Orbach's examination of the work of Lashley and Hebb is of great historical interest, but it illustrates a vast gap, both past and present, between research on the nervous system and research on behavior. Grand strides in the neurosciences have taken place with insufficient attention to the behavior of the organisms that are the hosts of nervous systems. In the final analysis, nervous systems are selected by evolutionary contingencies on the basis of the behavior that they engender. If we fail to understand the behavior, we will probably also fail to understand how the brain serves it. As we move away from the Decade of the Brain into the Decade of Behavior, those unfamiliar with the properties of behavior will be at a disadvantage when they seek its sources in the brain, because they will not know what they should be looking for. Lashley was on the right track when he used the properties of serial order in behavior to make inferences about the nervous system, but too often both Lashley and Hebb speculated about the nervous system without firm grounding in what was even then known about learning and behavior. We now know much more, and neuroscience and the science of behavior have each reached a point at which a modern synthesis holds great promise.
2. Some of the experiments in my undergraduate course in Experimental Psychology were on the psychophysics of vision. We obtained dark adaption thresholds and measured acuity at different locations in the visual field. When our instructor, W. N. Schoenfeld, related the discontinuities in our dark adaptation curves and our different foveal and peripheral acuities to rods and cones in the retina, he stressed that the visual behavior came first, in the sense that the anatomical differences between rods and cones became significant only in the context of what was known about the different kinds of seeing.
3. As a graduate student, I ran experiments in B. F. Skinner's pigeon laboratory. One of them was concerned with interocular transfer in the pigeon. Skinner had urged me to make that work my doctoral dissertation, because he thought it might help to correct the misapprehension that he and his students had no interest in physiological issues. His critique of physiologizing in his book, The Behavior of Organisms (Skinner 1938), had been about premature physiological hypothesizing, such as Pavlov's irradiations and Hull's Conceptual Nervous System, and not about actual studies of physiological systems. Like Schoenfeld, he was consistent in arguing that behavior has priority, in the sense that it is the guide to what neuroscientists must look for in the brain. There were marvelous precedents for this view: for example, Sherrington (1906) needed his behavioral data to arrive at his theory of the synapse.
4. Skinner made his point explicit in replies to commentaries in the Behavioral and Brain Sciences (BBS) treatments of some of his classic papers (Catania and Harnad 1984). For example: "It is the function of a science of behavior at the present time to give neurologists their assignments, as it was the function of genetics prior to the discovery of DNA to give modern geneticists their assignment with respect to the gene" (Skinner, 1988, p. 60, and cf. pp. 239 and 334; these pages correspond to BBS pages 507, 607 and 660). Skinner elaborated on the point this way: "The use of operant techniques in the brain science laboratory is the best demonstration I can offer of the contribution of an independent science of behavior in making the task of brain science clear. Valid facts about behavior are not invalidated by discoveries concerning the nervous system, nor are facts about the nervous system invalidated by facts about behavior. Both sets of facts are part of the same enterprise, and I have always looked forward to the time when neurology would fill in the temporal and spatial gaps which are inevitable in a behavioral analysis" (Skinner, 1988, p. 128, and cf. pp. 461 and 470; these pages correspond to BBS pages 543, 707 and 722).
5. I published my research on interocular transfer (Catania 1965), though I chose a different topic for my doctoral dissertation. The transfer research required the presentation of lateral stimuli as well as the usual anterior stimuli that appeared on the key on which the pigeon pecked. I learned that the pigeon's visual field consisted of an anterior myopic field and lateral hyperopic fields or, in other words, that the pigeon was near-sighted in front of its beak and far-sighted on either side (Catania 1964). Years after, I was pleased to learn of anatomical work that confirmed a differentiated region of the pigeon's retina corresponding to the myopic anterior field. The behavior had come first.
6. As I read Orbach's book, these examples repeatedly came to mind. Here were accounts by (and about) two brilliant and influential researchers who failed to see the many ways in which the analysis of behavior that was evolving at the same time as their own work could have guided them to more coherent treatments of the issues with which they were concerned. Instead of moving from behavior to the nervous system, too often they moved in the other direction.
7. I found it ironic to read the following: "The new formulation suggested that it is not the stimulus that produces a response. More accurately, the stimulus excites the organism, and it is the organism that initiates the response" (Orbach 1998, p. 5). Orbach describes this as a movement away from stimulus-response or S-R theories, but that movement had already been initiated in the analysis of behavior when Skinner began speaking of emitted behavior and of behavior that was occasioned by discriminative stimuli rather than being elicited by them. Nevertheless, in Lashley's and Hebb's writings as well as in Orbach (1998) and some commentaries on it (e.g., Robinson, 1999, paragraph 5), behavioral accounts are usually identified with S-R bonds and/or associations and are either explicitly or by implication characterized as obsolete. It is correct that associationism remains the centerpiece of some contemporary learning theories, but then as now there were other alternatives. By the 1950's Skinner's behavior analysis had already become a selectionist rather than an associationist account of behavior. Within the lifetime of the individual organism, behavior is selected by its consequences, much as organisms are selected over generations by evolutionary contingencies. This ontogenic selection, or the shaping of behavior by its consequences, creates functional units of behavior.
8. Here is another example: "Imagine the revelation that Hebb must have experienced, a man who had been committed to a Pavlovian connectionist point of view, and had begun to struggle himself with Lashley's neuropsychological problems of stimulus and response equivalence and the problems of selective attention and thought. The idea that activity in the cerebral cortex could be sustained independently of sensory and motor processes must have intrigued Lashley" (Orbach, 1998, p. 49). Compare Skinner's reflections on the influence of Sherrington and Pavlov as he groped toward the concept of the operant. "The concept of the reflex had served them well, and in my thesis I had said that it was all that was needed in the study of behavior. I knew better by the time I began to write my book [Skinner 1938]. My field was the operant rather than the respondent.... I could not break my chains, however. I went on talking about reflexes" (Skinner 1979, p. 201). But Skinner had begun to replace the language of reflexes with the language of operants, the emission of behavior and their selection by their consequences well before Hebb's "Organization of Behavior" appeared (Hebb, 1949), and soon after he made the ties between behavior and biology more explicit in his treatment of evolutionary contingencies (e.g., Skinner 1953, p. 90). Later work elaborated on the interactions of phylogenetic and ontogenic variables in the determination of behavior (e.g., Skinner 1966, 1981), and parallels between phylogenic natural selection and the ontogenic selection of behavior by its consequences have since been explored in considerable detail (e.g., Catania 1978, 1987; Smith 1986). The emission of behavior provides the variations upon which operant selection operates. When Hebb wrote "All mammalian behavior involves learning; but the innate factor is always there, determining what learning occurs and how" (quoted by Orbach, 1998, p. 38), Skinner's grounds for disagreement, if any, would have been only with respect to how learning should be characterized and assessed, not with whether there exist phylogenic ("innate") contributions to behavior.
9. I was most struck by the lost opportunities as Orbach's text and the reprinted articles reminded me about the features of Lashley's treatment of sensory and motor equivalences. Here were discussions of stimulus and response classes that could not be identified with any common neural pathways. A triangle at different locations, orientations and distances in the visual field stimulates different populations of visual receptors, and the legible writing of a word does not demand any particular muscle group (as when the word is written with either right hand or left). Lashley and Hebb were concerned with the organization of these stimulus and response entities at the level of the nervous system, but they either did not recognize or did not acknowledge that much (though probably not all) of the organization with which they were concerned derives from the interaction of the organism's behavior with its environment. Operant classes, which can be defined in behavioral terms without appeal to physiological processes, have precisely the properties that could have been coordinated with the classes that were characterized by Lashley and by Hebb (1949) in terms of sensory and motor equivalences.
10. The operant, as a class of behavior selected by its consequences, is a fundamental unit of behavior (Skinner 1935; Catania 1996). If a pigeon's pecks produce food, for example, pecking may become established as an operant. As an operant, it must be distinguished from classes of pecking that have other sources (e.g., elicited pecking). The class is defined in terms of both response properties (e.g., force of the key peck) and the stimuli in the presence of which responses occur (e.g., key pecking in the presence of green may be established as a different discriminated operant from key pecking in the presence of red). The stimulus term of the discriminated operant sometimes remains implicit (e.g., key pecking depends on various stimulus properties of the key even when the stimuli displayed on it remain constant), but all operants participate in three-term contingencies in which discriminative stimuli set the occasion on which responses have consequences (e.g., at a traffic intersection, the consequences that may follow from stepping on the gas or stepping on the brakes vary with whether the traffic light is green or red). To conflate these relations with those of the stimulus and response terms in the reflex is to misconstrue the nature of operant behavior (cf. Shimp, 1989).
11. Operant classes are defined functionally rather than topographically. For example, a rat's press of a lever with left paw, right paw, both paws, chin or rump is a member of the class of lever pressing provided only that the same contingencies operate for all of these topographies. The common contingencies define the class. We tend not to think of all of these variations as arbitrary, but it is appropriate to do so because they all depend on the arbitrary environment that exists within the experimental chamber.
12. Now consider another example, in which the variants that enter into the common contingencies are not different response topographies, as in the lever-pressing example, but instead are arbitrary stimulus sets. In Vaughan (1988), photographic slides were divided into two arbitrary sets of twenty each. The slides were presented one at a time, and pecks were reinforced given slides from one set but not the other. Occasionally the correlation between slide sets and reinforcement was reversed. After several reversals, pigeons began to switch responding from one slide set to the other after only a few slides. The common contingencies arranged for the twenty slides in a set made them functionally equivalent, in that changed contingencies for just a few slides in the set changed behavior appropriately for all of them. This procedure created two arbitrary classes of discriminated operants, pecks to one slide set and pecks to the other, by arranging common contingencies for the members within each set. The slides within each set had no common physical properties, and so, as for the triangle in the visual field, there can be no appeal to any common sensory pathway. But the origins of these operant classes are known: they are the product of common contingencies of ontogenic selection (selection at a cellular level via cell death, as in Edelman 1987, is a different variety of ontogenic selection). If such classes should ever be identified with neural structures, whether they be cell assemblies or reverberating circuits or whatever, the next task would be to show how their neurophysiological properties are engendered by environmental contingencies. (It may be worth noting in passing that the properties of operant classes seem to have more in common with the presumed properties of Hebb's cell assemblies than those of Lashley's reduplicated traces and interference patterns.)
13. Operant classes can also be embedded in higher-order classes, or classes on which higher order contingencies operate (as when consequences are arranged for imitation in general rather thatnfor specific imitative instances: cf. Catania 1995a). A discussion of higher-order classes is beyond the scope of this review, but it may be useful to note that higher-order classes provide opportunities for the emergence of novel behavior (some have called verbal instances of such emergence productivity).
14. Another example: In imprinting, a duckling follows a stimulus, whether mother duck or some arbitrary moving object, if the stimulus is introduced under appropriate conditions early in the duckling's life (for a more nuanced treatment, see Hoffman 1996). It is sometimes said that the duckling's following is elicited by the imprinted stimulus, but the language of elicitation is misleading. A natural consequence of the duckling's walking in different directions is to change its distance from its mother. If closeness to the mother is a reinforcer, it is no surprise that the duckling walks toward rather than away from her. It follows that if the closeness requires some response other than walking, the walking should be replaced by that other response. In one such experiment (Peterson, 1960), ducklings could produce a moving imprinted stimulus by pecking at a disk on the wall or by standing still on a platform. The ducklings emitted responses that produced the stimulus as a consequence even when those responses were incompatible with following. The critical property of the imprinted stimulus was not that it could elicit responses such as following or pecking or standing still, but rather that proximity to the imprinted stimulus could reinforce those responses. In other words, imprinting does not make the imprinted stimulus an elicitor of following; instead, it makes the stimulus a reinforcer.
15. One point made by this example is that the criteria for deciding whether a response has been elicited or shaped by its consequences are behavioral, not physiological. Obviously imprinting must have some physiological basis, but the researcher who searches in the brain for some neurological instantiation of elicitation will probably be at a disadvantage relative to one who searches for a neurological instantiation of either the operant selection of response classes by their consequences or the acquisition of reinforcing properties by the imprinted stimulus. Once again, the behavior comes first by showing the neuroscientist what the neurophysiology must do.
16. Another point of this example is that it illustrates behavioral explanation. Orbach (1998, p. 112; see also p. 277) quotes Lashley: "It is a sad commentary on his [Pavlov's] influence that his most ardent followers both in Russia and America are psychologists who hold that an adequate explanatory system for behavior can be worked out without reference to the nervous system" (Orbach parenthetically inserts: "evidently Skinner and Hull in America are being singled out"). But demonstrating that the phenomenon of imprinting can be characterized as an instance of operant selection (or shaping) rather than of elicitation is a variety of explanation, and it works without appeal to how the duckling's brain accomplishes imprinting. It works because the analysis of behavior calls upon a taxonomy of behavioral processes (including emission, elicitation and operant selection among other members) and the analytic task is one of identifying which processes operate in specific instances (Catania 1993). This does not rule out an eventual explanation based on brain processes (cf. the Skinner quotations in paragraph 4), but that reductive explanation is explanation of a different kind. It is perfectly legitimate not to be satisfied with any particular level of explanation, but explanation at one level need not negate or invalidate explanation at another (e.g., a satisfactory explanation of how a prism converts white sunlight into a spectrum can be offered without appealing to a quantum account of light). In general, we judge explanations to be effective or valid when they relate what is to be explained to other familiar and well-established phenomena. At any level, explanation is showing how something works. One kind of account of imprinting explains how things work at the level of behavior, which involves the ways in which the organism interacts with its environment; another could explain how things work at the level of physiological processes. Neither is complete. For example, neither would give an account of the phylogenic contingencies that over evolutionary time presumably selected those ducklings capable of imprinting over those not.
17. Some accounts of behavioral phenomena appeal to structure. For instance, one approach to the analysis of imprinting is to study which features of imprinting stimuli (e.g., movement, color, size) make imprinting more or less likely; similarly, an analysis of the topography of the duckling's following would be concerned with motor structure rather than with its function. Stimulus structure is often characterized in terms of common features. But in the absence of common features, derivatives of stimulus properties (e.g., information) or sets of features that are weighted probabilistically (e.g., fuzzy sets or polymorphic classes or prototypes) may be invoked. If no common physical features can be identified, however, all approaches that look to stimulus properties to define how such classes are formed must fail. We must look instead to the behavioral processes that created these classes, and the only consistent features of their members are the common contingencies they enter into. There just isn't anything else. Thus, when Lashley and Wade stated that "The dimension itself is created by or is a function of the organism and only secondarily, if at all, a property of the physically definable character of the stimuli" (quoted by Orbach 1998, p. 100), they properly identified the environmental antecedents of behavior but omitted the shaping role of contingencies (i.e., the consequences of behavior).
18. In biology, the study of structure is called anatomy and the study of function is called physiology. The priority of one or the other was an issue in the history of biology (e.g., Bonner 1961; Russell 1916; cf. Catania 1973). Behavior also has both structure and function, and questions of which has priority are raised at many points in Lashley's and Hebb's work (e.g. Orbach 1998, pp. 179, 282, 340). In any case, when a horse runs flexions of individual muscles combine to produce the movement of each leg, which in turn are coordinated with each other and with other parts of the horse's body. The details of these complex coordinations change as the horse shifts from one gait to another. All gaits, either natural (trotting) or trained (the rack), are constrained by neurophysiological and mechanical factors. Such constraints constitute a grammar of the horse's running, but this grammar does not contain the functions of the horse's running: when it runs; with which gait; where it goes; how its running interacts with its other behavior; what consequences follow from running. Just as organs differ in anatomy and physiology, so also varieties of behavior differ both in what they look like and in what they do. One horse may overtake another at a lope or at a gallop, and it may gallop either in overtaking another horse or in escaping from a predator. In the former case, actions of different form have similar functions; in the latter, actions of similar form have different functions. (Language too has both structure and function, but some linguistic accounts have so emphasized grammatical structure that they have almost completely neglected the functions of language; for example, it is a challenge to locate any material relevant to language function in Pinker .)
19. The problem of serial order is concerned with the structure of behavior in time. I still assign Lashley's (1951) paper on serial order in my graduate course on learning as one of his most original and enduring contributions. My familiarity with this paper was probably an important factor in bringing me to write this review. Perhaps the longevity of the paper in my teaching can be attributed to the role behavior plays in it. In other papers, especially his later ones, Lashley sometimes seems to give mere lip service to behavior, even when the term appears in the title. For example, Lashley's 1958 paper called "Cerebral organization and behavior" is mostly about the mind-body problem and consciousness and it is hard to find much about behavior in it (similarly, the space devoted to behavior in Hebb's 1949 book is in my view scanty relative to that devoted to other issues). But in the paper on serial order, Lashley clearly began with the properties of behavior and then considered their implications for organization in the nervous system. It set such an excellent precedent for how to proceed in studying relations between behavior and brain that it is curious that the precedent was so rarely followed in other work by Lashley or by Hebb.
20. But there was still another difficulty. Lashley appreciated the complexity of the Pavlovian conditioned reflex and sought an alternative to it: "Bechterew, Pavlov and the behaviourist school in America attempted to reduce all psychological activity to simple associations or chains of conditioned reflexes" (Lashley, as quoted in Orbach 1998, p. 204). Lashley changed that. In his ruling out of chaining accounts for some kinds of behavior sequences, however, he was interpreted as having ruled out such accounts for all. Furthermore, Lashley did not distinguish between Pavlovian chained sequences and operant chains.
21. In an operant chain, each response in the sequence produces a stimulus that sets the occasion for the next one. For example, pressing an elevator button is followed by the opening of the elevator door, which sets the occasion for entering the elevator, where there is now an opportunity for pressing a floor button, and so on until one reaches one's destination. An operant chain is a succession of different operants, each defined by the reinforcing consequence of producing a stimulus that provides an opportunity to engage in the next. The structure of the environment dictates the organization of such sequences (you cannot enter the elevator until the door has opened), and the integrity of each unit in the chain is demonstrated whenever the parts can be separated from each other. For example, an elevator door that is already open sets an occasion for entering that is independent of the button press. Some behavior sequences can be reduced to smaller units in this way. Lashley was interested in the ones that cannot.
22. Lashley's argument, then, was that some sequential patterns of responding cannot be reduced to a succession of stimulus-response or S-R units. When a skilled typist rapidly types the letters t-h-e, they cannot be discriminative stimuli for the next stroke, first because the typist typically executes that next stroke even before seeing the new letters on the page and second because these letters cannot function as discriminative stimuli if they can be followed by any of a variety of other keys depending on the word being typed (e.g., the, their, then, these, thermometer). The problem was that, faced with Lashley's argument, some concluded that they had to chose between assuming that all sequential behavior consisted of stimulus-response chains and assuming that all consisted of temporally extended units of behavior not reducible to such chains, and then opted for the latter. But the choice is necessary only in specific cases. For any given sequence the issue is deciding which type it is. Some sequences clearly can be put together in such a way that each response produces stimulus conditions that set the occasion for the next, whereas others must be integrated so that responses appear in the proper order without each depending on stimuli produced by the preceding one.
23. The consequences were unfortunate. For example, it has sometimes been assumed that behavioral accounts of language appeal exclusively to chaining. For that reason, Skinner's (1957) account of verbal behavior has sometimes been seen as incompatible with Lashley's incisive analysis (cf. Bruce 1994; Catania 1995b; Chomsky 1959; MacCorquodale 1970). But Skinner was well aware of Lashley's arguments, and though he dealt with some aspects of verbal behavior in terms of response chains, he did not argue that all sequential behavior is produced by such chaining. The most relevant evidence is that his taxonomy of verbal behavior included verbal responses attributable to chaining as only one of several distinct functional classes of verbal behavior. He called such responses intraverbals, distinguishing them from verbal behavior that came about through other processes (examples of intraverbal sequences include the completion of a line of poetry given the first few words or rote-learned arithmetic facts such as "four" as a response to "two plus two"). Consistent with Lashley's views, Skinner also argued that sequences learned intraverbally could become units in their own right or, in other words, that they could become well enough established that they no longer depended on chaining. The misapprehensions presumably persisted at least in part because Skinner's selectionist approach was not clearly distinguished from the associationist approaches that characterized other then extant behavioral positions (cf. Hailman 1988; Hunt 1988).
24. To a rough first approximation and at much risk of oversimplification, researchers in the contemporary psychology of learning can be divided into two groups. Those in one group, sometimes identifying themselves with animal learning, trace their intellectual antecedents to associative learning theories and the work of Hull and Spence, among others; some of them, concerned mainly with structural issues, also or instead identify themselves with the area of animal cognition and the work of Tolman. Those in the second group, identifying themselves with the experimental analysis of behavior, attribute their origins to operant behavior and selectionism in the work of Skinner. The research of former tends to use nonhuman rather than human preparations in group studies, whereas that of the latter tends to examine behavior in both human and nonhuman preparations in studies with individual organisms; the two groups tend to publish in different journals.
25. Just as the psychologies of nonhuman and human learning went their separate ways over much of the twentieth century (Catania, 1985), selectionist and associationist accounts of learning have developed different methodological and interpretive styles and have only interacted in limited ways. It is worth noting that associationism is not equivalent to connectionism, because associationism assumes associations among behavioral units such as stimuli or responses whereas the units involved in connectionism operate at a very different level (cf. Donahoe and Palmer 1989). Furthermore, selectionism can accommodate associationism because associations, like response classes or discriminated operants, can be selected, whereas an accommodation does not easily work in the other direction. After all, even associationist accounts must have a mechanism by which some associations are selected over others, but ontogenic selection has the advantage that it need not be restricted merely to associations. Though behavior analysis has sometimes been declared moribund or already dead, as befits operant units it remains lively. (Those who would like to learn more about behavior analysis will recognize my bias if I cite my own work [Catania 1998] but I can also strongly recommend the insightful and thoroughly selectionist text by Donahoe and Palmer 1994. Reviews that have dealt extensively with the relations between behavior analysis and research in biology and other disciplines have been anthologized in Catania and Hineline 1996.)
26. As already noted, the three-term contingency, in which a discriminative stimulus sets the occasion upon which a response has consequences, evolved over some time as a relation distinct from the respondent signalling by one stimulus of another. The joint dependence of behavior on both stimulus antecedents and on consequences distinguishes operant from respondent relations. The three-term contingency brought signalling and consequential functions together in such a way that effects on behavior were not reducible to pairwise relations among the three terms. Even researchers in the respondent tradition have finally recognized that operant relations are better characterized by the three-term contingency than by separate pairwise relations among the possible combinations of discriminative stimulus and response, discriminative stimulus and reinforcer, and response and reinforcer. Their justification is that the three-term contingency is not derivable from combinations of these two-term relations. (Unfortunately, they chose to describe their findings in terms that obscure the continuity with behavior analytic precedent, so that, for example, discriminative stimuli have been renamed as occasion-setters: e.g., Holland 1983; perhaps for that reason, they have also occasionally failed to recognize anticipations of their work in the earlier literature: e.g., Morse and Skinner 1958; Rescorla 1994.)
27. In any case, the criteria for distinguishing between these associative and selective processes are behavioral, and Lashley left other clues that for him behavioral data were primary besides his paper on serial order. For example, his reference to reflexes in paramecia (Orbach 1998, p. 184) makes it evident that his criteria for a reflex relation included an observed correlation between stimuli and responses and did not require a nervous system. But neuroscience has come a long way since the days of Lashley and Hebb. It has moved from gross anatomy to the biochemical details of changes at intercellular and intracellular levels. It would be surprising if their conceptions of the workings of the nervous system had not become obsolete. What is probably more relevant now are their stances with regard to the relations between behavior and brain. In my view, Lashley was more likely to start from behavior (as he did early in his own career in his research with the jumping stand and other learning preparations) and then to move to the brain. Hebb was more likely to move in the other direction (though many features of the brain that he started from were hypothetical). This view of Hebb is supported, for example, by his mistaken conclusions about the lability of early learning (e.g., Orbach 1998, p. 87), presumably influenced by phenomena in mature organisms such as learning set.
28. There is still another sense in which behavior comes first. Nervous systems service the behavior of the organisms that are their hosts. They (and other organ systems) have been selected by evolutionary contingencies on the basis of the behavior that they engendered. It may be useful to consider parallels between the ways in which organisms interact with their environments (as when the organization of the environment drives the shaping of discriminated operants) and the ways in which nervous systems interact with the bodies in which they find themselves (as when the organization of peripheral receptors drives the innervation of sensory systems: cf. K. C. Catania and Kaas 1995, 1997; Provine 1988).
29. The phenomena studied in the analysis of behavior have varied time courses. The reinforcement of a single instance of a response has immediate effects (as does its punishment). A novel response can be shaped in just a few minutes, but once established the new response will still be available if the organism is returned to the apparatus only many months later. In a standard classroom demonstration, I start with a pigeon that has been trained to peck a red but not green key. By shaping pecks on green with food reinforcers and then reinforcing movements away from the red key with the onset of green, I then reverse the discrimination, so that the pigeon now pecks green but not red, usually within only six to eight minutes. Such changes in behavior, at least in mammalian and avian species, can be produced at virtually any time in the organism's life (barring pathologies) and are readily reversible (via extinction or the reinforcement of other behavior). Other changes are more constrained in when they can occur, in how long-lasting they are, and in their reversibility, though the nature of the constraints in many cases remains an open question (e.g., see Hoffman 1996 on imprinting, or Jenkins et al. 1990 on sensory reorganization). Furthermore, procedures exist for standardizing measures of rates of acquisition and changes in behavior during shaping (e.g, Boren and Devine 1968; Eckerman et al. 1980; Galbicka, Kautz and Jagers 1993; Platt 1973). In addition, various temporal parameters of learning procedures have substantial effects on behavior. Delay of reinforcement, for example, which relates rates of operant behavior to the time between the reinforced response and its reinforcer, is a potent variable, and anomalies in the form of the delay of reinforcement gradient have been implicated in attention deficit/hyperactivity disorder in both humans and in a spontaneously hyperactive rat model (Sagvolden 1996; Sagvolden and Sargeant 1998).
30. Research in the neurosciences now offers brain changes at varying cellular sites and with varying temporal properties both in their initiation and their persistence (e.g., Kandel and Schwartz 1982; Yang et al. 1999; Segal, Korkotian and Murphy 2000). But, if as a behavior analyst I may be forgiven for saying so, extremely sophisticated physiological and genetic procedures have sometimes been accompanied by relatively crude behavioral ones. In Yang et al. (1999), for example, it is difficult to assess rates of acquisition across such different kinds of procedures as object recognition, fear conditioning and water mazes. Differences among outcomes with these procedures may stem not just from differences in capacity to learn but also from complex interactions among parametric variables and such mundane behavioral variables as levels of activity.
31. Evolutionary environments have selected learners, and that selection has made learning different in different situations. For example, sources of food and water vary but prey organisms do not get repeated opportunities to learn to escape from predators, so the acquisition of behavior that is shaped by consumables as reinforcers is likely to be much more labile than that of behavior that involves escape or other interactions with aversive stimuli (cf. Bolles 1970). Furthermore, advantages in the rapidity of acquiring new behavior must be traded off against the effects of coincidental correlations between behavior and environmental events (see Skinner 1953 on superstition; cf. Dawkins 1988). Contemporary organisms are highly evolved (perhaps rodents more so than primates, if the proper scale is number of generations rather than calendar time), so it may be worth considering the possibility that genetic variants similar to those produced by the procedures of Yang et al. (1999) are not more optimal for learning but instead have already lost out in selection to those that exist in natural populations. None of this is intended to demean the very substantial contribution made by Yang and colleagues, but rather to suggest that much is to be gained from a more detailed analysis of the behavioral changes that their elegant genetic procedures have engineered. Such analyses would enable us to say more precisely just what those genetically induced neural changes do.
32. Now that neuroscientists can look at actual events in the nervous system, it is worthwhile to be reminded that behavior still has priority, in the sense that it is the guide to what neuroscientists must look for in the brain. The missed opportunities that I cited above have not been irreversibly lost. In this endeavor, behavior analysis and brain science are allies and not competitors and will together fill in and perhaps write over some of the neurophysiological aspects of the organism's interaction with its environment that were sketched out in necessarily broad strokes by Lashley and by Hebb. It is appropriate to close with another quotation from Skinner (1988, p. 470, corresponding to BBS p. 722): "A behavioral analysis has two necessary but unfortunate gaps - the spatial gap between behavior and the variables of which it is a function and the temporal gap between the actions performed by an organism and the often deferred changes in its behavior. These gaps can be filled only by neuroscience, and the sooner they are filled, the better."
I am grateful to Terje Sagvolden for comments on a preliminary draft of the manuscript and to Stevan Harnad for recognizing long before I did just how strong my interest was in writing this review.
Azar, B. (1999) "Decade of Behavior" moves forward. APA Monitor (Online) 30(3) March 1999 http://www.apa.org/monitor/mar99/decade.html
Bolles, R. C. (1970) Species-specific defense reactions and avoidance learning. Psychological Review 77: 32-48.
Bonner, J. T. (Ed.) (1961) On Growth and Form by D. W. Thompson: An Abridged Edition. Cambridge: Cambridge University Press.
Boren, J. J., and Devine, D. D. (1968) The repeated acquisition of behavioral chains. Journal of the Experimental Analysis of Behavior 11: 651-660.
Bruce, D. (1994) Lashley and the problem of serial order. American Psychologist 49: 93- 103. Catania, A. C. (1964) On the visual acuity of the pigeon. Journal of the Experimental Analysis of Behavior 7: 361-366.
Catania, A. C. (1965) Interocular transfer of discriminations in the pigeon. Journal of the Experimental Analysis of Behavior 8: 147-155.
Catania, A. C. (1973) The psychologies of structure, function, and development. American Psychologist 28: 434-443.
Catania, A. C. (1978) The psychology of learning: Some lessons from the Darwinian revolution. Annals of the New York Academy of Sciences 309: 18-28.
Catania, A. C. (1985) The two psychologies of learning: Blind alleys and nonsense syllables. In S. Koch and D. E. Leary (Eds.), A century of psychology as science (pp. 322-335). New York: McGraw-Hill.
Catania, A. C. (1987) Some Darwinian lessons for behavior analysis. A review of Peter J. Bowler's The Eclipse of Darwinism. Journal of the Experimental Analysis of Behavior 47: 249-257.
Catania, A. C. (1993) The unconventional philosophy of science of behavior analysis. Journal of the Experimental Analysis of Behavior 60: 449-452.
Catania, A. C. (1995a) Higher-order behavior classes: Contingencies, beliefs, and verbal behavior. Journal of Behavior Therapy and Experimental Psychiatry 26: 191-200.
Catania, A. C. (1995b) Lashley on serial order and Skinner on verbal behavior. American Psychologist 50: 178.
Catania, A. C. (1996) On the origins of behavior structure. In T. R. Zentall and P. Smeets (Eds.) Stimulus Class Formation in Humans and Animals (pp. 3-12). Amsterdam: Elsevier.
Catania, A. C. (1998) Learning (4th ed.). Upper Saddle River, NJ: Prentice-Hall.
Catania, A. C. and Harnad, S. (1984) (Eds.) The canonical papers of B. F. Skinner. Behavioral and Brain Sciences 7: 473-724.
Catania, A. C., and Hineline, P. N. (Eds.) (1996) Variations and Selections: An Anthology of Reviews from the Journal of the Experimental Analysis of Behavior. Bloomington, IN: Society for the Experimental Analysis of Behavior. Available from http://www.envmed.rochester.edu/wwwvgl/jeab/vascontents.htm
Catania, K. C. and Kaas J. H. (1995) The organization of the somatosensory cortex of the star-nosed mole. Journal of Comparative Neurology 351: 549-567
Catania, K. C. and Kaas, J. H. (1997) The mole nose instructs the brain. Somatosensory and Motor Research 14: 56-58
Chomsky, N. (1959) Review of B. F. Skinner's Verbal Behavior. Language 35: 26-58.
Dawkins, R. (1988) Replicators, consequences, and displacement activities. In Catania, A. C. and Harnad, S. (Eds.) The Selection of Behavior: The Operant Behaviorism of B. F. Skinner (pp. 33-35). New York: Cambridge University Press.
Donahoe, J. W. and Palmer, D. C. (1989) The interpretation of complex human behavior: Some reactions to Parallel Distributed Processing edited by McClelland, Rumelhart and the PDP Research Group. Journal of the Experimental Analysis of Behavior 51: 399-416.
Donahoe, J. W., and Palmer, D. C. (1994) Learning and Complex Behavior. Boston: Allyn and Bacon.
Eckerman, D. A., Hienz, R. D., Stern, S., and Kowlowitz, V. (1980) Shaping the location of a pigeon's peck: Effect of rate and size of shaping steps. Journal of the Experimental Analysis of Behavior 33: 299-310.
Edelman, G. M. (1987) Neural Darwinism. New York: Basic Books.
Galbicka, G., Kautz, M. A., and Jagers, T. (1993) Response acquisition under targeted percentile schedules: A continuing quandary for molar models of operant behavior. Journal of the Experimental Analysis of Behavior 60: 171-184.
Hailman, J. P. (1988) Ethology ignored Skinner to its detriment. In Catania, A. C. and Harnad, S. (Eds.) The Selection of Behavior: The Operant Behaviorism of B. F. Skinner (pp. 431-432). New York: Cambridge University Press.
Hebb, D.O. (1949) The Organization of Behavior. New York: Wiley.
Hoffman, H. S. (1996) Amorous Turkeys and Addicted Ducklings: A Search for the Causes of Social Attachment. Boston: Authors Cooperative.
Holland, P. C. (1983) "Occasion-setting" in Pavlovian feature positive discriminations. In M. L. Commons, R. J. Herrnstein, and A. Wagner (Eds.), Quantitative Analyses of Behavior: Volume IV. Discrimination Processes (pp. 183-206). Cambridge, MA: Ballinger.
Hunt, E. (1988) A case study of how a paper containing good ideas, presented by a distinguished scientist to an appropriate audience, had almost no influence at all. In Catania, A. C. and Harnad, S. (Eds.) The Selection of Behavior: The Operant Behaviorism of B. F. Skinner (pp. 250-252). New York: Cambridge University Press.
Jenkins, W. M., Merzenich, M. M., Ochs, M. T., Allard, E., and Guic-Robles, T. (1990) Functional reorganization of primary somatosensory cortex in adult owl monkeys after behaviorally controlled tactile stimulation. Journal of Neurophysiology 63: 82-104.
Kandel, E. R., and Schwartz, J. H. (1982) Molecular biology of learning: modulation of transmitter release. Science 218: 433-443.
Lashley, K. S. (1951) The problem of serial order in behavior. In L. A. Jeffress (Ed.), Cerebral Mechanisms in Behavior: The Hixon Symposium (pp. 112-146). New York: Wiley.
MacCorquodale, K. (1970) On Chomsky's review of Skinner's Verbal Behavior. Journal of the Experimental Analysis of Behavior: 13, 83-99.
Morse, W. H., and Skinner, B. F. (1958) Some factors involved in the stimulus control of operant behavior. Journal of the Experimental Analysis of Behavior 1: 103-107.
Orbach, J. (1998) The Neuropsychological Theories of Lashley and Hebb. Lanham, MD: University Press of America.
Orbach, J. (1999) Precis of: The Neuropsychological Theories of Lashley and Hebb. PSYCOLOQUY 10(029). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.029.lashley-hebb.1.orbach http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?10.029
Peterson, N. (1960) Control of behavior by presentation of an imprinted stimulus. Science 132: 1395-1396.
Pinker, S. (1994). The Language Instinct. New York: Morrow.
Platt, J. R. (1973) Percentile reinforcement: Paradigms for experimental analysis of response shaping. In G. H. Bower (Ed.), The Psychology of Learning and Motivation. Volume 7 (pp. 271-296). New York: Academic Press.
Provine, R. R. (1988) A hierarchy of developmental contingencies: A review of Purves and Lichtman"s Principles of Neural Development. Journal of the Experimental Analysis of Behavior 50: 565-569.
Rescorla, R. A. (1994) Control of instrumental performance by Pavlovian and instrumental stimuli. Journal of Experimental Psychology: Animal Behavior Processes 20: 44-50.
Robinson, D. N. (1999) The Snark Is Still a Boojum. PSYCOLOQUY 10(45). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.045.lashley-hebb.2.robinson http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?10.045
Russell, E. S. (1916) Form and Function. London: John Murray.
Sagvolden, T., (1996) The attention deficit disorder might be a reinforcement deficit disorder. In J. Georgas, M. Manthouli, E. Besevegis and A. Kokkevi (Eds.), Contemporary Psychology in Europe: Theory, Research and Application (pp. 131-143). Gttingen: Hogrefe and Huber.
Sagvolden, T., and Sergeant, J. A. (1998) Attention deficit/hyperactivity disorder - from brain dysfunctions to behaviour. Behavioural Brain Research 94: 1-10.
Segal, M., Korkotian, E., and Murphy, D. D. (2000) Dendritic spine formation and pruning: common cellular mechanisms. Trends in Neurosciences 23: 53-57.
Sherrington, C. (1906) The integrative action of the nervous system. New York: Scribner's.
Shimp, C. P. (1989) Contemporary behaviorism versus the old behavioral straw man in Gardner's The Mind's New Science: A History of the Cognitive Revolution. Journal of the Experimental Analysis of Behavior 51: 163-171.
Skinner, B. F. (1935). The generic nature of the concepts of stimulus and response. Journal of General Psychology 12: 40-65.
Skinner, B. F. (1938) The Behavior of Organisms. New York: Appleton-Century-Crofts.
Skinner, B. F. (1953) Science and Human Behavior. New York: Macmillan.
Skinner, B. F. (1957). Verbal Behavior. New York: Appleton-Century-Crofts.
Skinner, B. F. (1966) The phylogeny and ontogeny of behavior. Science 153: 1204-1213.
Skinner, B. F. (1979) The Shaping of a Behaviorist. New York: Knopf.
Skinner, B. F. (1981) Selection by consequences. Science 213: 501-504.
Skinner, B. F. (1988) Responses to commentaries. In Catania, A. C. and Harnad, S. (Eds.) The Selection of Behavior: The Operant Behaviorism of B. F. Skinner. New York: Cambridge University Press.
Smith, T. L. (1986) Biology as allegory: A review of Elliott Sober's The Nature of Selection. Journal of the Experimental Analysis of Behavior 46: 105-112.
Vaughan, W., Jr. (1988) Formation of equivalence sets in pigeons. Journal of Experimental Psychology: Animal Behavior Processes 14: 36-42.
Yang, Y.-P., Shimizu, E., Dube, G. R., Rampon, C., Kerchner, G. A., Zhuo, M., Liu, G., and Tsien, J. Z. (1999) Genetic enhancement of learning and memory in mice. Nature 401: 63-69.