Pavlov’s fundamental worldview and scientific approach never changed, but during his final years he began to rethink one core view: by New Year’s Day 1936 he had decisively rejected the identification of “conditional reflex” with “association” that had for three decades undergirded his attempt to integrate physiology and psychology. This change resulted from a rich brew of observations and thoughts in the years 1929 to 1936 related to his increasing attention to psychiatry, his sustained engagement with Gestalt psychology, and his observations of the chimps Roza and Rafael.
With his explicit “turn toward Man,” Pavlov paid increasing attention to “systematicity,” that is, to the ways in which all reflexes constantly interacted and influenced one another so that “the system is always in some way a whole.” He referred to that whole as the “dynamic stereotype,” an awkward term expressing his view that this system was changeable yet durable, exercising an inertial influence of the whole over the parts.
As always in Pavlov’s investigations, this more synthetic perspective was related to—but not determined by—experimental data. Two standard experiments highlighted the importance of systematicity. The first showed that if a series of CSs was established, any variation in the order of those stimuli changed the response to each. The dog, in other words, responded not just to a single exciter but to the system of exciters as a whole. The second experiment demonstrated that once this series was established, the first CS alone elicited the same series of salivary responses as did exposure to the entire set. Only after a few such trials did the new circumstances elicit a change in these responses, an alteration in the dynamic stereotype.
Earlier, Pavlov had worked almost exclusively “from the bottom to the top,” from the parts to the whole, from the dynamics of isolated individual reflexes toward a characterization of higher nervous activity, behavior, and the psyche as a unit. Now he sought also to move from the top to the bottom, from the whole to the parts, from the dynamic stereotype to its component individual reflexes. Earlier, he had compared unconditional reflexes (URs) to a direct line between two telephones and conditional reflexes (CRs) to a temporary connection established through a switchboard (the cerebral cortex); now his attention shifted toward the switchboard itself.1
This new focus was an attempt to incorporate within his own associationist framework elements of the holistic perspective of Gestalt psychologists. Pavlov’s “dynamic stereotype” stood in for the Gestaltists’ active subject and “perceptual field.” As he was making his turn toward psychology and psychiatry, he had been struck by the popularity of Gestalt at international congresses of psychology in 1929 and 1932. In his discussions of systematicity, he attempted to incorporate within associationism (with its devotion to the analysis of the parts) the element of truth in Gestalt (its attention, however mystified, to the dynamics of the whole).2
The new emphasis on systematicity also exemplified Pavlov’s practice of explaining troublesome experimental data by introducing new variables and enlarging the frame of interpretation. For more than a quarter of a century, he had sought a set of laws that would reconcile the discrepancies that piled up constantly as his growing army of coworkers conducted thousands of experiments on hundreds of dogs. As he put it during a Wednesday discussion of November 1933 on the nature of excitation and inhibition: “We elicit various conditions when the excitatory and inhibitory processes manifest themselves. Everything comes down to this. The time will come, and it will be such a wonderful moment, when suddenly everything becomes clear, when we will know precisely all the conditions that elicit the process and weaken it.”3
Yet this “wonderful moment” had constantly eluded him. From 1917 through 1926, Pavlov’s inability to contain his data within a few basic laws had postponed completion of his monograph, turning its composition into a frustrating exercise and manifesting itself in the unsatisfying organization, endlessly open-ended discussions, contradictions, and downbeat conclusion of Lectures on the Work of the Large Hemispheres of the Brain. From the mid-1920s through the mid-1930s, he struggled to contain the data through his concept of nervous types, constantly amending his typology and enlarging its interpretive frame through the distinction between heredity and experience, temperament and character. Yet experimental results remained contradictory and their relationship to his dogs’ behavior and personality stubbornly paradoxical. Pavlov’s typology became increasingly ad hoc, eliciting the frustrated admission—as he struggled to type his dogs at Koltushi—that “to chase reality into boundaries is difficult.”
By the late 1920s and early 1930s, even some simple and presumably well-established principles had fallen victim to a seemingly endless conditionality. Solid ground was increasingly difficult to find. The relationship between UR and CR, for example, turned out to be complex and fluid; contrary to long-standing lab doctrine, a CR sometimes elicited more salivation than the UR upon which it was based (Pavlov explained: “sometimes expectation is more intense than experience”); the amount of food used as a US had an “enormous influence” upon the size of the CR, which was also sensitive to the length of time the two were presented simultaneously. And all these interactions played out differently in different dogs. Baffling variations in the “law of strength” led Pavlov to announce in October 1935 that it required fundamental reinvestigation. The “law of the summation of CRs” was discovered to depend upon the limitations of each dog’s capacity for cortical excitation. Excitation and inhibition exhibited an ever-increasing array of phases in their intensity and interaction. The formation of new reflexes was now seen to be complicated by the preservation in the cortex of traces of previous stimuli (memories). Wednesday sessions in 1929–1936 were filled with these and other complexities that seemed always to take Pavlov farther away from that “wonderful day” when a few crisp laws of higher nervous activity would bring order to the data from decades of research.4
Most troubling, perhaps, was the failure of periodic attempts since 1905 to establish second- and third-order CRs in the lab. This cast doubt on Pavlov’s fundamental assumption that CRs—as the equivalent of psychologists’ associations—combined end-to-end in long chains to produce complex behaviors, thoughts, emotions, and knowledge about the environment. After a decade on the back burner, this problem became increasingly pressing in the 1930s as Pavlov addressed human psychology and psychiatry. He assigned two trusted coworkers, Alexander Lindberg and Nikolai Podkopaev, to revisit the issue. Each succeeded in developing an apparent second-order CR in his dog—but each time it proved short-lived, and each time it turned out that this “second-order CR” was actually formed not on the basis of a first-order CS but rather on the basis of the original US. In other words, it did not really constitute a second-order CR at all. No stable chain—not even one composed of a mere two links—could be established experimentally. In the new context of his thinking about systematicity, Gestalt, and humans, Pavlov no longer brushed off this failure. Rather, he conceded reluctantly to a Wednesday gathering of November 1932 that the feeble, inconstant, pseudo-second-order reflex occasionally obtained in experiments “is generally not analogous to our associations.”5
He thus faced a challenge to a founding axiom of his quest. From the very beginning of his investigations, Pavlov had assumed that “conditional reflex” and “association” were but different terms for the same phenomenon. Their identity underlay his entire effort to integrate physiology and psychology. He was still not prepared to surrender this general point—which he repeated in the authoritative article on CRs that he wrote in December 1934 for the Great Medical Encyclopedia—but, almost thirty years after his lab’s first baffling failure to develop second-order reflexes, he was groping toward an alternative conception.
His ruminations about this problem were reflected and facilitated by a shift in metaphor. The long-standing image of “chains of reflexes” remained, but it was increasingly subsumed within that of the cortex as a “mosaic” filled with the glittering light of excited cortical points, the darkness of inhibited ones, and the grays formed by the traces of past experiences. The mosaic metaphor, which appeared in the brief discussion of systematicity in his monograph of 1927, made it easier to envision a different kind of nervous connection: not just a link between a subcortical US and a cortical CS, but multiple connections between the cortical points of the “mosaic”—for example, between two indifferent stimuli (such as a light and a metronome that had not been joined to a UR and so had no status as signals).
Pavlov elaborated his concept of systematicity and positioned his doctrine as an extension of associationism to encompass the holism of Gestalt in four presentations of 1930–1932: a speech to Leningrad’s Institute for the Improvement of Physicians, an article for the English-language volume Psychologies of the 1930s, and addresses to the International Physiological Congress in Rome and the International Psychology Congress in Copenhagen. The new synthetic dimension of his research, he emphasized, elucidated the sometimes baffling variations of experimental data and joined physiological processes to subjective phenomena in dogs and humans, enabling CR methodology to address “the total external as well as internal activity of a higher animal.”6
He now emphasized that the cerebral hemispheres were constantly bombarded by innumerable stimuli both from the external world and from the animal’s own “internal milieu.” This latter addition reflected his interaction with the Polish Pavlovian Jerzy Konorski, who, in a 1928 article, in correspondence with Pavlov, and through participation in Pavlov’s lab from 1931 to 1933, pressed his case for a second type of CR, which he termed “motor conditioning” as opposed to Pavlov’s “classical conditioning.” (Skinner would later term this “instrumental” or “Type 2 conditioning.”) When Konorski and his collaborator moved a dog’s leg and reinforced this action with food, a CR was formed and the dog eventually moved its own leg to obtain food. Pavlov had initially resisted Konorski’s conclusion that this process differed fundamentally from his own “classical” CR, but had, by 1932, successfully repeated Konorski’s experiments and largely accepted his argument. As Pavlov put it in his speech in Rome: “There is this everyday fact, reproduced by us in the laboratory—the formation of a temporary connection between any external excitation and passive movements, with the result that the animal makes active movements in response to certain signals.” Not only did this provide the basis for investigation of volitional movements as a “conditional, associational process,” it turned Pavlov’s attention to a new area of associations governed by the cerebral cortex: those between internal bodily sensations and external signals. This insight would prove critical to his chimp studies and to his rejection of the longstanding identification of all associations with CRs.7
The dynamics of systematicity were extraordinarily complex. External and internal stimuli produced “the most varied unconditional and conditional” effects—“encountering one another, clashing, and interacting.” Through the “grandiose work” of the cerebral cortex these were constantly systematized and balanced in the dynamic stereotype—a holistic structure that complicated the dynamics of any single CR within it. Existing stereotypes were constantly challenged by new stimuli, yielding slowly and partially. “Consequently, there is a certain layering of stereotypes and a competition among them.” (This produced yet another set of variables for the interpretation of experimental data.)8
The whole, then, partially governed the parts; the existing stereotype influenced any single CR within it. Yet this stereotype was also “dynamic,” responding to changes in its parts; otherwise, CRs would be inert and unable to perform their adaptive role. After repeated trials in which the experimenter deployed only the initial CS in a series, for example, the old stereotype finally yielded to new conditions.
It did so unevenly, differently in different dogs, and, Pavlov thought, with clearly psychological effects. (A connection captured and facilitated by the very metaphor “stereotype,” which he drew from the realm of human industrial practices and mental habits to characterize a physiological phenomenon.) For example, one “strong, balanced dog” successfully incorporated a new exciter into its existing stereotype, but, clearly discomforted by this strenuous nervous work, became “extraordinarily excited, bursting from the stand, tearing off the [experimental] apparatus, barking,” refusing food, and failing to respond to previously established CSs. This disturbed state lasted for two or three months, after which the dog proved able to establish a second, simpler stereotype with much less difficulty. Strong but unbalanced dogs and “more or less weak dogs” often proved able to establish only the very simplest stereotypes.9
The disturbances among dogs struggling to change a dynamic stereotype evoked the challenges that a violation of established routines posed for human beings—most palpably, no doubt, for Pavlov himself. Were not human feelings in general often the subjective reflection of the difficult physiological process of changing a dynamic stereotype? “It seems to me,” he told the psychologists at Copenhagen, that:
Here are the feelings of difficulty or ease, energy or exhaustion, satisfaction and disappointment, triumph and despair.... Often the heavy feelings during a change in the usual form of life, with the end of habitual activities, with the loss of a loved one, to say nothing about intellectual crises and shattered beliefs, have their physiological basis to a great degree specifically in the change, in the disturbance of the old dynamic stereotype and in the difficulty of establishing a new one.10
Reflecting upon this theme at a Wednesday meeting a few years later, he empathized with one dog’s difficulties changing an established stereotype. “I am accustomed to analyzing myself. I had a son Viktor.... He was a remarkably pure soul, the embodiment of honesty and sincerity, a rare nature. Impressions of him created a very strong trace in my subjective world. And when he died I could find no peace for an entire year. I understand this physiologically in this way: this was a sharp disturbance of the stereotype, a great emptiness was created in the system and made itself felt.”11
The relationship to mental illness was obvious. Confronted with the challenge of changing a dynamic stereotype, lab dogs often developed chronic neurosis. The oft-told tale of Pavlov’s old friend Bystrov—the star student of the humanities at seminary who had failed miserably in science studies at university—could easily be recast in this new language. Bystrov’s (intellectual) stereotype did not correspond to the demands of science; the strain of attempting to change it and his failures in the classroom led inevitably to dark moods and several suicide attempts. Only after his friends marched him over to the juridical faculty—where his existing stereotype suited the course of study—did he recover and conquer his melancholy forever.12
Pavlov saw his concept of systematicity as the incorporation of the whole—which he referred to as “synthesis” or “structuredness”—into associationism, and so as a scientific response to the confusions of Gestalt. Speaking to a medical audience in Leningrad shortly after his return from the International Psychology Congress of 1929, he noted that the conflict between associationists and Gestaltists had divided psychologists into “two sharply warring camps.” The former analyzed the process by which the brain analyzed and joined individual sensations; the latter insisted that the brain always acted as a unit and that the various elements of psychic activity were understandable only as a function of the structure and dynamics of that unit.
Pavlov insisted that his own approach, while firmly rooted in associationism, respected the role of both the parts and the whole: “For us, it is entirely clear that the cortex of the large hemispheres represents a most complex functional mosaic of separate elements, each of which has a defined physiological significance, either positive or inhibitory. On the other hand, it is equally indisputable that all these elements are united at each moment into a system where each of the elements is in interaction with all the others.” Just as the chemist employed both analysis and synthesis to understand an unknown compound, so the physiologist analyzed both the formation of individual CRs and the systematicity of cortical processes as a whole. It was senseless to oppose analysis to synthesis or to insist upon the primacy of one over the other. That same theme dominated his remarks to coworkers upon his return from the International Psychology Congress in 1932, where he had delivered his address on systematicity. That congress, too, witnessed the “dominance of Gestaltism,” to which he attributed “many reports undeserving of attention.”13
Pavlov had by this time heard enough about Gestalt to disapprove of it heartily, but had not yet engaged with it critically. He had visited Wolfgang Köhler’s lab at the Psychological Institute in Berlin, probably in 1927 when traveling there from Karlovy Vary in the vain hope of a rendezvous with the exiled Vsevolod. The cofounder of Gestalt theory struck his visitor as “an entirely reasonable person, very knowledgeable, and well educated in natural science.” He had heard Köhler’s address to the Psychology Congress in New Haven in 1929, where Köhler and Pavlov were the two speakers at the main plenary session. Over the next few years, he read at least parts of Köhler’s Mentality of Apes (1917) and Gestalt Psychology (1929, 1933), commenting briefly about them at a few Wednesday sessions. In November 1929, he compared the “fragmented” thinking of a “feebleminded schizophrenic” patient with Köhler’s description of an ape unable correctly to stack boxes in order to reach a hanging fruit. One year later, he used Köhler as an example of psychologists who insisted upon “either giving a too complex and confused explanation or completely refusing to explain simple facts,” and in October 1931 he dismissed the Gestaltist’s recent criticisms of his own views as a reflection of “dualism.”14 A confluence of factors now pushed him toward a more sustained and consequential intellectual engagement with Gestalt. Most important was the unexpected arrival at Koltushi of two anthropoid guests.
* * *
In August 1933, Petr Denisov brought the chimpanzees Roza and Rafael to Koltushi. This gift reflected the attempt of Pavlov’s Communists to challenge his reductionist, mechanistic positions by weaning him from exclusive reliance on the dog and engaging him in the investigation of anthropoids and humans.
Denisov’s timing proved fortuitous. Pavlov was increasingly preoccupied with psychology and Gestalt, whose leading proponent, Köhler, frequently invoked research on apes. Furthermore, he was deeply attached to Koltushi but lacked an absorbing scientific project there. Progress on his genetics and breeding project was mired in vexing problems with typology and stalled by slow construction of the science village—after which he faced another long wait while dogs were interbred and new generations produced.
Roza and Rafael immediately engaged his curiosity. He visited them daily after they first arrived and, beginning in fall 1933, during regular Friday sojourns to Koltushi and lengthier stays during summer and winter breaks. By early 1934, he was hooked on his anthropoids and had experiments with them systematically filmed. Intent upon expanding this line of investigation, he planned to send Denisov to study Yerkes’s methods for the care and investigation of primates.15
Pavlov’s chimp studies were shaped by his preoccupation with Gestalt’s challenge to associationism. He observed Roza and Rafael while reading—and, at Wednesday meetings, often polemicizing against—Köhler’s Mentality of Apes and Gestalt Psychology. For the first year and a half, Denisov’s experiments were based largely on those Köhler described in that former work, as Pavlov wanted to examine Köhler’s evidence for ascribing “insight” to chimps and denying that their behavior could be explained in associationist terms.
Köhler argued that the human mind engages the world not by accumulation of small perceptions as postulated by associationists, but rather by grasping wholes, configurations, or Gestalten. The key concept here was “insight”—apprehension of the structure of an entire “perceptual field,” of the general lay of the land that lurked beyond simple experiences. A melody, for example, is not the simple sum of its notes, and apprehension of a melody does not involve a simple summing of individual nervous traces but rather a grasp of the melody’s organization, of the relationship between its notes. If the notes are changed but the organization remains, the melody is preserved and recognized. There was nothing necessarily “mystical” (as Pavlov would have it) about this idea; for Köhler and his fellow Gestaltists, the perceiving subject simply had grasped the whole.
From 1914 to 1917, Köhler examined insight in apes by conducting experiments in which an obstacle prevented the direct and simple solution of a problem. For example, food was visible, but access to it was obstructed by a fence, confinement in a box, or by the height at which it hung from a tree. Watching his chimps solve such problems, he concluded that they indeed demonstrated insight, as manifested by the lack of a smooth learning curve and by a pause (when, presumably, the chimp was thinking) before taking a decisive action. “Where a new ‘idea’ leads to new action in subjective experience, the observer will see one stream of visible development ceasing and another one starting independently.”
In The Mentality of Apes, Köhler emphasized the inability of associationists to explain such learning and the need for a psychology based upon perception of form and structure, but largely avoided explicit theoretical arguments. In Gestalt Psychology, he drew upon his studies of anthropoids and research by other Gestaltists to criticize associationism and develop Gestalt theory. Here he blended criticisms of “the theory of original and acquired reflexes” into a pointed critique of Gestalt’s main antagonist, behaviorism. For him, Pavlovian doctrine was part of the conservative consensus that retarded progress in the young science of psychology. For those “who are deeply convinced that the theory of original and acquired reflexes is the whole truth about the nervous system, there is no real incentive for the further observation of natural behavior, since they do not feel the need of any new information or new functional concepts.” He also expressed doubts about the “law of contiguity” that underlay Pavlov’s research (questioning whether the temporal contiguity of two stimuli led to a CR independent of the stimuli’s qualities) and dismissed Pavlov’s schema as a mere rephrasing of associationist doctrine: “From the viewpoint of machine theory it sounds a little nicer if we talk about conditional reflexes instead of associations. I do not find, however, that the first concept is clearer than the second.”16
During the first phase of their primate experiments—from about August 1933 until spring 1934—Denisov and Pavlov attempted mainly to acquaint themselves with their new experimental subjects.17 They discovered that the chimps were distinguished by their “extraordinarily great development of the investigative reflex” and that Rafael was more easily motivated by food and so was a better experimental subject than Roza (for whom a more effective reward was freedom). According to Pavlov, Rafael conceived a dislike for him, so the chief’s presence interfered with the chimp’s pursuit of an experimental task by exciting him and generating “waves of negative induction.” When both chimps were motivated to undertake the same task, Roza learned more quickly—but the more easily manipulated Rafael became the favored experimental subject.18
Between May and December 1934, the experiments at Koltushi followed Köhler’s basic design, challenging Roza and Rafael with a variety of tasks that defied simple, direct solutions. In one set of experiments, Denisov had the chimps watch while he placed some fruit in a sturdy cubical box, which was then covered with a lid that had either a round, triangular, or square aperture. Scattered before the chimps were fifteen rods (keys) with the same three shapes; these could be inserted in the corresponding aperture to open the lid. Once a chimp had done so successfully, the lid was replaced. (This experiment was altered for Roza in order to motivate her with the prospect of freedom.) Over a period of two or three months, the chimps learned to put the proper key in the aperture. In another experimental series, food was suspended from the ceiling of the cage beyond the reach of the chimp and six rectangular boxes were scattered in the vicinity. After three or four months, Rafael succeeded in piling the boxes and climbing them to obtain the food. In one variant of this experiment, boxes of various sizes were used to test the chimps’ ability to build a serviceably stable configuration.19
Observing these trials, Pavlov agreed with Köhler that chimps seemed unable to cooperate in the solution of problems and distinguished between forms quite poorly. Roza and Rafael’s successes sometimes led him to enthuse about the “elements of scientific creativity” they displayed, but Rafael’s difficulties when a task required a recognition of form led Pavlov to dismiss him as a “fool” and a “ninny.” “What a porridge he has in his head!” The crucial question, however, was not what the chimps could accomplish, but rather how they accomplished it.20
Pavlov’s observations and ruminations led him in two related but contradictory directions. On the one hand, he satisfied himself that Köhler was incorrect—that the chimps’ performance was completely explicable as the result of associations governed by CRs and that the concept of “insight” was a fictitious muddle. He polemicized enthusiastically, and sometimes angrily, at Wednesday meetings from October 1933 onward against Köhler’s interpretations, the “mysticism” of Gestaltists, their flimsy arguments against associationism, and their strange attachment to overly complex analyses. Yet he also had a second reaction, an undercurrent that grew increasingly stronger: contemplating the behavior of his chimps, deploying his doctrine to encompass their actions, and confronting Gestalt theory forced Pavlov to reexamine and then amend his own views.
Armed with his observations of Roza and Rafael, then, Pavlov completely rejected Köhler’s claim that anthropoids learned not through the accumulation of associations but through an insight into the general perceptual field. In experimental trials, Pavlov insisted, his chimps’ learning process began with what Americans termed “trial and error.” Rafael had no doubt associated sticks with tools of action from previous experience, and he tried first one key, then another, to open the box that contained food. When one key did not work, it was “differentiated” (that is, became a CI), so he discarded it. When another key worked, its use was reinforced. When this occurred several times, there was formed “a connection between the visual sight of this stick and success”—the key had become a CS. When Denisov changed the shape of the aperture, Rafael first employed the reinforced key, which after several failures was “differentiated,” and so forth.
This behavior, Pavlov insisted, was all based on simple associations, and did not differ essentially from that of dogs and humans. Since a chimp’s lower extremities performed the same functions as its upper ones, it could of course manipulate a stick in more ways than a dog. Yet this greater mechanical ability did not result in a different path to the solution of a problem, and one could not attribute to chimps an intelligence absent in the dog.21 Addressing Köhler’s notion of “insight,” Pavlov professed astonishment. “This Köhler, in my opinion, saw nothing that the apes actually showed him.” He mocked him at a Wednesday meeting:
When the ape is given the task of obtaining high-hanging fruit, and when it needs a tool to attain its goal...all the unsuccessful attempts to attain the goal do not, in Köhler’s opinion, prove rationality. So much for the method of trial and error. After many failures, fatigued, the ape goes to the side and sits, undertaking nothing else. After a period of sitting, of peace, it returns to work and achieves the goal. Köhler considers this sitting to be proof of its rationality. Literally, gentlemen! According to Köhler, while the ape sits it is conducting rational work. How do you like that! The silent inaction of the ape is the proof of rationality. But that the ape acts with a stick, piles up boxes—all this is not rational.
Köhler’s “insight” was, at most, a term for the point at which various associations came together in the form of “new knowledge, a new connection, new connections of things.” This was the moment “when the physiological and the psychic completely merge,” when the associations formed by trial and error yielded an idea, an understanding, knowledge. “That is all there is. And all our human thought occurred also in just this way.”22
Gestalt psychology was not merely wrong—it was a retreat from real science, a “word game” with profoundly negative consequences for psychology. Adding nothing to real knowledge, it threatened to destroy what was “most basic and most true” in psychology: the analytic approach of associationism and the research on processes of connection and synthesis that it entailed. “These psychologists are strange!” Pavlov exclaimed in November 1934. Nobody “doubts wholeness,” but no true scientist could doubt that a biological whole could also be “broken down into systems of circulation and digestion; that digestion can be broken down into stomach, intestines, gastric glands, and so forth.” The Gestaltist slogans that the mind is not “a simple sack of potatoes, apples, cucumbers” and that “our behavior does not consist only of the sum of reflexes” merely expressed a long-standing and obvious truth. “Once you have an organism, it is clear that all its elements interact with one another, just as in a chemical body oxygen, hydrogen, and carbon act according to the manner in which they are placed in the molecule. All this has long been known. Nobody said that it is a simple sum. Once it is a system, those elements of course interact one with the other, and our task is to begin the investigation of their interaction.”
What was new—Gestalt’s “scientific sin”—was its opposition to associationism, to analysis of the parts. “But why should the product of the highest animal organism, the phenomena of our subjective world, be studied by a different approach [than in chemistry and biology], one that forbids disaggregation and excludes analysis?” The only possible explanation was the continued hold of “dualism, in the form of animism, that is, the notion of a special substance opposed to the rest of nature.” This subject put Pavlov in a most militant mood. At a Wednesday session in May 1934, he characterized both Köhler’s and Yerkes’s “attempts to depart from the truth” through the empty phrases of Gestalt as “repulsive,” “nonsense,” and “childish.” In January 1935, he introduced a discussion of Gestalt and chimps with the words “We are at war with Köhler. This is a serious struggle with the psychologists.” He would refute them completely at next year’s International Congress of Psychologists.23
Yet his confrontation with Gestalt and observations of chimps had raised some important new challenges for his own doctrine. Confident that these could be accommodated within his views, he addressed them experimentally and interpretively. Three separate but related lines of investigation—each focused upon a different organism—would converge from late 1934 through early 1936. Further observations of chimps raised questions about his long-held view that “conditional reflex” and “association” were but physiological and psychological terms for the same phenomenon. Addressing that same problem, and responding to Gestaltists’ insistence that organisms responded to relationships (rather than objects), he conducted trials on dogs using a new experimental design. Meanwhile, his observations of humans also complicated his thinking about the nature of CRs and their relationship to associations and perception. These lines of investigation and Pavlov’s final ruminations about these key issues culminated in two unfinished manuscripts: The Intellect of Anthropoid Apes and Psychology as a Science.
While rejecting Köhler’s argument that anthropoids displayed a qualitatively different form of intelligence than that of dogs, Pavlov did, however, concede over time that there were important differences between the two organisms that influenced their intellect and learning behavior. At a Wednesday gathering of September 1934, he noted that anthropoids differed by virtue of having four hands that allowed them “to enter into very complex relations with objects” and so to form “a mass of associations, which does not exist in other animals.” The chimp was an “ideal tightrope walker,” and, watching it “confirm empirically” the stability of its stacked boxes, one saw the complex interplay of its multifold associations—“tactile, muscular, visual, and so forth.” Roaming Koltushi freely and tackling experimental tasks, Roza and Rafael constantly displayed the multidimensionality of their interaction with their environment—simultaneously manipulating, inspecting, and smelling objects (and people) in their domain.24
Pavlov did not say so, but, regardless of its species characteristics, the experimental dog harnessed to the stand and exposed to one stimulus at a time could only manifest some of its qualities. Constrained by experimental design, that dog embodied Pavlov’s long-standing model of the organism responding to individual stimuli at specific times and so developing—or failing to develop—unilinear chains of reflexes. The unconstrained chimp, on the other hand, was free to express its powerful “investigative instinct” simultaneously with eyes, ears, nose, and hands, and so embodied the emerging model of systematicity and the multidimensional mosaic.
By January 1935, Pavlov and Denisov had completed the second phase of their primate experiments. Having tested and re-interpreted Köhler’s observations, they now began phase three: encouraging Rafael, as Pavlov put it, to “broaden his ‘natural science.’” The chimp had already learned a great deal—to evaluate the significance of apertures and keys, to construct a building from boxes and climb onto it, and even to extinguish fire with water. For Pavlov, these “scientific achievements” represented the fruit of many “more or less elementary associations.” Now the experimenters placed before him more complex tasks involving “the association of associations.” Having already learned that water extinguishes fire, and confronted with a “zone of fire” (candles) blocking his way to food, could Rafael learn to associate a faucet with water, turn it on, and set the water flowing toward the fire? Having learned about correctly shaped keys and stacking boxes, could he open a door with a suitable key, enter a room, extinguish a fire that prevented him from exiting onto a platform, climb to the platform, and stack the necessary boxes to obtain a high-hanging fruit?25
* * *
Sometime in early 1935—probably in January, during Christmas break—Pavlov gathered his thoughts in a manuscript titled The Intellect of Anthropoid Apes. This never-completed essay reflected his confidence about some issues and his difficulties with others. He confidently repeated his comments at the Wednesday meetings about Köhler’s fallacies; about the essential continuity between the learning process in dogs, chimps, and humans; about the superior intelligence of the chimp due to the use of its four hands to develop “extraordinarily more complex mechanical interrelations” with its environment; and about his own ability to explain the chimp’s problem-solving process in terms of associations.
He was much less confident, however, doing so in terms of conditional reflexes. It was a crucial distinction. For Pavlov, a CR was a connection between a subcortical and cortical point. He sometimes used the equivalent formulation that it was a connection between an unconditional exciter of the subcortex (such as food, pain, and sex) and an indifferent agent (such as a buzzer, light, or metronome) that excited a point in the cortex and so became a signal for this subcortical exciter. Moreover, this signaling always resulted from a temporary linkage due to simultaneity. These specific physiological features underlay both the laws that Pavlov had identified as governing the conditional reflex and its biological role as the mechanism through which organisms adapted to a changing environment. “Association” was a looser, more general term than “conditional reflex,” a reference to the connection between two or more experiences, sensations, or images. Pavlov himself had long assumed that this psychological term was synonymous with his physiological term “conditional reflex,” that they referred to the subjective and objective aspects of the same process. But now he was not so sure.
In The Intellect of Anthropoid Apes, he analyzed the “elementary associations, or knowledge, or ideas” involved when the chimp piled boxes to reach a high-hanging fruit. He identified ten, describing some in more detail than others in this rough draft intended for his eyes only. Here, in paraphrase, is his list, with direct citations (in italics) of some notes to himself: 1. Boxes must stand exactly under the fruit; 2. They must be placed one upon the other. But how?; 3. The chimp crawls onto the box to test its stability; 4. It moves a second box near, places it on the first box, and tests for stability by crawling on the boxes. This is a kinesthetic association; 5. When a number of boxes are piled up, it tests for stability by inspecting the boxes, and instead of the kinesthetic association, there is another visual association; 6. Over time, the chimp looks at the structure from a greater distance to see if it is of sufficient height; 7. The chimp tests the height by climbing on the boxes; 8. If rectangular boxes are used in the trial, the chimp might reach the necessary height without using all the boxes; 9. If the boxes are not rectangular, the chimp needs to use all six boxes; 10. If boxes of various sizes are used in the trial, the chimp must undertake the difficult task of stacking them in the correct order, a process that is probably governed by sight.26
Here, then, were the associations involved in the chimp’s successful stacking of boxes to obtain a high-hanging fruit. There was no need to resort to Köhler’s “insight” or other Gestaltist obfuscations. In this “visible and indubitable act of thinking, recognized as such by psychologists, there is nothing other than simple and complex associations.” But what was the nature and mechanism of these associations? As Pavlov wrote in the margin of his manuscript: “What is united with what?” In his text, he reasoned: “Nevertheless, the first condition for the formation of associations, as we saw with the conditional reflexes of our dogs, is the constant existence of interest, that is, of a certain tonus of the cortex.” CRs were rooted in subcortical drives, so, assuming that the “associations” he was pondering were CRs, they were rooted in the chimp’s quest for the fruit. “All associations must represent stages toward the achievement of the goal.” That is, not only the fruit itself, but increased proximity to the fruit reinforced an association. Each action that enhanced the height and stability of the chimp’s stacked boxes was reinforced, and eventually became the type of volitional action described by Konorski: “A movement that initially, accidentally, produces a certain change in the mechanical relations of things, when it is repeated with the growing proximity to the fruit is reinforced and produced in reverse by the animals themselves, as when the paw [of Konorski’s dog] was raised and this was reinforced with food. Other exciters are also united with this, most frequently visual ones—that is, a certain image of the successful arrangement of things. Since kinesthetic and visual stimuli are linked one with the other, the visual stimuli or their traces are directed toward the kinesthetic [stimuli] and condition fulfillment of certain movements.”27
This explanation relied, then, upon Konorski’s “motor conditioning,” upon various visual and kinesthetic associations, and upon relations between these associations. Pavlov referred to “a lawful chain of these associations, that is, the association of associations,” but what he described was not so much a chain as an interconnected mosaic. “Of course there is the obvious fact of the association of associations. The elementary [associations] are united among themselves in the most varied way...In this manner, there is acquired a system of nervous processes that with repetition is implemented more and more easily, becoming more and more fixed.”28
Pavlov relies heavily here upon the terms “associations” and “the association of associations,” and much less upon “conditional reflexes.” The key problem, as he had noted in his marginalia, was: “What is united with what?” The “association of associations” he describes differed fundamentally from his old definition of the CR as a temporary connection between a subcortical and cortical point, a signal established through simultaneity. Rather, he repeatedly invokes the connection between, for example, the chimp’s kinesthetic and visual associations, which are “united among themselves in the most varied way.” He could explain the chimp’s complex interaction with its environment and its progress toward a high-hanging fruit by using his notions of “the association of associations,” systematicity, and the cortical mosaic. But this explanation drove a wedge between the allegedly synonymous “association” and “conditional reflex.” He understood this clearly and pondered the implications over the next months.
He concluded his manuscript on a poetic note: Roza and Rafael had revealed the origins of that most exalted product of subjectivity, human science itself. The fruit of “extraordinarily complex mechanical interrelations with surrounding conditions, multidimensional actions upon their environment,” their achievements represented “the beginning of the study of the mechanical side of nature, which is the embryo of science. With mechanics, with its laws, began also humanity’s precise science, both ancient (Archimedes) and modern (Galileo).”29
Meanwhile, Pavlov’s confrontation with Gestalt had also generated a new set of trials with dogs. In the winter of 1934, he was reading Columbia University psychologist Robert Woodworth’s Contemporary Schools of Psychology (1931) and commenting agitatedly about it at Wednesday meetings. Woodworth described one experiment that Gestaltists touted as evidence that animals responded not to separate exciters but to relationships. The experimenter placed in a chamber two gray boxes, one light and one dark, and trained the animal to approach the lighter one. These boxes were then replaced with two others of different gray hues, with one lighter than the other. The animal chose the lighter one, though this was a different shade than the box it had been trained to choose in the first trial. So, according to Gestaltists, the animal responded not to separate exciters but to relationships. “They consider this a strong argument,” Pavlov commented dismissively. He and his coworker Sergei Kleshchev repeated this experiment with a dog—establishing one tone as a CS and another as a CI, then using two other tones that stood in the same relationship. The dog indeed formed a CS and a CI to the second set of tones much more quickly than it had to the first set. So, Pavlov concluded, CRs could indeed be formed to relationships—but “there is nothing special in this,” and it hardly constituted a criticism of associationism.
Yet, intrigued, he assigned Kleshchev and Dolin to investigate further. At Wednesday sessions of March and October 1935 he reported on their successful development of CRs to the relationship between two exciters. For example, they first established M120 as a CS and M60 as a CI. When a light flashing at different speeds was then used as an exciter, the dog responded to intervals of 120 per minute as a CS and to intervals of 60 per minute as a CI. Very interesting, but hardly unexpected, Pavlov pointed out; he had long maintained, after all, that “any specific state” accessible to an animal’s sensory apparatus could become a CS. That was of course true for intensities of color or time intervals. One could call this “Gestalt” or use an English term such as “form or figure.” “Name it as you please,” but, again, it hardly undermined associationism.30
Pavlov’s routine was disrupted from March through August 1935 and he was separated for months from his lab, but he continued to ruminate about Gestalt, chimps, and associations, and he resumed discussion of these issues immediately upon resumption of the Wednesdays in October 1935.
One month later, he dropped a bombshell on his coworkers: contrary to three decades of lab doctrine, he announced, “conditional reflex” and “association” were not synonymous. The immediate occasion for this revelation was Dolin’s report about an experiment he had conducted on humans. Having kept his subjects in darkness for half an hour, he then flashed a beam of light at them. Their vision suffered for a period of time, recovering gradually according to an identifiable curve. When they had recovered and he flashed the light again, they again suffered a decline in their vision. He then sounded M120 at the same time as he flashed the light. Over time, M120 elicited precisely the same effect as the light—causing diminished vision. Very nice. The auditory cells in the cortex had clearly become associated with the visual cells, forming a “single functional unit.”31
This was clearly an association, but, Pavlov asked, was it a conditional reflex?
You see, “association” is a generic concept, that is, the uniting of that which was earlier separated, the unification or generalization of two points into a functional relationship, their merger into one association; but “conditional reflex” is a specific concept. This, too, of course, is the linkage of two points that were not earlier united, but it is a partial case of such a linkage with a specific biological significance. In the case of the conditional reflex, we have essential characteristics, constant traits of a certain subject (food, an enemy, and so forth) replaced by temporary signals. This is a partial case of associations.
And here is another example: when phenomena are linked due to their simultaneous action upon the nervous system, linking two phenomena that are actually constantly joined in reality. This is another species of that very same association; this is the basis of our knowledge, the basis of the main scientific principle of causality. This is another species of association that is perhaps no less, and probably more, significant than conditional reflexes, than the connection of a signal.
And, finally, a simple case...when, for example, psychologically, two sounds are linked that have nothing in common, that are linked only by the fact that one is repeated after another, and they finally are linked, one elicits the other.32
One must distinguish between all these cases, he informed his coworkers, for their mechanisms and biological significance might well differ.
So had Dolin really developed a conditional reflex in his human subjects? Pavlov had initially doubted it, but now thought that indeed he had, “because the light produces a chemical reaction [causing the deterioration of vision]...and instead of the light the metronome does the very same thing.” Thus, M120 had become a temporary signal for the light, and this phenomenon was indeed a conditional reflex.33
By this same logic, however, many of the associations that Pavlov had deployed in his earlier manuscript to explain Rafael’s successful stacking of boxes to acquire fruit did not qualify as CRs. He continued:
But when an ape builds its structure in order to obtain fruit, one cannot term this a “conditional reflex.” This is a case of the formation of knowledge, of the grasping of the normal connection between things. This is another case. Here one must say that this is the beginning of the formation of knowledge, the grasping of a constant connection between things—that which lies at the basis of all scientific activity, of the laws of causality.34
When Rafael associated the scent of fruit with something good to eat, this was a conditional reflex; but when he associated that scent with the way a fruit felt in his hand, this was not a conditional reflex. No temporary signals were involved. Similarly, when he climbed upon his stacked boxes, sensed kinesthetically their stability or instability, and joined this to his visual sense of how they looked, this association was not a conditional reflex. The chimp, rather, was forming an association between two permanent, “objective” characteristics of an object. This was empirical “science,” the recognition of permanent causal relations in the environment. Furthermore, although Pavlov didn’t develop this point in his brief commentary at the Wednesday session, Rafael’s “associations of associations” (as the chief had emphasized in his unpublished essay) involved linkages between cortical points, not connections between the subcortex and the cortex.
Concluding his extended remark, Pavlov posed a question to his coworkers that apparently struck them dumb, coming as it did some thirty years after the issue had supposedly been resolved: “What is a conditional reflex?”35
Shortly thereafter, at Koltushi during the 1935–1936 winter break, he synthesized his thoughts about Gestalt, CRs, and associations in a manuscript titled Psychology as a Science. This was perhaps a draft of the presentation he eagerly anticipated delivering in Madrid to the upcoming April meeting of the International Psychology Congress.36
His central theme was the physiology of associations and its relationship to the great achievements of associationist psychology. The “most important and unquestionable achievement of psychology as a science,” he began, was its establishment of “the fact of the connection of subjective phenomena—of the association of words,...thoughts, feelings, and impulses to action.”
That achievement was threatened by “the new, fashionable current in psychology, Gestalt psychology,” which undermined psychology’s great achievement at the same time as physiology was affirming it.37 He then launched into his customary criticisms of Gestalt, attributed its popularity to an animistic belief in a “special substance” that distinguished mind from the rest of nature, and proposed to illustrate the scientific fruitfulness of associationism with his own investigations of “the objective phenomenon that corresponds to the psychologists’ ‘association.’”38
He was now able to invoke experiments with dogs that he had initiated with the turn to systematicity. These adopted a design that he had avoided when preoccupied earlier with “chains of reflexes,” but which had been developed by other researchers schooled in his methods. In the mid-1920s, two of Pavlov’s disciples interested in psychiatry and child development, Krasnogorskii and Ivanov-Smolenskii, had conducted experiments on humans in which they demonstrated that humans routinely formed connections (associations) between two cortical points—that is, between two indifferent stimuli, even if these were not reinforced by a subcortical exciter. For example, when a child was exposed simultaneously to light, a buzzer, and a metronome, and the light was then established as a CS for food, the buzzer and metronome also elicited a food response. Pavlov’s independent former coworker Zelenyi had attempted analogous experiments on dogs, but had failed to produce similar results.
From October 1932 through at least January 1934, Pavlov’s coworkers I. O. Narbutovich and N. A. Podkopaev attempted similar experiments, and by October 1934 they had acquired positive results where Zelenyi had failed. Using movement reflexes, they demonstrated that dogs, like humans, could form associations between two indifferent cortical stimuli. First, by simultaneously exposing a dog to a rotating figure and a tone, they established an association between the two. Then they established the rotating figure as a CS for electrical shock. (That is, the rotating figure, like shock itself, now elicited a “defensive reaction.”) Finally, they demonstrated that the tone alone now elicited this same defensive reaction. The CR of light and shock had been combined with the association between light and tone. The experimenters then made the same point in a different fashion. In separate trials, they flashed a light in front of a dog, which responded by looking intently at the light source (a specific “orientational response”), and sounded a tone behind the same dog, which turned toward the noise (another, equally specific, orientational response). They then repeatedly flashed the light and sounded the tone simultaneously to form a connection between the two. The dog now responded to each stimulus with both orientational responses. Narbutovich and Podkopaev wrote up the first part of their experiments for the chief’s in-house journal, concluding, according to longstanding lab doctrine, that they had demonstrated that “the terms ‘conditional reflex,’ ‘association,’ and ‘temporary connection’ are entirely identical.”39
As Pavlov worked on his essay a year later, however, he interpreted their results very differently. In the spirit of his comments at the earlier Wednesday gathering, he distinguished between the general term “association” and the specific “conditional reflex.” For him, in their initial trials the experimenters had created a CR (between the rotating figure and electrical shock), a cortical association (between the rotating figure and the tone), and, by combining the two, an “association of associations.” The defining general characteristic of all these associations was the linkage of different cells through temporality. The CR was just one “obvious example” of this process.
In his manuscript, Pavlov explored the physiology of associations—and the distinction between CRs and cortical associations—by tying Narbutovich and Podkopaev’s trials to the classical puzzle-box experiments of American psychologist Robert Thorndike. Pavlov had long recognized Thorndike’s priority in setting an objectivist path for psychology, but Thorndike’s research—and that of the American behaviorism that followed—had influenced Pavlov’s intellectual development little if at all. In winter 1934, however, Pavlov had read a summary of Köhler’s criticism of Thorndike’s conclusions and so had himself engaged the American’s earlier experiments.40
As he pondered the relationship of his lab developments to associationist psychology, Pavlov translated the lessons that he had learned from Konorski, chimps, and his coworkers’ recent experiments into an analysis of Thorndike’s puzzle-box experiments. Thorndike had confined hungry dogs and cats within a box fitted with a door that could be opened by various devices, and had observed them as they attempted to open the door and escape to a food bowl. The animals, he concluded, learned by “trial and error.” When an animal’s action happened to bring a favorable outcome (“satisfaction to the animal will”), it was reinforced and repeated with increasing frequency; when an action brought an unfavorable outcome (“discomfort to the animal will”), it was weakened and gradually disappeared. After some resistance Pavlov had incorporated this principle, which Thorndike termed the “Law of Effect,” into his own interpretive repertoire by way of Konorski, who had translated it into Pavlovian terms and convinced the chief that it represented a “second type” of CR. And Pavlov had deployed this “Law of Effect” in partial explanation of chimp behavior in his manuscript on the Intellect of Anthropoid Apes.41
Now, in Psychology as a Science, Pavlov analyzed the behavior of Thorndike’s dogs and cats as the result of the same non-CR associations that he had identified in the puzzle-solving of Roza and Rafael. Thorndike’s animals had opened the door of their puzzle box—just as Pavlov’s chimps had fitted keys to corresponding apertures and stacked boxes to obtain fruit—by “acquiring knowledge of the relations between the material objects of the surrounding environment—and so power over it.” This type of knowledge (non-CR associations) represented the “embryo of science” and was accumulated through the same trial-and-error method by dogs, cats, chimps, and humans. Science differed only by its systematization of specialized knowledge and the focused inquiries through which it built upon that already achieved.42
How, then, did CRs differ from these other associations? Both were connections between cells that repeatedly fired simultaneously, but a CR did not represent the “embryo” of scientific knowledge. In forming a non-CR association, the dogs in Thorndike’s puzzle boxes and the chimps stacking boxes acquired knowledge about “the relationship of external things among themselves.” For example, they learned that a particular image of stacked boxes corresponded to a stable structure or that a specific movement would open a door with a particular lock. In forming a CR, on the other hand, an animal gained knowledge about the “relationship of separate objects in the environment to itself.” For example, it learned that a flashing light signaled an imminent shock or opportunity to eat.
A second difference between the CR and the non-CR association resided in their physiological natures. The CR involved the connection between a point in the subcortex (excited by unconditional drives for food and sex) and a point in the cortex that responded to a particular external exciter. In a non-CR association, “certain kinesthetic, tactile, and visual excitations from certain external objects and their position become connected with other, also visual [excitations] and perhaps also with kinesthetic excitations from one or another external objects.” These were connections between two cortical points. A third difference followed from this: for any association to be formed, the cortex must have the necessary “tonus” (energy, excitability). In the CR, that tonus was provided by the involvement of the subcortex (by the energy of primal drives). Non-CR associations did not involve the subcortex and so required an alternate source. Pavlov illustrated this through Podkopaev’s and Narbutovich’s experiments. They had initially failed to produce an association between two indifferent cortical exciters, and had succeeded only when they varied these exciters sufficiently (for example, allowing the tone to waver a bit) in order to maintain the animal’s orientational reflex (that is, its interest). The energy (or excitability) that they thus acquired was a necessary substitute for that provided in the CR by the primal drives of the subcortex.43
In Psychology as a Science, Pavlov also sketched some thoughts for further research. The scope of CRs as a phenomenon had been narrowed, but that of Pavlovian method and explanation had been broadened dramatically. He was preparing to analyze systematicity and all associations—and associations of associations—in the same manner as he had analyzed CRs, and thus to expand his research into new areas of psychology.
The problem of second- and third-order CRs had now been transcended by a new model, not of serial exciters and chains of reflexes, but of a structured system of interrelated CRs and associated associations distributed across the cortical mosaic. What were the different types of association and their laws? What were the dynamics and mechanisms of systematicity? By January 1936, Pavlov’s coworkers were tackling new subjects opened up by this new perspective—for example, on the reciprocal relationship of the subcortex and cortex, and on the animal’s perception of musical sounds and melodies.
At the Wednesday meeting of February 12, 1936, Dolin’s further experiments on the dog’s response to rhythms and intervals elicited a lively discussion between him, Podkopaev, Maiorov, and the chief about the nervous mechanisms responsible for the animal’s formation of CRs to “relationships, form, Gestalt.” Was this due to mutual induction or, perhaps, to “zones of inhibition”? The next Wednesday session ended with Pavlov’s return to this question and his tentative endorsement of the latter explanation. “Think about it,” he said, concluding the meeting.44
His thinking remained firmly framed by the worldview of a lifetime and his continued faith in the power and precision of his methodology. Indeed, the very notion of the “conditional reflex” had become largely metaphorical during his long quest. It had long ceased to denote the relatively simple stimulus-and-response of a sensory-motor nervous unit or a dog salivating to the sound of a buzzer, and had come to stand in for Pavlov’s determinist ideal itself—for a mechanistic explanation of the complex phenomena that he confronted. Pavlov’s quest had outstripped the rich potential of that metaphor and led him to subsume it within the even broader metaphor of “associations.”
At age eighty-six, he had shown himself both true to a lifetime’s mental habits and remarkably open to changing his mind. He had transcended the accumulated problems of his old model by adopting a new one that generated new, exciting research possibilities and a broader path to the horizon.