writes of functionalism as follows: “The essential or defining feature of any type of mental states is the set of causal relations it bears to … bodily behavior” (1984, p. 36). This functionalist notion is similiar to the behaviorist idea that reference to behavior and to stimulus/response relations enters centrally and essentially into any account of what it means for a creature to behave or to be subject, in the scheme of analytical behaviorism, to the attribution of mental states.

Remnants, however, are remnants. Behaviorism has lost strength and influence. Why?

The deepest and most complex reason for behaviorism's demise is its commitment to the thesis that behavior can be explained without reference to non-behavioral mental (cognitive, representational, or interpretative) activity. Behavior can be explained just by reference to its “functional” (Skinner's term) relation to or co-variation with the environment and to the animal's history of environmental interaction. Neurophysiological and neurobiological conditions, for Skinner, sustain or implement these functional relations. They do not serve as ultimate or independent sources of behavior. Behavior, Skinner (1953) wrote, cannot be accounted for “while staying wholly inside [an animal]; eventually we must turn to forces operating upon the organism from without.” “Unless there is a weak spot in our causal chain so that the second [neurological] link is not lawfully determined by the first [environmental stimuli], or the third [behavior] by the second, the first and third links must be lawfully related.” (p. 35) “Valid information about the second link may throw light on this relationship but can in no way alter it.” (ibid.) It is “external variables of which behavior is a function.” (ibid.)

Skinner was no triumphalist about neuroscience. Neuroscience, for him, more or less just identifies organismic physical processes that underlie animal/environment interactions. Therein, it rides epistemic piggyback on radical behaviorism's prior description of those interactions. “The organism”, he says, “is not empty, and it cannot adequately be treated simply as a black box” (1976, p. 233). “Something is done today which affects the behavior of the organism tomorrow” (p. 233). Neuroscience describes inside-the-box mechanisms that permit today's reinforcing stimulus to affect tomorrow's behavior. The neural box is not empty, but it is unable, except in cases of malfunction or breakdown, to disengage the animal from past patterns of behavior that have been reinforced. It cannot exercise independent or non-environmentally countervailing authority over behavior.

For many critics of behaviorism it seems obvious that, at a minimum, the occurrence and character of behavior (especially human behavior) does not depend primarily upon an individual's reinforcement history, although that is a factor, but on the fact that the environment or learning history is represented and how (the manner in which) it is represented. The fact that the environment is represented by me, to me, constrains or informs the functional relations that hold between my behavior and the environment and may, from an anti-behaviorist perspective, partially disengage my behavior from its reinforcement history. No matter, for example, how tirelessly and repeatedly I have been reinforced for pointing to or eating ice cream, such a history is impotent if I just don't see a potential stimulus as ice cream or represent it to myself as ice cream or if I desire to hide the fact that something is ice cream from others. My conditioning history, narrowly understood as unrepresented by me, is behaviorally less important than the environment or my learning history as represented to me.

Similarly, for many critics of behaviorism, if representationality comes between environment and behavior, this implies that Skinner is too restrictive or limited in his attitude towards the role of neurophysiological mechanisms in producing or controlling behavior. The brain is no mere passive memory bank of behavior/environment interactions (see Roediger and Goff 1998). The central nervous system, which otherwise sustains my reinforcement history, contains systems or sub-systems that implement or encode whatever representational content the environment has for me. It is also an active interpretation machine or semantic engine, often critically performing environmentally untethered and behavior controlling tasks. Such talk of representation or interpretation, however, is a perspective from which behaviorism—most certainly in Skinner—wished to depart.

One defining feature of traditional behaviorism is that it tried to free psychology from having to theorize about how animals and persons represent their environment. This was important, historically, because it seemed that behavior/environment connections are a lot clearer and more manageable experimentally than internal representations. Unfortunately, for behaviorism, it's hard to imagine a more restrictive rule for psychology than one which prohibits hypotheses about representational storage and processing. Stich, for example, complains against Skinner that “we now have an enormous collection of experimental data which, it would seem, simply cannot be made sense of unless we postulate something like” information processing mechanisms in the heads of organisms (1998, p. 649).

A second reason for rejecting behaviorism is that some features of mentality—some elements in the inner processing of persons—have characteristic ‘qualia’ or presentationally immediate or phenomenal qualities. To be in pain, for example, is not merely to produce appropriate pain behavior under the right environmental circumstances, it is to experience a ‘like-thisness’ to the pain (as something dull or sharp, perhaps). Behaviorist creatures may engage in pain behavior, including beneath the skin pain responses, yet completely lack whatever is qualitatively distinctive of and proper to pain (its painfulness). (See also Graham 1998, pp. 47-51 and Graham and Horgan 2000. On the scope of the phenomenal in human mentality, see Graham, Horgan, and Tienson forthcoming).

The philosopher-psychologist U. T. Place, although otherwise sympathetic to application of behaviorist ideas to matters of mind, argued that qualia cannot be analyzed in behaviorist terms. He claimed that qualia are neither behavior nor dispositions to behave. “They make themselves felt,” he said, “from the very moment that the experience of whose qualia they are” comes into existence (2000, p. 191; reprinted in Graham and Valentine 2004). They are instantaneous features of processes or events rather than dispositions manifested over time. Qualitative mental events (such as sensations, perceptual experiences, and so on), for Place, undergird dispositions to behave rather than count as dispositions. Indeed, it is tempting to postulate that the qualitative aspects of mentality affect non-qualitative elements of internal processing, and that they, for example, contribute to arousal, attention, and receptivity to associative conditioning.

The third reason for rejecting behaviorism is connected with Noam Chomsky. Chomsky has been one of behaviorism's most successful and damaging critics. In a review of Skinner's book on verbal behavior (see above), Chomsky (1959) charged that behaviorist models of language learning cannot explain various facts about language acquisition, such as the rapid acquisition of language by young children, which is sometimes referred to as the phenomenon of “lexical explosion.” A child's linguistic abilities appear to be radically underdetermined by the evidence of verbal behavior offered to the child in the short period in which he or she expresses those abilities. By the age of four or five (normal) children have an almost limitless capacity to understand and produce sentences which they have never heard before. Chomsky also argued that it seems just not to be true that language learning depends on the application of reinforcement. A child does not, as an English speaker in the presence of a house, utter “house” repeatedly in the presence of reinforcing elders. Language as such seems to be learned without, in a sense, being taught, and behaviorism doesn't offer an account of how this could be so. Chomsky's own speculations about the psychological realities underlying language development included the hypothesis that the rules or principles underlying linguisitic behavior are abstract (applying to all human languages) and innate (part of our native psychological endowment as human beings). When put to the test of uttering a grammatical sentence, a person, for Chomsky, has a virtually infinite number of possible responses available, and the only way in which to understand this virtually infinite generative capacity is to suppose that a person possesses a powerful and abstract innate grammar (underlying whatever competence he or she may have in one or more particular natural languages).

The problem to which Chomsky refers, which is the problem of behavioral competence and thus performance outstripping individual learning histories, seems to go beyond merely the issue of linguistic behavior in young children. It appears to be a fundamental fact about human beings that our behavior and behavioral capacities often surpass the limitations of our individual reinforcement histories. Our history of reinforcement often is too impoverished to determine uniquely what we do or how we do it. Much learning, therefore, seems to require pre-existing or innate representational structures or principled constraints within which learning occurs. (See also Brewer 1974, but compare with Bates et al. 1998 and Cowie 1998).


Benbob said...

As someone who is working on a PhD in applied behavior analysis, it is clear that you, like Chomsky, do not understand the science, literature, theory and philosophy you are criticizing. You are giving credence to the hypothetical constructs of psychology, which have no scientific evidence to support them whatsoever; you then use these baseless theories to criticize the hard science of behavior. Consider this excerpt from Palmer (2009) as you fall exactly into to this category:

"The patent failure of Skinner’s approach is often cited, usually to help justify an alternative approach. Even authors who find themselves drawn to quite compatible positions take pains to argue that there are fundamental differences(e.g., McClelland & Rumelhart, 1986; Snow, 1996), apparently lest the reader dismiss their claims out of hand. Of course much of this animus can be dismissed as merely recycled dogma: There is seldom evidence that the dismissal of Skinner’s work is based on familiarity with it. A few prominent critics (e.g., 33 Chomsky, 1959, 1971; Dennett, 1978) tend to be cited as authoritative to the neglect of primary sources."

In contrast, the seductive metaphors of cognitive science raise as many questions as they answer. As Machado, Lourenco, and Silva (2000) have asked of one such metaphor:
If we say that a rat navigates a maze efficiently because it scans with the mind’s eye a stored representation of the maze, a cognitive map, and then admit that there is literally no mind’s eye, literally no internal action of scanning, literally no map, at least in the sense that we usually conceive of eyes, scanning actions, and maps, then how does our account explain the rat’s behavior? (p. 30)
Not only should we like our explanations to rest on independent principles, we should like them to be smoothly integrated with the rest of biology. In particular, we should like them to be compatible with what is known about physiology and evolutionary biology, for otherwise, an account carries the extra burden of accommodating any discrepancies. Here too a behavioral account satisfies, for plausible neural mechanisms of behavioral processes have been identified, and the adaptive significance of such processes is conspicuous (Donahoe & Palmer, 1994).

It is still surprising to me, however, that the dominate paradigm of psychology is to dismiss the only portion of the field that has scientific evidence to support it.

Benbob said...

On another note, you should look up the term 'mentalism'; examine why behaviorism rejects it; then reexamine all the mentalistic theories you are using to criticize behaviorsm.

You wrote:

"Unfortunately, for behaviorism, it's hard to imagine a more restrictive rule for psychology than one which prohibits hypotheses about representational storage and processing."

Behaviorism is a hard science. These restrictions are what make it 'good' science. This one of greatest assets that behaviorism holds over the rest of the field of psychology.

Imagine if we applied similar hypothetical constructs and mentalistic theories, free of such restrictions, to other hard sciences like physics. It would be awfully hard to fly a plane. I would guess that you wouldn't want to ride in it. Yet, you clearly prefer to ride in this one. Good luck with that.