BEHAVIORISM

Traditional notions of the mind have tended to treat mental states as "private" and "subjective," not accessible to the public and objective methods of science. With the failure of an "introspectionist" psychology in the early twentieth century, the only recourse seemed to be either to deny that mental states had any role to play in any serious science, or to try to find a way to understand talk of mental states that was entirely objective. The first option is called the "eliminativist" strategy, and Radical behaviorism was a monumental effort to realize it. The eliminativist strategy proposed to explain all human and animal behavior in terms of physically specified stimuli, responses, and reinforcements. It is to be distinguished from the second, "reductionist" strategy, which attempts not to eliminate mental phenomena, but rather to save mental phenomena by identifying them with some or other existing physical phenomena. Analytical behaviorism was the specific reductionist view that mental phenomena could be identified in one way or another with dispositions to overt behavior. Both Radical behaviorism and Analytical behaviorism dominated Anglo-American philosophy, and especially psychology, from roughly 1920 through 1970.

Although the two views are similarly motivated, they are independent. As will be seen in section one, Radical behaviorism is a specific scientific hypothesis, to be assessed according to the usual scientific criteria of how well it predicts and explains its intended range of phenomena. Analytical behaviorism is essentially a semantic, or philosophical hypothesis, to be assessed according to how well it captures the mental notions it purports to analyze (sec. 2). A person could subscribe to one and reject the other: Strict radical behaviorists might be skeptical of semantic proposals of analytical behaviorists; and many analytical behaviorists might reject the scientific proposals of Radical behaviorism.

There is also a third view, methodological behaviorism, according to which the only evidence for any mental phenomena must be behavioral. As a claim about evidence, this is actually independent of both the other views, although it often accompanied them. Indeed, one of the lasting positive contributions of the entire behaviorist movement was a much higher standard of evidence than had been observed previously, discouraging the kind of reliance on empathic intuitions that was characteristic, for example, of clinical psychotherapeutic claims. Unlike Radical behaviorism and Analytical behaviorism, methodological behaviorism survives in some quarters to this day, although some problems for methodological behaviorism are raised at the conclusion of section three.

Radical Behaviorism

the law of effect

Since this is a philosophical encyclopedia, the treatment of Radical behaviorism will perforce be brief (for a more thorough discussion in which references can be found to the experiments cited here see Gleitman et al. 2004, Gallistel 1990, and Rey 1997). However, the treatment of Radical behaviorism is not philosophically irrelevant since a substantial number of twentieth-century philosophical views often relied on it, most famously those of the American philosopher W. V. Quine.

Radical behaviorism emerged from the work of the Edward Thorndike (1874–1949), John Watson (1878–1958), and Ivan Pavlov (1849–1936), receiving its most energetic development in the work of B. F. Skinner (1904–1990) and attaining considerable precision in the work of Clark Hull (1884–1952). It has its source in traditional empiricist theories of the mind, according to which the mind at birth is a tabula rasa, or blank tablet on which experience forms sensory impressions. Ideas are derived from experience and are welded together to form complex ideas by a process of association, which closely tracks the presentation of those experiences in reality. Thus, certain sights, sounds, and tactile sensations become associated in experience to form the idea of a material object, and certain associations of "contiguity, succession and constant conjunction" form the idea of causation (Hume 1734).

This traditional suggestion, though regarded by radical behaviorists as right in spirit, suffered from a major defect—namely, a reliance upon peculiar private entities, ideas, and impressions, which did not seem to radical behaviorists to be proper objects of scientific inquiry. To remedy this situation, they proposed studying not associations among ideas but among physically characterizable stimuli to sense organs and responses of the motor system. The specific law that linked stimuli and responses was Thorndike's Law of Effect, which for purposes of this entry may be stated thus:

The Law of Effect: The probability of a response R following a stimulus S is increased/decreased if pairs 〈R, S〉 have been followed by positive/negative reinforcements, F, in certain patterns (e.g., intermittently) in the past.

For example, should a particular movement like pressing a paw on a lever (=R) when a light is on (=S) be followed intermittently by the presentation to a hungry animal of a food pellet (=F), then the probability of the animal pressing its paw on the lever when the light is on in the future will be increased. Such are rewards. Negative reinforcements are either the absence of positive reinforcements, or actual punishments, which also reinforce, but in the opposite direction: the probability of the R given S is reduced if pairs of S and R have been followed by punishment in the past. Radical behaviorism is essentially the bold hypothesis that all intelligent human and animal behavior can be explained by the Law of Effect.

As Skinner frequently stressed, the Law of Effect is nearly the biological principle of natural selection, extended now beyond the persistence of traits that are genetically inherited to the persistence of acquired behaviors in individual animals. Just as from a random generation of genetic mutations certain ones are selected by virtue of meeting an environmental test of "survival of the fittest," so from an essentially random generation of responses in an animal certain ones are selected by virtue of being reinforced when they occur after certain stimuli. The responses that are selected in this way Skinner (1938) called "operants," since they involved ways that an animal "operated" on an environment that secured reinforcement. This process of "operant conditioning" was Skinner's distinctive contribution over "classical conditioning," where the response was elicited (e.g., salivation by hunger in Pavlov's dogs), rather than being spontaneously emitted.

How could such a simple law as the Law of Effect possibly stand a chance of explaining the full range of animal behavior? The central idea was an extension of the associationist strategy of building complex ideas from simpler ones, only now it was a matter of building not complex ideas, but complex responses. These could be built up out of simpler responses by "response chaining," whereby stimuli associated with a reward themselves become ("secondary") reinforcers, and so available for the conditioning of further responses. Thus, a pigeon conditioned to peck a lever on hearing a bell could now be conditioned by the sound of the bell itself to produce further responses given further stimuli, say, doing a little dance on seeing a red light, which is then followed by the bell, which is then followed by food if the pigeon pecks again at the lever. Discrimination of complex stimuli would similarly be built from discrimination of simpler stimuli, through either a chain of discriminations of simpler stimuli, or by "stimulus generalization," whereby novel stimuli are treated as "of the same kind" as earlier ones.

The Law of Effect is likely true of some animal behavior. Skinner achieved remarkable successes using it to train animals to engage in all manner of curious behavior: for example, rats to run mazes, pigeons to play Ping-Pong, and pigs to push shopping carts around supermarkets. And the Law of Effect seems to play a role in explaining a variety of persistent behavioral patterns, such as gambling and drug addiction, as well as in extinguishing them, as in "behavior modification therapy." For the purposes of this entry, the issue is not whether such applications occur or are a good idea, but whether they offer a theoretically adequate paradigm for understanding the full range of intelligent animal behavior.

inadequacies of the law of effect

Problems with the Law of Effect emerge in the first instance from the radical behaviorist experiments themselves. Contrary to popular belief, it is not only human behavior that resists radical behavioristic explanation; the theory does not even really work for the rats. The main problem is that the probability of a response can be increased in ways other than by the Law of Effect. There are at least four classes of phenomena that the law has trouble explaining: latent learning, passive learning, spontaneous alteration, and improvisation.

Latent learning occurs when an animal learns without reinforcement. Rats that were well sated with, for example, food and water were allowed to run around in a maze for ten days without any reward, sometimes being placed in the maze at arbitrarily different points. Subsequently, when they were hungry again they were introduced into the maze and were able to find the food much faster than rats not previously exposed. Similarly, Harry Harlow showed that monkeys presented with a complex hinge, requiring the undoing of several pins and bolts to free it, learned to undo it with no special reward other than "the fun of it." Further, indigo buntings learn something about the position of the stars while still in the nest, despite not using this information for navigation (and so, a fortiori, not for any reward) until they are much older. In all of these cases the probability of the animal producing the appropriate response was greater than that of animals that had not been previously exposed to such stimuli, but without any history of reinforcement. In a related vein, passive learning occurs when an animal learns without antecedently producing the requisite response. Thus, rats can learn a maze merely by being pulled through it in a transparent trolley car, not executing anything like the responses that will take them through the maze when they are tested later.

Not only can rats learn without reinforcement or response, they can sometimes respond in ways that defy their conditioning history. In "spontaneous alteration," an animal actually avoids emitting the response that has recently been reinforced. After having found food at a particular location, for example, hummingbirds will go somewhere else to find more food. Rats presented with a number of paths of equal length to a goal will vary their routes, although invariably in ways that advance their approach to the goal. The phenomenon is most dramatically displayed by rats in a "radial maze," consisting of eight pathways radiating out in all directions from a central location, with baits placed at the end of each arm. The Law of Effect should predict that the rats should return to an arm in which they have found food. What they do instead, however, is to avoid an arm they have already visited until they had—at random—visited all the others. That is (as we might put it mentalistically), they seem not to be matching responses to stimuli, but "keeping track" of "where the baits are," and, knowing they had consumed one, no longer "expected" it to be there. The Law of Effect seems not only inadequate to account for such cases; it actually seems to be disconfirmed by them

Animals also produce appropriate behaviors that have not even previously been produced, much less reinforced. Thus, rats trained to take a circuitous route to a goal box will immediately take a shortcut if it is made available. Indeed, animals apparently refuse to be tied to specific physical responses: Rats will swim a flooded maze after being taught to run it, and—moving to the wild, outside the confines of a structured maze—desert ants will forage in a winding path up to one hundred meters from their nests, and then, once they find food, will take a beeline home.

Passing beyond issues of navigation, it has been noted with regard to latent learning that monkeys presented with a novel, complex hinge, figure out how to undo several pins and bolts to free it. Köhler demonstrated even more remarkable improvisation in chimpanzees: They would use sticks as rakes to secure food that was outside a fence; they would then use these sticks as poles, which they would climb up in order to snatch food that was out of reach, grasping the food just as the stick toppled over. In all of these cases, the responses—that is, the sequence of muscular motions required to execute the acts—are by no means physically type-identical between prior and test trials. So the animals must have learned something other than merely to repeat certain physically typed responses.

Of course, these inadequacies with the Law of Effect become even more glaring in the human case. Picking up on an example of Skinner's (1957, p. 38), Daniel Dennett (1975) provides an apt and amusing discussion of the difficulties besetting a radical behaviorist attempting to explain why someone mugged in New York hands over his wallet: Why doesn't the person instead do any number of things that were more likely to have been previously reinforced with the stimulus "Your money or your life," such as giggling, or yawning? Of course, it is not impossible that there is a story of prior threat stimuli and responses of the requisite sort. But the burden is squarely on the radical behaviorist to supply it.

Radical behaviorists, of course, did not take challenges to the inadequacies of the Law of Effect lying down. They often made ingenious replies to them involving elaborate emendations of the theory—for example, by Clark Hull (1943). But these emendations were then subject to further tests, showing animals to be more ingenious than the Law of Effect allowed. A consensus began to emerge that what animals learn is not any mere sequence of responses to stimuli, but rather to the development of what Edward Tolman called an "internal map" (1948). Such talk of "insight" and "maps" of course, begins to imperil any eliminativist ambitions of Radical behaviorism: such a map would be an inner representation, involving an internal mental state.

structured responses and language

An important problem in principle for Radical behaviorism was raised by Karl Lashley (1951): Serial responses like those involved in tying shoes or riding a bicycle seemed to be structured in a way that it did not seem possible to explain by local response-chaining alone. A domain of behavior that exhibits particularly striking structure is language. Skinner (1957) tried to sketch an account of linguistic behavior, but it was soon subject to a devastating review by the then young linguist Noam Chomsky (1964). Among other things, he pointed out:

1. along lines indicated by Lashley, language is structured in units that cannot be captured by response chaining. For example, a sentence of the form "Either … or …," or "If … then … " can involve waiting indefinitely for novel items to be inserted in the blanks;
2. language is creative : most of the sentences people encounter and produce are constantly novel —it is why people bother to converse, read, and write—all contrary to the Law of Effect's commitment to a prior history with the stimuli and responses;
3. language is productive : in grasping a grammar, even small children know how to produce a potential infinity of novel, structured sentences, as in "This is the house that Jack built," "This is the rat that lived in the house …," "This is the cat that chased the rat …," without any history of conditioning each component in this way; indeed:
4. the complex set of rules that constitute the grammar is acquired effortlessly by practically all human children by the age of three, without (and sometimes despite) any efforts at explicit instruction.

(For more detailed discussion see Chomsky 1972 and Pinker 1994).

Although Chomsky's review was (to many minds) a definitive blow by itself, what really led to the end of Radical behaviorism was the spectacular positive research program that he and others (e.g., Fodor 1968, 1975) had begun to develop, what has come to be called the cognitive revolution, associated with computational-representational theories of mental processes.

Often recognizing the difficulty of avoiding mentalisms in the explanation of animal behavior, radical behaviorists sometimes allowed mentalisms to creep into their explanations, postulating "exploratory" and "curiosity" drives, or "drives to perceive" or "know." Of course, if the theory was to remain true to its goal to avoid reference to subjective mental states that it regarded as unscientific, it would be obliged to define these postulations in terms of overt behavior. It was in this way that Radical behaviorism invited Analytical behaviorism, to which we now turn.

Analytical Behaviorism

Analytical behaviorism was motivated by two related philosophical trends of the twentieth century that persist into the twenty-first century: the well-known verifiability theory of meaning (or verificationism ) and the less well-known doctrine that might be called irreferentialism. Because the latter serves as something of a background for the former, it will be considered first.

irreferentialism

Irreferentialism is a novel suggestion that arose from Bertrand Russell's (1905) famous theory of definite descriptions, according to which expressions like "the present king of France" should not be construed as referring to any (in this case) nonexistent entity, but as rather shorthand for some logically complex expression, only some of the most basic parts of which manage to refer. Perhaps the most obvious deployment of such a strategy is the in the case of a sentence such as "The average American family has 2.5 children," which, of course, does not entail that there is some family somewhere in America that has a half of a child. A proper analysis of the grammar of the claim reveals that it is simply a way of expressing the ratio between American families and their children.

The view begins to be applied as a claim about mental expressions in the work of the later Wittgenstein (1953) and Gilbert Ryle (1949). They argued that philosophers too often think about the phenomena that people introspect in their "inner mental worlds" on the model of the objects in the familiar "outer" one. The temptation to this analogy arose, Wittgenstein and Ryle claimed, from an excessively referential conception of the functioning of the human mental vocabulary, treating words like "belief," "thought," and "sensation" as referring to "inner," "private" objects, in the way that words like "cat" and "rock" refer to outer, public ones. It is not that, like "Zeus," they do not happen to refer to anything; rather, like "the average American," they do not even purport to refer to anything. The view is perhaps best known from Ryle's attack on the idea of the "ghost in the machine": A mind is not some sort of thing that could be a ghost, or any other thing. Not surprisingly, this irreferentialism was often associated with an antipathy one finds in Wittgenstein (1953) and Ryle (1949) toward a psychology that suggests a "promise of hidden discoveries yet to be made of the hidden causes of our actions and reactions" (Ryle 1949, p. 325).

Irreferentialism is an essentially negative thesis about the analysis of mental terms. Understandably, many people might want to hear something more positive: if the analysis of mental terms does not involve the postulation of mental entities, what does it involve? For Wittgenstein and his followers, in particular, this question (like, in their view, most philosophical questions) was the wrong one to ask: it exhibited a somehow inappropriate "craving for generality" about the nature of thought and language. That may in the end be so, but many wanted to see a greater effort made toward some systematic account than he and his followers provided. The use of mental language does not seem entirely capricious and chaotic, and, if it is not, then it is not unreasonable at least to ask what the principles might be that guide its use.

verificationism

According to verificationism, the meaning of a claim consists in the method by which it could be tested (see Ayer 1952 for a classic statement). For example, claims about something's being an acid might be defined in terms of its turning litmus paper red. Or claims about the existence of material objects might be analyzed as logical construction of claims about those sense experiences that people ordinarily take to confirm such claims (e.g., that people would have certain experiences of color, shape, and resistance to touch). Hypotheses such as those of the possibility of human lives being a dream, or other people having radically different mental lives were to be ruled out as "meaningless" if there really were in fact no evidence in principle that could make a difference to their truth or falsity.

There are myriad problems with verificationism: It is by no means obvious how to apply it to the claims of logic, mathematics, ethics, or aesthetics, or even to itself (what is the test for assessing where it is true?). Even in the supposed parade cases of natural science verificationism did not fare well: Scientists often do not know how to seriously test a hypothesis (as in contemporary string theory in physics) and often change their tests as their theories evolve (as new tests are devised for a disease). But the most serious problem is confirmation holism, or the fairly obvious observation that claims are not tested by experiment individually, but only as parts of whole theories (see Quine 1960). As will be seen, Analytical behaviorism offered a vivid case in point.

analytical behaviorist proposals

Analytical behaviorism was largely motivated by verificationism and the observation that the vast majority of human mentalistic claims are tested by observing overt behavior (of course, this does not seem to be true in the case of first-person reports, which were always a problem for Analytical behaviorism, although these represented a small minority of claims). Moreover, it did seem that those ascriptions were by and large indifferent discoveries that might be made about the actual physical aetiology of mental life. If one were to open up the heads of familiar people and discover that they were empty or full of sawdust, one would not conclude that these familiar people did not have the mental states that seem to be constituted by the familiar behavior observed. Consequently, it seemed reasonable to suppose that mental claims should be understood as equivalent to various sorts of dispositional or conditional claims about how an agent would behave if she were in such and such circumstances. One particular model that impressed behaviorists was that of dispositional claims that arise elsewhere: "Salt is soluble" presumably means something like, "If salt is put into water in certain normal conditions, then it dissolves"; "Glass is fragile," something like, "If struck in normal circumstances, it breaks." Analogously, wanting something should be taken to mean something like "trying to get it, if the occasion were to arise."

Successfully spelling out the appropriate dispositions in the case of mental terms turned out, however, to be none too easy. Ryle was never precise, but he offered a strategy, exemplified by the following characterization of belief:

To believe that the ice is dangerously thin is to be unhesitant in telling oneself and others that it is thin, in acquiescing in other people's assertions to that effect, in objecting to statements to the contrary, in drawing consequences from the original proposition, and so forth. (Ryle 1949, 134–135)

action vs. "colorless movement"

A formidable problem arises, however, as soon as one considers the "behavior" on which people normally rely, namely of distinguishing actions from mere movement : To take a famous example from Wittgenstein, it is the difference between raising one's arm and one's arm rising. The rising of an arm might occur as a result of, say, some machine moving the arm up and down; it is only the raising of an arm if it was the result of the person whose arm it is intending to raise it. Ryle may think that he is describing mere behavior in talking about someone being "unhesitant" and "acquiescing," "objecting" to certain "assertions," but a moment's reflection reveals these as only slightly covert mentalisms: hesitation, acquiescence, and the possible involvement of any of an indefinite variety of bodily movements (or none); all that is crucial is that whatever the agent does or does not do is a result of a certain psychological attitude.

This point was often most seriously missed by the radical behaviorists who, as has been noted already, often resorted to mentalisms to deal with the apparent counterexamples to their Law of Effect. Thus, Skinner wrote:

The artist … is reinforced by the effects his works have upon … others … [his] verbal behavior … reach[ing] over centuries or to thousands of listeners or readers at the same time. The writer may not be reinforced often or immediately, but the reinforcement may be great. (Skinner 1957, pp. 206, 224)

But, as Chomsky (1964) noted, the term reinforcement here has degenerated to only a ritual function, being used as a cover term for "X wants Y," "X likes Y," "and X wishes that Y were the case."

insuperable problems

For every thought experiment arguing for Analytical behaviorism there are compelling ones against it as well. Consider not only people with sawdust heads, but people who turn out to be robots cleverly controlled by radio waves produced by some ingenious scientists at MIT: Such creatures would seem to have no more of a mental life than do marionettes. Or consider a race of "Super-Spartans" who, as matter of training and principle, refuse to flinch or complain in response to even the most excruciating pain and are inarticulate about an enormous range of their psychological repertoire (Putnam 1975). Surely it is possible that these Super-Spartans have by and large the full range of psychological states of the more expressive and articulate.

All of these objections become more evident when one considers an underlying technical difficulty noticed by Roderick Chisholm (1957): Every effort to define most mental states by behavior seems to require citation of other mental states. Typically, any particular mental state causes a particular behavior only in conjunction with (an often large number of) other mental states. Beliefs, hopes, and expectations issue in behavior only in conjunction with (at least) desires; desires issue in behavior only in conjunction (at least) with beliefs and expectations. To take a proposed example from Tolman, suppose a person tried to define a rat's expectation that there is food at L in terms of the rat's moving toward L: This only if the rat wants food; and the rat's wanting food can be defined in terms of its moving toward L only if it expects there is food at L. Insofar as this is true, the prospects of a definition of a single informational or single directional state in terms of behavior seem dim. This problem is an instance of the aforementioned confirmation holism emphasized by Quine. Indeed, a philosophically influential (and disconcerting) way of understanding this and related difficulties with Analytical behaviorism is provided by Quine's (1960, ch. 2) "thesis of the indeterminacy of translation," according to which there is no fact of the matter about the content of mental states, a thesis that has influenced later philosophers such as Donald Davidson, David Lewis, and Dennett.

Methodological Behaviorism

In the twenty-first century few, if any, philosophers or psychologists would be prepared to defend either Radical behaviorism or Analytical behaviorism. However, there is a weaker view that survives, called "methodological behaviorism," according to which the only permissible evidence for a psychological claim can be behavioral. It is not, like Radical behaviorism, an explanatory scientific hypothesis, but neither does it, like Analytical behaviorism, offer analyses of mental terms, although it is motivated by vaguely verificationist concerns like those that motivated Analytical behaviorism. Methodological behaviorism is perhaps best expressed by Wittgenstein's famous dictum, "An inner process stands in need of outward criteria," (1953, sec. 580) but without any of the analytical behaviorist commitment to defining an inner process in terms of some specific criteria. It has most recently been defended by Dennett (1993), who describes its motivation as not a "village" but an "urbane verificationism" that is merely trying to avoid "epiphenomenalism, zombies, conscious teddy bears, [and] self-conscious spiders" (1993, p. 461). Indeed he "unhesitatingly endorse[s] the claim that necessarily, if two organisms are behaviorally exactly alike, they are psychologically exactly alike" (1993, p. 922).

For all methodological behaviorism's urbanity, however, it is hard to find a convincing argument for it. Why shouldn't psychologists avail themselves of evidence that may go beyond ordinary overt behavior, as they indeed seem increasingly to do when they investigate the finer structure of the brain? Consider, for example, the nice case Dennett (1991, pp. 395–396) discusses of the familiar plight of the adult beer-drinker who wonders whether in coming to like it since childhood, it is his experiences or his preferences that have changed. It can seem obscure what further considerations should settle the matter, and it is not implausible to suppose that current behavioral discriminations or even introspection would not suffice. Dennett concludes that that there is in such a case "no fact of the matter" about "the way the beer tastes" to such a person.

But suppose it turns out that children have more taste buds than adults. One might have independent evidence that both children and adults have the same preferences for bitter titillation, but that consequently children reach a painful threshold sooner with the same quantity of a bitter substance. It tastes differently because, arguably, more intense sensation is caused by their tongues and/or gustation subsystems. However, such cases would clearly transcend mere behavioral evidence.

So, in the end, even methodological behaviorism seems problematic given the increasingly rich conceptual and evidential resources of cognitive science, and especially of a computational/representational theory of thought. Behaviorism in all its forms seems a heroic theory that was ultimately defeated by the high standards of theory and experimentation that it encouraged. Eliminativist strategies survive in ambitions to replace mental talk by neurophysiological descriptions, and reductionist strategies abound, along either neurophysiological lines or computational ones. But the effort to replace mental talk with behavioral talk, or reduce it to it, can safely be said to have passed with the twentieth century, in which it first appeared.

See also Chomsky, Noam; Fodor, Jerry A.; Functionalism; Mind-Body Problem; Physicalism; Quine, Willard Van Orman; Reductionism in the Philosophy of Mind; Verifiability Principle.

Bibliography

Ayer, A. Language, Truth and Logic. New York: Dover, 1952. A classic statement of the "Verificationist Theory of Meaning," which provided the general semantic view underlying Analytical behaviorism.

Chisholm, R. Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press, 1957.

Chomsky, N. Language and Mind. New York: Harcourt Brace Jovanovich, 1972. Chomsky's earliest and most accessible statement of the relation of his theory of grammar to theories of the mind generally.

Chomsky, N. "Review of Skinner's Verbal Behavior." In The Structure of Language: Readings in the Philosophy of Language, edited by J. Fodor and J. Katz, 547–578. Englewood Cliffs, NJ: Prentice Hall, 1964. The devastating review of Skinner's effort to incorporate linguistic behavior into Radical behaviorism.

Dennett, D. Brainstorms. Cambridge, MA: MIT Press, 1975. Several essays ("Skinner Skinner," "Why the Law of Effect Won't Go Away") provide excellent discussions of the attractions and limitations of Radical behaviorism.

Dennett, D. Consciousness Explained. New York: Little Brown, 1993. A recent defense of methodological behaviorism (which he calls "urbane verificationism").

Fodor, J. The Language of Thought. New York: Crowell, 1975. The classic statement of the computational/representational theory of thought that, he argues, is presupposed by virtually all work in cognitive and perceptual psychology.

Fodor, J. The Modularity of Mind: An Essay on Faculty Psychology. Cambridge, MA: MIT Press, 1983. The classic statement of the now highly influential view of perception and language processing as consisting of computational "modules."

Fodor, J. Psychological Explanation: An Introduction to the Philosophy of Psychology. New York: Random House, 1968. An early, powerful philosophical critique of both Radical behaviorism and Analytical behaviorism.

Gallistel, C. The Organization of Learning. Cambridge, MA: MIT Press, 1990. An important non-behaviorist, computational theory of animal navigation, bringing together extensive studies of animals in the wild.

Gleitman, H., A. Fridlund, and D. Riesberg. Psychology. 6th ed. New York: Norton, 2004. An excellent, up-to-date text in psychology. Chapter four provides a detailed account of Radical behaviorism.

Hull, C. Principles of Behavior: An Introduction to Behavior Theory. New York: Appleton-Century-Crofts, 1943. A serious, concerted effort to state the laws of behavior in terms of "colorless movement" in ways designed to accommodate many of the experimental problems Radical behaviorism faced.

Hume, D. A Treatise of Human Nature, (1739), edited by L. A. Selby-Bigge. 2nd edition revised by P. H. Nidditch. Oxford: Clarendon Press, 1975. The classic statement of an associationist psychology, of which Radical behaviorism was an instance.

Lashley, K. S. "The Problem of Serial Order in Behavior." In Cerebral Mechanisms in Behavior, edited by L. A. Jeffress, 112–136. New York: Wiley, 1951. Widely regarded as raising a crucial problem for Radical behaviorism by calling attention to highly structured behaviors that could not be the result of mere response chaining.

Pinker, S. The Language Instinct: How the Mind Creates Language. New York: Harper and Row, 1994.

Putnam, H. "Brains And Behavior." In Mind, Language And Reality: Philosophical Papers. Vol. 2. Cambridge, U.K., and New York: Cambridge University Press, 1975, 325–341.

Quine, W. V. Word and Object. Cambridge, MA: MIT Press, 1960. Chapter two presents his influential discussion of the "indeterminacy of translation," which he draws as an interesting consequence of Radical behaviorism.

Rey, G. Contemporary Philosophy of Mind: A Contentiously Classical Approach. Oxford: Blackwell, 1997. Chapters four and five present detailed discussions of Radical behaviorism and Analytical behaviorism along lines of the present article. Chapters seven and ten present arguments against methodological behaviorism (called "superficialism" there).

Russell, B. "On Denoting" Mind 14 (1905): 479–493.

Ryle, G. The Concept Of Mind. London: Hutcheson, 1949.

Skinner, B. F. The Behavior of Organisms: An Experimental Analysis. New York, Appleton-Crofts, 1938. The official statement of Skinner's famous version of Radical behaviorism.

Skinner, B. F. Verbal Behavior. New York: Appleton-Century-Crofts, 1957. Skinner's effort to provide a radical behaviorist framework for understanding language.

Tolman, E. "Cognitive Maps in Rats and Men." Psychological Review 55 (1948): 189–208.

Georges Rey (2005)