Back to CQ Homepage

The Effects of Atomistic Ontology

 on the History of Psychology


by Teed Rockwell

(CSU Hayward)




 This article articulates the presuppositions that psychology inherited from logical positivism, and how those presuppositions effected the interpretation of data and research procedures. Despite the efforts of Wundt, his most well known disciples, Titchener and Külpe, embraced an atomistic view of experience which was at  least partly responsible for many of their failures. When the behaviorists rejected the introspectionism of Titchener and Külpe, they kept their atomism, using the reflex  as the fundamental psychological unit, rather than the sense datum. When cognitive psychology embraced functionalism, it made the most radical break from atomism in psychology's history, which is reason for optimism. However, there are still certain presuppositions which make cognitive psychology vulnerable to some of the weaknesses of atomism, and research in a variety of areas is starting to uncover them.



            There is much justification for the psychologist's claim that it is possible and proper for the psychologist to ignore philosophy. If it is possible for psychology to be a science at all, it must also be possible to take for granted that certain kinds of things exist, and then start measuring them and/or theorizing about them. But whenever a science enters into a crisis, whenever its research program starts to disintegrate in some way, this usually means that one has to start questioning the set of assumptions about one's subject matter that formerly made the research project work so well. This set of assumptions is called an ontology  by philosophers. Ontologies include implicit commitments to what kind of things exist, what determines the essential nature of those things, and what things are more "ontologically fundamental" than others. When A is more ontologically fundamental than B, it is sometimes convenient to say that B is made of "A"s in some sense, although that phrase contains ontological assumptions of its own which are far from necessary. Those assumptions are an essential part of an ontological position which has been so widely accepted by scientifically minded people that it is often assumed to be the only possible scientific ontology. This position is called atomism, and the following commitments are usually considered to be essential to it.


            According to atomism, the universe consists of certain fundamental individuals, each of which "is what it is" 1) without reference to any of the others, or 2) without reference to any of its parts[1]. What physicists call atoms are not thought of today as being atoms in the second sense, for they are ontologically dependent on ("made of") quarks. But physicists gave them this name originally because it was assumed that they were atoms in this sense, and the fact that physics continues to search for something like quarks shows that the epistemological goals of atomism still motivate modern physics. And because psychology has often used physics as the highest exemplar of a science, it has been widely assumed that psychology must have atoms of its own if it is to become a science. I will try to show in this paper that this assumption is dangerously misleading, and continues to be so because it has been preserved through many of psychology's scientific revolutions.


            Atomistic ontology's greatest danger comes from the fact that most forms of it deny the existence of ontology altogether. It seems plausible, after all, that if atoms are the only things that exist, and each atom is genuinely independent of every other atom, there should be nothing that needs to be said about Being itself. Those philosophers who have defended atomism have usually admitted that an atomistic ontology is something of a paradox (As Wittgenstein's Tractatus did with its ladder metaphor), have asserted that we must be skeptical about all ontological and metaphysical claims (Hume), or have been driven by atomism's inconsistencies to eventually reject it, (essentially all of the logical positivists). However, for those that presuppose a philosophy rather than actively formulate one, atomism has always had a strong appeal. Indeed, the crudest forms of atomism are merely semiverbalized hostility towards philosophy itself, which are made explicit only to justify spending as little time thinking about philosophical issues as possible. This hostility makes it extremely difficult for an atomist to see what sort of problems his ontology is causing him, because it is only by explicitly formulating an ontology that one can conceive of alternatives to it.


            Many researchers in the physical sciences can do their job without having to think about ontology at all (although high level theoretical physics frequently has to discuss specific ontological claims). For this reason, psychologists are often uncomfortable discussing ontology and other philosophical issues. It reminds them of their differences from the physical sciences, and seems to imply a regression to the days of Wundt and James, when psychology was considered to be a branch of philosophy and not a "real" science. But now that philosophers have again been admitted into the psychological discussion with the creation of the cognitive science community, hopefully there will be less denial when ontological crises arise, and less likelihood of their festering into stalemates.


            Admittedly, once one does seriously attempt the ontological enterprise, it is hard not to sympathize with the atomist's skepticism towards the whole project. The use of ponderous Teutonic neologisms never successfully hides the fact that all ontologies are a ragtag collection of poetry and hunches. And most of the time differences in ontological assumptions are purely academic in the worst sense, utterly lacking in what James called cash value. And even when consistent adherence to certain ontological principles can be genuinely pernicious, it is often possible for practicing scientists to avoid these dangers simply by being inconsistent.


            There are, however, times in the history of science when certain stalemates can be best understood by articulating the ontologies of the researchers.  If the following analysis is even partially correct, atomistic ontology has been a great source of trouble for psychology, and consequently any hopeful alternative deserves serious consideration.


            The Introspectionists and the Myth of the Given


            If one denies the existence of some kind of a priori  knowledge[2], there is only one way we can ever build our theories on a foundation of certainty. The "this" which confronts us in any given moment must be present to us in its entirety, distinct and independent from each other "this" we confront in the past and future. Thus 'tabula rasa" epistemologies like Locke's must claim that our experience is divided up into discrete and independent sense data, and if we believe our knowledge reflects the world, we must believe that the world is similarly divided. This further implies that our innate ignorance of the world can be laboriously overcome by carefully observing these directly observed moments, and then piecing them together like parts of a jigsaw puzzle. If this were the nature of things, the introspective psychology founded by Edward Titchener and Oswald Külpe in the early twentieth century would have been the most likely candidate to discover and categorize those directly observed moments.


            Titchener and Külpe considered their work to be built on the principles and laboratory techniques of the great German psychologist Wilhelm Wundt.  In fact, the term "introspectionist", which was coined by the behaviorists to give a label to what they were rebelling against, is usually considered to be applied primarily to Wundt, with Titchener and Külpe being seen as Wundt's disciples. But Wundt, like Marx and Jesus, had disciples who often violated his principles while speaking in his name.  Wundt himself, being trained in the philosophy of Kant, did not accept the atomistic picture developed by his disciples, and in fact denounced this aspect of their work in print. The atomism of Titchener and Külpe was created by quoting Wundt out of context on those few points where he expressed some agreement with the fashionable postivism of the time, and ignoring or contradicting him on almost everything else. Nevertheless, there were elements of atomism in Wundt's approach that make it difficult to completely untangle his ideas from the more simplistic atomism of his two most influential followers.


            Wundt believed that experience was shot through and through with structure and interrelationships, and that many of those structures were conditioned by what he called affective components. "Affect"  was Wundt's generic term for the emotional aspects of experience, which derived their significance and impact from the volitions and projects of the self as a whole. (see Blumenthal 1975 p. 1085) Wundt called our awareness of this aspect of experience apperception, which like Kant's use of the same term, implied that there was a unified sense of self underlying the individual sensations. (although unlike Kant, Wundt stressed Schopenhauer's point that the unifying force that constituted the self was the will.)


             Wundt did feel that certain individual sensations presented themselves directly to us, but because of the influence of the affective components of the apperceptive self, they rarely (if ever) presented themselves in their true guise. When we notice one idea following another in ordinary experience, we are not, according to Wundt, experiencing the process of two atomic thought elements associating with each other. According to Blumenthal, the term that Wundt used for the ideas that we encounter in ordinary experience is gebilde, which in other contexts is often translated as "structure" "formation" or "creation."   Wundt goes to great lengths to argue that the laws of association described by the British empiricists cannot fully explain how gebilde  come into being and interrelate. (Wundt 1897 pp. 224-228). In fact, Wundt did not believe that experimental laboratory methods would be able to explain these higher level associations at all. He dealt with them in a more descriptive and anthropological manner in his ten volume Volkerpsychologie, which described the effects on consciousness of such cultural factors as myth, religion, law, and art.


              Wundt did believe that gebilden  were made up of what he called elementary associations, and accepted the "necessity of elementary associations as antecedents to all complex combinations" (Wundt 1897 p.202). But he also said that "Psychological elements as isolated persisting entities have no other reality than as products of our conceptual abstraction which for reasons of research we may wish to think of as elemental qualities" (quoted in Blumenthal 1979). Because both Baars (1986 p.31) and Danziger[3] (1979 p. 221 ) admit that Wundt did sometimes use introspection to study simple sensations, I think that Wundt was ambivalent on this point and/or changed his mind later in life. But there is no question that Wundt always believed that most of psychology's subject matter would be ignored if it concerned itself only with analyzing experience into discrete sensations. And unfortunately this was how future historians (especially Titchener's student Boring)  ended up describing Wundt's research goals.  "Complex combinations" in the first quote above is almost certainly a translation of some form of "gebilde" (according to Blumenthal 1975, the most common translation of gebilde in Wundt texts is "compound"). But the fact that this choice of translation stresses the analyzability of gebilden, rather than their uniqueness and wholeness, shows how Wundt's English speaking disciples misinterpreted his work. For them, Wundt's Volkerpsychologie  was not the capstone of his psychology, but a change of subject.


            Both Titchener and Wundt agreed that it was impossible to study the higher level gebilden  in the laboratory. The difference between the two was that Wundt concluded from this that laboratory science must be supplemented by the more humanistic and historical Volkerpsychologie , and Titchener concluded that the higher level functionings could be ignored. In order for Titchener to justify his methods, he had to conclude that the higher mental functions could be analyzed without remainder into the lower level functions. This was why Titchener usually translated "gebilde"  with words like "compound" or "aggregate".  These words implied an analogy with an atomistic interpretation of the concepts of chemistry, an analogy which, according to Blumenthal, Wundt had considered but ultimately rejected. This analogy with chemistry was unacceptable to Wundt for two reasons. First of all, it ignored the emergent nature of conscious experience by seeing gebilde  as mere aggregates or compounds of elementary sensations. And secondly, it implied that experience was the passive byproduct of the interactions of its parts. Wundt believed that "Volition is the paradigm psychological phenomenon"  (quoted in Blumenthal 1979) which meant that the primary organizing principle for consciousness was "purpose, values, anticipations of the future" (ibid). Even the most apparently immediate experience consisted not only of sensations (which Wundt apparently did think of as being something like psychological atoms) but also feelings. Feelings were the fundamental components of emotions, and were always constituted by their relationship to the affective component of experience i.e. to the volitions of the unified self given in apperception.


simple feelings depend on the relations in which each single feeling stands to the whole succession of psychical processes. (Wundt 1897 p.84)


despite the fact that Wundt did believe that in some sense there were such things as simple feelings, he did not mean to imply by this that they could be understood without reference to a subject that felt them.


            Titchener was able to see himself as following in Wundt's footsteps only by de-emphasizing certain aspects of Wundt's thought, and rejecting others outright. Anderson 1975 points out that Titchener doubted that Wundt considered the Volkerpsychologie  to be all that important, and that Titchener only translated volume one of Wundt's Grundzüge der Physiologischen Psychologie  (foundations of physiological psychology), because it was relevant to the task of analyzing experience into its elementary associations. This meant that Titchener's introspectionism was influenced by only one third of the Grundzüge, and also ignored the dozens of other volumes written by Wundt on the more complex aspects of human experience. Titchener also took exception to Wundt's theory of feeling and discarded his concept of apperception. (Anderson 1975 p.383). Titchener eventually concluded that the purposive elements which were the essential source of Wundt's psychic causality actually interfered with scientific research. Titchener thought that the goal of psychology was to study the "existential experience" that was left when 'purpose , value, social utility, in short, meaning, have all been removed." (Danziger 1979 p. 223) What was left, of course, was nothing but naked sense data. He thus turned Wundt's  perspective, which was originally heavily influenced by Kant and Schopenhauer, into something like Humean atomism. The goal of Titchener's laboratory psychology became to construct what Güzeldere 1995 calls "an atomic table of the human mind." (p.38)  Titchener believed that he could discover these elementary psychological atoms with disciplined introspective techniques he saw as being similar to those developed by Wundt. And then, Titchener hoped, we could find laws of aggregation and combination that would explain (without positing emergent properties) how these elementary associations could give rise to more complex psychological properties.


            This seemed like a reasonable goal to those who accepted  the positivism of Mach and Avenarius, which was very much in the air at the time.  Their assumption that the immediately given was describable by observation sentences required an atomism that enabled those sentences to stand on their own feet without any support from higher level theories. And that level of epistemic independence would only be possible if there were somewhere in the flux of our experiences moments that brought us into direct contact with the world. Many of Wundt's students, especially Külpe and Titchener, considered themselves to be positivists. Wundt himself, however, was at best ambivalent towards positivism. He did agree with the positivists that metaphysics and science should be separated, and saw his goal as the "exclusion of all metaphysics from psychology" (Wundt 1897).  But unlike the positivists, he did not think that the observation sentences of the natural sciences were independent of theories.


Natural science seeks to discover the nature of objects without reference to the subject. The knowledge that it produces is therefore mediate or conceptual. . . . This abstraction makes it necessary continually to supplement reality with hypothetical elements. . . .Science makes up for this lack of direct contact with the objective processes, by forming supplementary hypothetical concepts of the objective properties of matter. (ibid.)


            He did, however, think that the subject matter of psychology, unlike that of the other sciences, was immediately given to each of us.


 The knowledge thus gained in psychology is, therefore, immediate and perceptual, -- perceptual in the broad sense of the term in which, not only sense-perceptions, but all concrete reality is distinguished from all that is abstract and conceptual in thought. Psychology can exhibit the interconnection of the contents of experience, as these interconnections are actually presented to the subject, only by avoiding entirely the abstractions and supplementary concepts of natural science (ibid.)


            Here, it seems to me, Wundt is trying to have it both ways. He acknowledges that the other sciences must relate their observations to things beyond immediate experience, and thus their knowledge is necessarily abstract. He also acknowledges that psychology does not study individual atoms of experiences, but the interconnections between experiences as well. And yet because the interconnections studied in psychology are between other experiences, he wants to claim that these connections are not abstract in the way that the concepts that connect entities in the other sciences are abstract. This idea that connections between experiences are every bit as present to us as the experiences themselves may have been an inspiration for James' radical empiricism, and Dewey's theory of experience. But Dewey, at least, was aware that this kind of unified experience was changed by the act of inquiring into it, and that inquiry required abstract theories which separated us from this immediacy. He did not think, as Wundt appears to have thought, that the structures in our experience could be both expressible in theories and immediate.


            In fact, there are also passages in Wundt's writings that indicate he sometimes thought that it was impossible for the elementary sensations to ever be directly given in experience. Psychological atoms were, in some strange way, as abstract and theoretical as the atoms of physics.


. . .psychical elements, or the absolutely simple and irreducible components of psychical phenomena, can not be found by analysis alone, but only with the aid of abstraction. This abstraction is rendered possible by the fact that the elements are in reality united in different ways. If the element a is connected in one case with the elements b, c, d . . ., in another with b', c', d' . . ., it is possible to abstract it from all the other elements. because [sic] none of them is always united with it.


 If this were the case, in what sense could psychology be seen as dealing with immediate, rather than abstract knowledge? This equivocation between the experienced and the abstract was ultimately the same inconsistency that destroyed positivism, despite Wundt's valiant attempt to separate himself from it.


            Titchener, in contrast, was apparently so under the spell of positivism that he could specifically state positivism's central inconsistency without noticing that there was any problem.[4]  He claimed that the atoms of subjective experience were the same atoms described in the observation sentences of physics, only under a different context.[5] Boring 1950 describes Titchener's position thusly.

Both psychology and physics work immediately with experience, but they regard it in different ways; physics takes the "point of view" of experience "regarded as independent of the experiencing individual", psychology the "point of view" of experience "regarded as dependent upon the experiencing individual". (p. 417)


To claim, as Wundt did, that one science yields direct knowledge while all others rely on abstractions is at worst a unjustified form of special pleading. But it is (from a modern perspective, at least) self-evidently contradictory to claim, as Titchener did, that experience can be both immediate and mediated by two different points of view.


            Nevertheless, if one ignores the inherent contradictions of atomism, (as so many people of that time did), it is plausible to assume that if elementary sensations existed, the techniques of the introspectionists would  discover them. After all, musicians learn how to hear elements in sound sensations that are unnoticeable to people who do not have their specialized training. Why shouldn't psychologists with special introspective training eventually be able to recognize the fundamental elements of all sensations? But alas, it was not to be. Titchener's laboratory in Cornell discovered 44,435 "fundamental" sensations and Külpe's laboratory in Leipzig discovered only around 12,000, with very little overlap between the two. (quoted in Güzeldere 1995 p. 39 from Boring 1942). The idea of 44,000 fundamental anythings is clearly a contradiction in terms, and the fact that two different laboratories got such different results made it impossible to take either set of data seriously. Something was clearly wrong here, but what? 


            Part of the problem was the apparent impossibility of mediating between two conflicting introspective reports. (see Güzeldere 1995 and Baars 1986 pp.32-33). I think, however, that the failure of Titchener's and Külpe's laboratories to find any elemental sensations can also be interpreted positively. The fact that no one was able to find these elemental sensations after so much hard and careful work is, I believe, pretty good evidence that there aren't any. The results they got were just the sort of results one would expect if perceptual experience was a process all the way down.[6] If elementary associations had been directly accessible to immediate experience, those associations  would probably have revealed themselves to the methods of the introspectionist school. But if consciousness is fundamentally a continuous process without discrete parts, in which each moment is shaped by the context in which it occurs, then any attempt to study it by performing distinct acts of observation will almost certainly create an artifactual "atom" with each observation. It's rather like trying to slice up water with a knife. You will succeed in dividing up the water if you run a knife through it, but it will flow back together again after you've stopped cutting. And the next time you cut it, it will flow differently.


            A more effective metaphor would be a liquid that flows more slowly and thus gives an illusion of concreteness. Let us suppose we are trying to find the fundamental components of a sauce pan full of oatmeal by throwing it against the wall. If we assumed that throwing was an experimental process guaranteed to reveal the true constituents of the oatmeal, we would measure each of the shapes that cohered against the wall, and then classify them. The next time we threw the oatmeal against the wall, we would get a completely different set of shapes, which, if we maintained our faith in this process, we would again measure and classify. If we became suspicious of the large number of different shapes we were getting, with no repetition of patterns, we might train ourselves to throw the oatmeal against the wall in exactly the same way every time, which would probably decrease the number of possible shapes down to a few thousand. We would also be greatly annoyed when we discovered that someone else was throwing oatmeal against a wall using a different wrist action, and getting a completely different set of fundamental components.


            This is a rather cruel caricature of what happened to the introspectionist psychologists, but it does give some sense of why the introspectionist laboratories could sometimes get consistent results "in house" which nevertheless conflicted with the results of other laboratories. In no sense, however, should my willingness to ridicule be taken as a lack of respect. Without the evidence provided by the introspectionist's heroic failures, the claim that experience is fundamentally a process would only be a philosopher's speculation. In fact, when William James claimed that experience was an absolute process in his Essays in Radical Empiricism, it was dismissed by most psychologists as a philosopher's speculation. The only way to determine the relative merits of atomistic versus process theories of sensation was to do what the introspectionists did: Choose an alternative and spend thousands of hours doing experiments to test it. We must not forget that if we consider falsifiability to be a scientific virtue, it should be to a scientist's credit when his theory is decisively falsified.


            There were two different and opposite reactions to the failure of atomistic introspectionism.


             1) Many European psychologists decided to keep introspection and reject atomism. They started a movement called gestalt psychology which, in the words of one of its founders, "objected to this premise, the thesis that the psychologist's thinking must begin with a consideration of such elements. . . it may be tempting to assume that all perceptual situations consist of independent, very small components. . . {but} Are we allowed to impose on perception an extreme simplicity which, objectively, it may not possess?" (Köhler 1959   p.727).  Note, however, that even in 1959, when gestalt psychology had been around for decades, that Köhler was only willing to be agnostic about elemental sensations, and could not quite bring himself to deny their existence altogether. The most the Gestalt psychologists were willing to claim was that there are lots of important perceptual properties that elementary sensations could not explain. Figure-ground distinctions, for example, were reversible depending upon expectations and context, and they seemed to be essential to almost all of the experiences of discreteness that we encounter in normal perception. But in spite of these and many other important discoveries, the Gestalt psychologists apparently did not feel they could dispense with the concept of elementary sensation and claim that all sensations were emergent gestalts, even though they felt most of the interesting ones were. Their fundamental position was thus not that different from Wundt's, although the widespread misinterpretation of Wundt by Titchener and Boring made it difficult to see this. (Green 1922 helps to perpetuate this myth when he says that "The aim of {Gestalt Psychologist Max Wertheimer's} criticism was the atomistic psychologys of Wundt. . . and other European psychologists of the time." ) Both Wundt and the gestalt psychologists believed that most of the experiential qualities that are most obviously present to us are emergent and cannot be reduced to their parts. The only difference between Wundt and the gestalt psychologists on this issue was that the latter wanted to study emergent properties in the laboratory, and Wundt studied many of the most important emergent properties outside of the laboratory.


            2)American psychology, in contrast to gestalt psychology,  concluded that the cause of the problem was the use of introspection, which was one reason why American behaviorism demanded that psychology should no longer concern itself with mental states at all. The behaviorists did not reject the atomist ideal of science, however. As Danziger 1979 puts it:

A later generation was to attribute the weakness of Titchener's foundations to his reliance on introspective experience. . .  The behaviorist revolt did not question Titchener's positivist ideals. On the contrary, it adopted them with enthusiasm; it simply considered that the wrong path had been chosed  to reach them. . .the switch from Titchenerian introspectionism to behaviorism must be characterized, not as a revolution, but as a reformation within the broader movement of positivist psychology. (p.219)


The behaviorist's devotion to this ideal, however, eventually caused them to experience a crisis of their own.


The Reflex as the Unit of Behaviorist Atomism


            That Titchener considered an atomistic analysis to be the only possible way of scientifically studying anything is evident from the following quotation


James' chapter on The Stream of Thought, which has already become a classic upon the anti-atomistic side, posits the fact that "thinking of some sort goes on." Its author then proceeds: "How does it go on? We notice immediately five important characters in the process." And this discrimination of characters is  obviously an instance of that mode of analysis which James later terms 'the process of abstraction.

There is no doubt, then, that a descriptive psychology must be analytical. (Titchener 1912 p. 496]


            Thus, for Titchener,  to say anything at all about a subject was to analyze it, and to analyze something was to break it down into fundamental atoms. This conflation was also accepted by the behaviorists, but because the behaviorists blamed all of the failures of introspective psychology on its acceptance of subjective experience, they could no longer base their analysis on elementary sensations or sense data. At the same time, they wanted to have an atom which was unique to psychology, which could provide a  foundation for a science that could be distinct from physics. They thought they had found such a foundation with Pavlov's concept of the reflex..[7]


            Pavlov originally thought of his concept as being genuinely neurological, but the American behaviorists rightly concluded that his references to neurology were purely speculative. They decided, under the influence of the logical positivists, that one could directly observe behavior within the context of an operationalist science of behavior, and each observation revealed a single reflex. When these reflexes were chained together, they could supposedly account for all of the behavior of organisms. The problems that eventually led to behaviorism's demise are too numerous to fully discuss here. But I maintain that at least some of them were caused by the same ontological commitment to atomism that destroyed Titchener's introspectionism. In the next sections of this paper, I will discuss two areas of behaviorist psychology where atomism caused problems, and the solutions that have been proposed to deal with those problems. One of these solutions, often called functionalism, is now the dominant ontology in cognitive science but suffers from a commitment to something like the atomism that plagued both introspectionism and behaviorism. The other solution, called either contextualism or ecological psychology, is, I believe, the best available alternative to atomism, but it requires us to recognize that psychology may be much harder to do than we feared. Many psychologists may find this conclusion disturbing, but if it is the closest thing we have to the truth, the only scientific choice is to accept it.


The Fall of Verbal Behaviorism


            In order for atomism to maintain any semblance of coherence, it has to downplay the significance of any properties that appear to emerge when the atoms are combined in any way. The most effective way of doing this is to give the atoms intrinsic properties that make it possible to predict what will happen when any possible set of atoms is combined. The physical science that is most atomistic is chemistry. It often operates at a level where it is possible to see the chemical elements as different components in a kind of submicroscopic lego set.  Because atom "A" has a certain number of electrons in its outer shell, and atom "B" has the right number of holes, they will combine to form the molecule "AB". Given the premises of chemistry, this follows with as much necessity as any conceptual truth. (Whether the premises of chemistry are themselves true is an empirical question, which is why chemistry is an empirical science.) Like Lego sets, chemical elements can be combined to produced properties that are not possessed by any of the individual elements. H2O is wet, even though neither hydrogen nor oxygen is wet, , and Lego pieces can be put together to resemble choo-choo trains, even though no individual piece looks like a choo-choo train. Nevertheless, classical reductionism claims that the wetness of the water, and the "choo-choo train"- hood of the lego ensemble, are nothing but the sum of those particular parts. I will probably win more enemies than friends for questioning that assumption, so I won't. But fortunately for this paper I don't have to, because behaviorism had a much tougher form of atomism, which made it impossible for it to grant the mind even as much structure as a Lego choo-choo train.


            The reason it is possible to make things like choo-choo trains with a Lego set is that there are more than two connectors on each Lego piece. This makes it possible to have structures with a variety of shapes in three dimensions (like choo-choo trains). The behaviorist reflex, however, has only two connectors: an input for the stimulus and an output for the response.  This makes possible only a one dimensional structure, of which there are only two possible forms: the chain and the loop (which is only a chain that connects back to itself.) This made it impossible to have even the kind of innocently emergent structures that can be made from Lego sets. This, however, was exactly the way the behaviorists wanted it. Any attempt to say anything about the mind itself was considered to be either a Cartesian superstition, or an unjustified speculation about neurology. As long as they were only talking about individual reflexes, they felt that all of their knowledge was based on direct observations of what subject X did at time T, and no references to mental properties would be necessary.  The behaviorists would admit, under duress, that neurological structures of some sort had to be responsible for the behavior of complex organisms. But they felt that the behavior itself could be analyzed into discrete atomistic stimulus-response connections, and that one could remain agnostic about what sort of neurology was responsible for those connections.


            There is a grain of truth to this which was expressed quite well by behaviorist Howard Rachlin in his interview in Baars 1986.


Take this analogy: You're Mario Andretti, the greatest racing driver in the world. To what extent does Mario Andretti know how a car is built? It may be useful to some extent, but there are many people who know much more than he does. But he knows how to drive the car. That is a different kind of knowledge. In psychology. . .we are treating each other as cars and predicting and controlling each other's behavior. To what extent does the knowledge of how the nervous system is built help you? Maybe a little. . . Somebody who knows about the engines of racing cars may be able to tell Mario Andretti not to drive the car that day because there is something wrong with it. But that is a different level of knowledge than Mario's expertise, which is how hard you push the accelerator on a certain curved road, when to turn the wheel, and when not to turn the wheel. People at Mario's level I consider to be behaviorists. The engine people are doing physiological psychology. (p.95)


If we follow out this analogy's implications, we can see both what is wrong and right about the behaviorist view of psychological properties. It is true that knowing how a system functions is a different kind of knowledge from knowing its physiology, although physiology is frequently useful in understanding function. But any attempt to codify Mario Andretti's knowledge in the atomistic terms demanded by behaviorism simply wouldn't work. It might be that Mario's knowledge is not stored in any sort of theoretical structure at all, but actually in a series of vector transformations performed by connectionist nets. (a possibility that could not even be considered without the benefit of concepts derived from physiology). But behaviorists cannot claim that their knowledge is this kind of unverbalizable "knowing-how" behavior. After all, they do write lab reports, and people read them, so the knowledge they produce is verbalizable knowledge. And an atomistic laundry list of stimulus-response connections could not explain all of the different reasons and ways Mario would turn the wheel or step on the accelerator with the precision of a cognitive science explanation that would posit an interacting system of functional modules. Wheel, accelerator, brake,  visual input from the road etc. would all have to interact with each other in ways that, as far as we know, can only be verbally expressed with some kind of decision tree. A simple list of behaviors could not make any sense of the factors that lead to his decision, and would probably be too long to be stored in the human brain even if it were possible to construct such a list. Of course, a lot of important things were learned about cognitive abilities using behaviorist methods. But this may merely illustrate the point I made earlier that scientists often adapt to bad ontologies by being inconsistent. Rachlin himself, a few pages after describing the analogy quoted above, admits that he uses the concept of value in his own research, which is not a directly observable stimulus-response connection. But he claims "It' s probably a mistake. . . I would assume that if my theory were perfectly developed, you could get rid of value." (Baars 1986 p.98). Most other psychologists, however, eventually lost faith in the usefulness of squeezing their results into the atomistic stimulus-response mold.


            Chomsky's famous review of Skinner's Verbal Behavior was probably the single most important document in discrediting behaviorism, and one of its most recurrent themes was the inability of individual S-R connections to account for the intricacies of language. Language learning could not be seen as learning how to respond to particular stimuli with particular responses, because there are many different appropriate ways to respond to any given stimulus, and many different stimuli that can produce the same response. The only way to account for this disparity is to posit some kind of structure more complicated than the stimulus-response chain, which could weigh and compare a huge variety of factors and make appropriate decisions based on those comparisons. According to Chomsky, this structure would  have to be capable not just of selecting the appropriate behavior from a pre-set repertoire, but also be able to  generate completely novel behavior. (hence the term "generative grammar"). Linguistics prior to Chomsky had tried to reduce language to the chain-like structure called Finite State Markov sources. (See Baars 1986 pp. 342-344) Once Chomsky had demonstrated that these chain-like structures were inadequate to explain language, it became respectable for a variety of psychological researchers to admit that their data was also too complex to be comprehensible as a chain of ontologically independent S-R connections. And this was what a made possible a shift from atomistic to functionalist explanations.


The Differences between Atomism and Functionalism


             Behaviorism and functionalism agree upon one crucial point: There are many things about the mind that cannot be learned by dissecting brains. Functionalist psychology studies behavior, not neurophysiology, and recognizes that its theories are to some degree distinct from neurophysiology (at least in the short run). But unlike behaviorism, functionalism admits that it posits its fundamental explanatory entities. Behaviorism's operationalism required it to claim that it only observed what it studied, and never posited anything. Even though behaviorism's anti-mentalism forbid it to talk about things like sense data, the operationalist claim that it is possible to observe without theorizing only makes sense if we accept what Sellars called the myth of the given. In order for a single entity, such as an S-R connection, to be immediately given to the observations of the operationalist, it must be ontologically independent of whatever might be observed later.


            When the functionalist divides the world up into comprehensible parts, however, these parts are not ontologically independent. Each of the parts is what it is because of its relationship to the other parts in some sort of systematic structure. Each part must be distinct from the other parts, for it is only by being distinct from the rest of the system that it can relate to it. But without those relations, the part itself could not exist. The players in the game of baseball, for example, are ontologically constituted by both their relationship to and their distinctness from the players. Unless there were distinct differences between what the pitcher does, what the batter does, and what the third baseman does, there would be no game of baseball, only an unseemly free-for-all. But that does not mean that before there was ever a game of baseball, there were batters, pitchers and third basemen sitting out in the universe,  waiting for some bright fellow to get the idea of putting them together into a game.[8]


            Consequently, Titchener was wrong when he claimed that analyses cannot be performed without accepting atomism. It is only the myth of immediately given experience that makes atomism seem even possible. Without it, whatever analysis we perform binds the analysanda more tightly together, for every separation clarifies their relationship to each other.  The functional analyses of cognitive psychology acknowledge that their goal is to unify the data they discover, not simply observe it, and that this unity can only be achieved by making inferences that go beyond the data that inspired them. In some sense, the whole must be greater than, or possess qualities not attributable to, the mere list of its parts. The dangerously misleading assumption that "real scientists don't theorize, they just look at the facts", has been abandoned, because a study of the history of science has revealed that even physicist's observations are always shaped by theories. Cognitive psychologists have thus ironically become more like the physical sciences than behaviorism or introspectionism ever were, and it has accomplished this by abandoning the atomism that positivism erroneously attributed to the physical sciences.


             Cognitive psychology is consequently a radical break from the atomism of both the introspectionists and the behaviorists. This means that it has now freed itself from a flaw in past psychological methods that has impeded psychology's growth for almost a century. A radical break from old methods that don't work is always grounds for optimism, at least initially. But no paradigm can last forever without being threatened by a revolution, and cognitive science may already  be showing signs of tensions that point towards its successor. In the next section, I am going to discuss certain inconsistencies in the ontology of functionalist cognitive psychology that may cause it problems in the future. I will then cite some texts that give some more reasons to believe that those problems will actually manifest.


Atoms, Systems, and Environments


            The problem I have in mind is this: The critiques of ontological independence that make functionalism preferable to atomism can be turned back, with only slightly weakened power, onto functionalism itself. If a single individual cannot stand on its own, but must get its ontological status from the function it performs in the context of a system, why should we assume that a system can stand alone without reference to the environment within which it performs a functional role? Isn't the relationship between a system and its environment exactly the same as that between any other part and the whole of which it is a part? Isn't the difference between seeing the forest and seeing the trees just a matter of how far you step back before you take a look? Isn't every tree an environment to the bugs that crawl across it, and every forest just a small part of an even larger ecosystem? So why should we assume that we can study a single functional system in isolation? Shouldn't such an attempt be at least somewhat vulnerable to the problems that arose when the introspectionists tried to isolate elementary sensations and the behaviorists tried to isolate S-R connections?


            Chomsky's project of isolating syntax from semantics assumes that language must be comprehensible as a closed system, separable in principle from the world that the language talks about. Fodor's classic essay "Methodological solipsism as a research strategy in Psychology" argues at great length that psychology must assume that organisms are comprehensible as completely isolated systems in order for psychology to be a science. Even if Fodor's arguments are sound, however, they don't prove that it actually is possible for psychology to be a science. Perhaps psychology, by the very nature of its subject matter, may not be capable of being comprehensible in the same way as the subjects of the physical sciences. Some may find that depressing.  But if we were able to understand exactly why psychology is so much more difficult than physics, it would free psychologists from the need to try to do what can't be done, and to concentrate on what can.


            The following are brief commentaries on  some research and theorizing which question the assumption that a functional system can be studied in isolation from its context. 1) J.J. Jenkins' description of how he rejected behaviorist atomism as a viable theory of memory, and why the alternative he proposed differs from Chomsky's and Fodor's 2) Ulrich Neisser's ecological critique of cognitive psychology 3) George Lakoff's critique of Chomsky's separation of syntax and semantics.


Jenkins Forgets the Old Theory of Memory


            J.J. Jenkins began his research on memory with a paradigm derived from the work of Clark Hull, which was both significantly different from, and significantly similar to, Skinner's. (See Skinner 1944 for a description of some of the differences). Like Skinner, Hull believed that the S-R connection was the fundamental atom of behavior, and that the chain was the only acceptable form of structure that could be made with those atoms. But unlike Skinner, Hull was willing to posit the existence of unseen theoretical S-R connections to account for more complicated mental processes. Because Hull had the courage to commit himself to falsifiable theoretical posits, his theories were usually eventually falsified. This was one of the main reasons that the allegedly theory-free operationalism advocated by Skinner became so widely accepted. Jenkins, however, did research which seem to confirm the existence of inner S-R connections, and this research became the basis of what was called the mediation theory of memory.


            Jenkins describes the principle of mediation thusly "If element A is associated with element B, and element  C is associated with element B, then element A will acquire some association with element C" (Jenkins 1963 p. 213.). Jenkins tested this principle by teaching laboratory subjects a list that paired words with nonsense syllables (such as zug/table). He then taught them another list that paired each nonsense syllable with another word that was commonly associated with the first word. (such as zug/chair).  He found that his subjects learned the second list much faster than they learned the first. This seemed to imply that an associative chain had been established linking "zug" to "chair" by means of the mediating association of "table" to both "zug" and "chair". Further research, however, was not as encouraging. For one thing, the attempt to link four items together broke down completely, yielding only chance results. This meant that even if Markov chains had been structurally sophisticated enough to account for memory functions, it would still have been impossible for  human minds to form the necessary associations to hold those chains together.  Also, it was discovered in research that " If a stimulus elicits two responses, the responses will acquire a tendency to elicit each other" (Jenkins and Palermo 1964 p.147)  This is not terribly surprising to anyone who has ever introspected, but it required a major change in the associationist ontology. Connecting A to B, then B to C, creates a chain, which is supposed to be the only possible structure for the associationists. But if connecting both  B and C to A creates a connection directly between B and C that is not mediated by A, we now have the possibility of creating a variety of structures. We have, in another words, a lego set with more than two connectors on it, and thus we can create things that have the complexity of choo-choo trains and not the simplicity of a Markov series.


            Jenkins could have seen this as an opportunity to formulate theories of memory along Chomskian cognitive lines, which would have posited functional systems divided into flow charts and block diagrams. In fact, Jenkins acknowledges in Baars 1986 that he was strongly influenced by Chomsky, and Jenkins and Palermo 1964 begins with a quote from Chomsky. But Jenkins eventually took a very different approach in his provocatively titled 1974 essay "Remember that old theory of memory? Well, forget it." He begins by explicitly rejecting the behaviorist ontology of fundamental units that can be connected only by chains. But he also claimed that his own research implied that " Memory is not a box in a flow diagram. . . it seems to demand an understanding of all of the higher mental processes at once" (p.794). One couldn't divide memory off from the rest of  mental processes, because every mental processes contributes to our ability to remember things in some way or another. Even more difficult for a functionalist analysis is that remembered events do not appear to remain distinct once they are "stored". "Once the fusion of strands into events has occurred. . . the subject cannot perform an analysis to recover the exact pattern of input" (p.790).


            Jenkins calls his position contextualism, and says " it has its roots in William James, C.S. Peirce, and John Dewey." (p. 786). (Remember that Titchener considered James the foremost defender of anti-atomistic psychology.) Like William James, Jenkins claimed that experience was fundamentally a process made up of events, and that "the quality of the event is the resultant of the interaction of the experiencer and the world". (ibid.). Jenkins also recognized that once we admitted that we cannot study memory processes independently from their context, "This means that being a psychologist is going to be much more difficult than we used to think it would be." (p.787).


            If we take this position as seriously for other mental processes as well as for memory, then we will see mentality as consisting of context-dependent processes, rather than self-contained systems with stable interrelated functions.  This compromises many of the goals of cognitive psychology, such as Chomsky's claim that syntax is a self-contained system that can be separated from semantics and phonology. And yet in Baars 1986, Jenkins doesn't see himself as being in conflict with Chomsky, despite these differences. I think this partly because Jenkins' greatest contributions have been the experiments he has developed, and he is therefore by temperament not as interested in fine-honed ontological arguments. His contextualism has served him well by keeping him from getting stuck in what Baars calls "Methodolatry".  But by his own admission, there is something a bit unsettling about rubbing one's nose in the sense of uncertainty implied by contextualism, and he doesn't appear eager to dwell on the subject . To quote his interview in Baars 1986: " You just keep saying 'well, everything depends', and people get a little tired of hearing that" (p.252)


Ulric Neisser and Ecological Psychology


            Ulric Neisser wrote one of the first books to exploit the computer metaphor in psychology, and its title, Cognitive Psychology,  probably helped to name the movement. Several years later, however, his book Cognition and Reality questioned many of the presuppositions of cognitive psychology, for many of the same reasons as Jenkins 1974. By Nesser's own admission. "the message I brought in Cognition and Reality was not as popular as the one I brought in {Cognitive Psychology}. . . now, I'm saying that what people want to do may not be worth doing; maybe they should be doing something else. That's not such a popular message. " (Baars 1986 p. 282).


            Like Jenkins, Nesser claims that it is impossible to study a system outside of its context.  The data in Jenkins 1974 mainly demonstrates that one mental function (i.e. memory) cannot be studied independently from the rest of the mind.  Nesser 1976 similarly claims (especially in Chapter 5) that there is no reason to assume that there is a single faculty of the mind that is responsible for what we call attention. We have a variety of different skills, and most of them require us to interact with the environment. But there is no reason to assume that all of our skills must get their information from a single centralized "attention  module", or that the different ways that each skill relates to the world must all share certain principles in common. Nesser also criticizes the idea that perception can be studied separately from the embodied activity that it guides.

the motionless observer with a fixed head who serves as a subject in so many perception experiments is in an unusual and remarkably unfavorable situation. The motion producing information that he lacks is crucial for normal vision (Nesser 1976 p.109)


            Nesser then steps back a bit and emphasizes another closely related point that Jenkins only touches on:  Even the entire mental system cannot be understood outside of its environment. He claims this is one of the main things wrong with cognitive science experiments. They create an artificial environment in the laboratory, and then extrapolate, without any real justification, that the laboratory behavior is somehow more fundamental than the behavior performed in a natural environment. Although this assumption is considered to be essential to both the possibility and the value of scientific experiments, it is far from certain.



            Lakoff's Cognitive Linguistics: Recombining

Syntax and Semantics


            Lakoff 1987 specifically criticizes Chomskian Linguistics[9] for many of the assumptions discussed above, and proposes an alternative similar to the ones advocated by Jenkins and Neisser. Lakoff points out, for example, that the following principles are essential to what he calls objectivist semantics.


A) every concept is either a primitive or built up out of primitives by fully productive principles of semantic composition.


B) All internal conceptual structure is the result of the application of fully productive principles of semantic composition.


C) The concepts with no internal structure are directly meaningful, and only those are.


(Lakoff 1987 p.279)


A) is an explicit statement of what we have been calling atomism, and which Lakoff also calls atomism in other parts of the book. B) is a way of saying that the rules  of semantic composition are a self-contained system i.e. they are governed by "fully productive principles" that would get the same results even if they were run automatically on a computer. C) Makes an intriguing claims about the presuppositions of a Chomskian system that perhaps Chomsky himself was not aware of: that his rule governed theory of mentality must rest on a foundation of directly evident atomistic particles, even though it cannot consist of nothing but such particles.


            Lakoff spent many years trying to discover linguistic principles that both met these criteria and explained the way real language actually works. He eventually concluded that there was a great deal of linguistic behavior that could only be explained by relaxing these criteria in certain key respects. Just as Jenkins claimed that memory, and Nesser claimed that perception and attention, were not self-contained faculties, so Lakoff claimed that syntax was not a self-contained faculty. It cannot be separated from the semantic factors that hook language up to the world. Lakoff states this most explicitly on page 466, where he lists two sides of ten issues, saying that Chomsky's generative approach assumes the first alternative, and that his newer approach provides empirical evidence for the second alternative. Here are the dichotomies in this list which relate to the issues we are discussing.


            7) must all syntactic constraints be accounted for only by rules of syntax that are oblivious to meaning, or can a great many syntactic constraints be accounted for on semantic grounds?


                  8) Is the meaning of every grammatical construction computable from its parts, or are there constructions whose meaning is motivated by the meanings of its parts but is not computable from them by general rules?


                  9)Is there a strict dichotomy between grammar and lexicon, or is there a continuum between the two?


                  10) is grammar a separate "module", independent of other aspects of cognition, or does it make use of other aspects of cognition such as prototype categorization, cognitive models, and mental spaces. (Lakoff 1987 p.466)


            Lakoff calls a linguistics that accepts the second of each of these pairs of alternatives cognitive linguistics. Given that the first of these alternatives is usually accepted by what Nesser called cognitive psychology, this label is a bit confusing. (I keep wanting to call Lakoff's work post-cognitive linguistics).  The name makes sense within the context of Lakoff's work, however. For one thing, it refers to Lakoff's claim that syntax cannot function independently of functions that tell us about the world (i.e. cognitive functions). For another, Lakoff claims that cognition is not purely a mental process in the traditional sense, but rather rests on a fundamental base of embodied experience within the world. Even when we are dealing with abstractions, we often make sense out of them by referring to our existence as living bodies that interact with a world and move through space.  Lakoff gives so many examples of so-called abstract thought that would make no sense to a disembodied computer program (one small example: the fact that we say "things are looking up" or "the stocks are down"). that he makes a compelling case that we could not think the way we do except for the fact that we have bodies that we know how to move through space and interact with objects. Thus Lakoff, like Jenkins and Nesser, claims not only that mental processes cannot be understood in isolation from each other, but also that they cannot be understood in isolation from their environment.




            The history of psychology reveals that there is real Jamesian "cash value" to the ontological arguments against both atomism and the more subtle atomism that assumes it is possible to study a system in isolation from its context. The successive failures of the introspectionists and the behaviorists show that an atomism based on the myth of immediately given observables ultimately leads to scientific crisis, regardless of whether it is explicit (as it was for the introspectionists) or implicit (as it was for the operationalism of the behaviorists). There now seem to be similar crises starting to brew with  what Fodor calls "Methodological solipsism" in cognitive science, which assumes that one can safely study a closed system and ignore its context. But admittedly, it is not easy to imagine what the alternative would be. If attention, perception, language, and memory are all actually parts of other faculties, what does psychology do if there are no faculties which can be studied separately from each other?  And if the mind cannot be separated from its environment, how can we study the mind at all? This is similar to the problem Fodor raised in the Modularity of Mind, when he said that cognitive science would be impossible if all cognitive processes were what he called "isotropic", (i.e. if every fact we know is in principle relevant to our understanding them) and "Quinean" (if our theories  must refer to emergent characteristics of our entire system of knowledge.)


            The Dynamic Systems Theory (DST) approach to cognitive science avoids many of these problems by treating processes, rather than particles, as fundamental, and by its nature continually reminds us that there really is no such thing as a closed system. (Although some researchers are better at remembering this than others). When Dewey critiqued the concept of the reflex arc in 1896, he offered an alternative style of explanation that sounds today like a prophecy of DST.


 In the physical process, as physical, there is nothing which can be set off as stimulus, nothing which reacts, nothing which is response. There is just a change in the system of tensions (Dewey 1896 p.365)  


Because DST  is a branch of physics, and can call on some sophisticated math to measure the changes in a system of tensions, it is being taken much more seriously now than was Dewey's evocative proposal a century ago. But as Andy Clark points out, DST is limited by the fact that it cannot analyze its subject matter into parts, even though it can  describe a whole system with relative precision (insofar as it is possible to isolate that system.) There will always be a place for analysis, of course. But because methods like DST acknowledge the existence of functionally distinct (and therefore in some sense emergent) properties of wholes, we will need to study both parts and wholes from now on, rather than assuming that the study of either one can ever tell the whole story.


            It is hard not to be tempted by the widely held belief that analysis can in principle deliver certainty once it reduces its subject to either fundamental atomistic particles or the enduring functions of a closed system.  But just because this kind of analysis is possible in principle does not mean that it is possible in fact. Certainty has never been more than an unrealized dream for psychology, and if we have good reason to believe that certainty is not obtainable in psychology, this is an important discovery. And even if the abandonment of the quest for this kind of certainty is cause for disappointment, it need not be cause for despair.  Psychologists do have a scientific obligation to find out exactly how much certainty is possible in their subject matter, and to come as close to that level of certainty as their skills and budgets permit. But no one has a scientific obligation to give their subject a level of certainty it is incapable of having.




Anderson, R.J. (1975) " The Untranslated Content of Wundt's Grundzüge der Physiologischen Psychologie " Journal of the History of the Behavioral Sciences  11 pp. 381-386


Baars, B. (1986) The Cognitive Revolution in Psychology  W. H. Freeman San Francisco


Blumenthal, A.L. (1975) "A reappraisal of Wilhelm Wundt" American Psychologist 30 1081-1088


_____________(1979) "Wilhelm Wundt, the founding father we never knew" Contemporary Psychology,  1979, Vol 24 #7 547-550


Boring, E. (1942) Sensation and Perception in the History of Psychology

D.  Appleton Century Company New York.


Boring, E. (1950) History of Experimental Psychology  Appleton Century Crofts

 New York.


Boring, E. (1953) "A History of  Introspection"  Psychological Bulletin  Vol 50 no.3


Boring, E. (1963) History, Psychology and Science  John Wiley and Sons New York and London.


Buss, A.R. (1978) "The Structure of Psychological Revolutions" Journal of Behavioral Sciences 14.


Chomsky, N. (1959) "Review of Skinner's 'Verbal Behavior'" Language 35 26-58.


Danziger, K. (1979) "The Positivist Repudiation of Wundt" Journal of the History of the Behavioral Sciences 15.


Dewey, J. (1896) The reflex arc concept in psychology. Psychological Review 3.


Fodor, J. A. (1965) Could "Meaning be an Rm  "  Journal of  Verbal Learning and Verbal Behavior 4 .


Green, C.,  Introduction to Koffka (1922)


Güzeldere, G. (1995) "Consciousness: What is is, how to Study it, What to Learn from its History"   Journal of Consciousness Studies  Volume 2, Issue 1


Jenkins, J.J. (1963) "Mediated Associations: Paradigms and Situations" in C.N. Coffer and B.S. Musgrave (eds.) Verbal behavior and Learning: Problems and Processes New York McGraw Hill.


__________(1965) 'Comments of Pseudomediation' Psychonomic Sciences Vol 2 pp. 97-100


__________(1974) "Remember that Old Theory of Memory? Well, Forget it!" American Psychologist 29(11).


Jenkins, J.J. and Palermo, D. (1964 "Mediation processes and the Acquistion of Linguistic Structure" in Bellugi, E. and Brown, R. (Eds.) The Acquistion of Language  Antioch Press, Yellow Springs, Ohio.


Koffka, Kurt (1922). "Perception: An introduction to the Gestalt-theorie". Psychological Bulletin, 19, 531-585.


Köhler, Wolfgang. (1959). "Gestalt psychology today". American Psychologist, 14, 727-734.


Lakoff, G. (1987) Women, Fire and Dangerous Things  University of Chicago Press, Chicago 


MacCorquodale, K. (1970) On Chomsky's Review of Skinner's Verbal Behavior Journal of Experimental Analysis of Behavior 13


Mc Kenszie, B.D.(1972) Behaviorism and Positivism Journal of the History of the Behavioral Sciences. 8.


Miller, G.A. (1956) "The Magical Number seven, plus or minus two: some limits on our capacity for processing information." Psychological Review 63.


Neisser, U. (1976) Cognition and Reality   W. H. Freeman, San Francisco.


Skinner, B.F. (1944) "Review of Hull's 'Principles of Behavior'" American Journal of Psychology 57.


Skinner, B. F. (1935). "Two types of conditioned reflex and a pseudo type". Journal of General Psychology, 12, 66-77.


Skinner, B. F. (1937). Two types of conditioned reflex: A reply to Konorski and Miller. Journal of General Psychology, 16, 272-279


Titchener, E. B. (1898). "The postulates of a structural psychology." Philosophical Review, 7, 449-465.


Titchener, E.B. (1912). "The schema of introspection". American Journal of Psychology, 23, 485-508.


Wundt, Wilhelm Max. (1897). Outlines of psychology  (Charles Hubbard Judd, Trans.)


Wundt, W. (1907) Lerctures on Human and Animal Psychology   The Macmillan Co. New York




[1]this quote from the Wittgenstein's "Tractatus" probably says it best:

1.2 The world divides into facts

1.21 Anyone can be the case or not be the case, and everything else remains the same

[2]   a priori   does not always mean the sense in which Kant used the term i.e. necessary and universal for all time. In order for our knowledge to be certain, we must be able to rely unconditionally on either observations or universal a priori principles. But Charles Peirce spoke of an a priori  which could change over time, even though it provided a structure that could condition individual moments of experience. This is similar to a common interpretation of Thomas Kuhn, who sometimes appears to be saying that paradigms provide an a priori  structure which changes only when there is a 'paradigm shift". This concept is an alternative to the extremes of both Kant (who claimed that a priori knowledge never changes) and the British empiricists (who claimed there was no a priori knowledge at all). However, both Kuhn and pragmatists like Peirce acknowledge that this kind of a priori  requires us to admit that all of our knowledge is in principle fallible.


[3] Danziger's point was that for Wundt  the analysis of elementary sensations was "a limited, preliminary, and subsidary task". This statement implies that the the analysis of sensations was nowhere near as important for Wundt as is widely believed, but it also unambiguously implies that he did such analyses.


[4]Danzinger 1979 quotes Boring as saying " The teaching of Mach and Avenarius seems to have been ingrained even into Titchener's everyday thinking" (p. 215)


[5] Note that the atomic sentences are different from the physical atoms whose behavior they might be obliquely describing. An observation sentence might be described as an epistemic atom, for it is supposedly a piece of knowledge that stands on its own and provides a foundation upon which to base our knowledge of other more theoretical entities--like the atoms posited by physicists. Which is more fundamentally real, the observational atoms, or the theoretical atoms?  Positivist philosopher/physicist Ernst Mach comes down firmly on the side of the observational atoms when he says " Thus perceptions, presentations, volitions, and emotions, in short the whole inner and outer world, are put together in combinations of varying evanescence and permanence, out of a small number of homogeneous elements. Usually,  these elements are called sensations" (quoted in Danziger 1979 italics added). Given that everything in the list preceding the words in italics is usually consigned to the inner world, rather than the outer, this description probably didn't sit well with many of Mach's fellow physicists, who no doubt thought that their beloved electrons were a lot realer than sensations. The debate between those who side with Mach and those who want to grant higher ontological status to the electrons continues to this day under the name of realism vs. instrumentalism. The fact that such metaphysical debates are unavoidable, even among postivists, was one of the reasons that positivism was eventually acknowledged to be incoherent.


[6]When I say "all the way down", I don't mean to take any position on the question of physical atomism. What I am saying is that there is no reason to assert, and good reason to deny, that the organic processes that give rise to consciousness operate by the interaction of psychological atoms. One can explain the flow of water by means of fluid dynamics without making reference to H20 molecules. There is no reason that psychological processes couldn't be explicable only by laws that posit processes, and if so, there would be no such things as elementary sensations.


[7] Skinner did make an important supplement to Pavlov's concept by discovering operant conditioning, which differs from Pavlovian conditioning in that the stimulus comes after the desired response. Sometimes the word "reflex" is used in psychological texts to refer only to Pavlovian conditioning, but I am using it here to refer to operant conditioning as well. Both kinds of reflexes are claimed to be independent atoms, so the difference between the two is not relevant here.

I use the term "reflex" to refer to functional S-R connections, not to neurological ones, because Skinner used the word that way in his article  "Two types of conditioned reflex and a pseudo type". (Journal of General Psychology, 12, 66-77. ) Skinner's behaviorism was atomistic, even though it wasn't reductionistic, because he believed that his observations were genuinely theory-free--i.e. each observation could stand on it's own epistemicly. What he claimed was that S-R connections were functionally atomistic, because they could be studied independently of the atoms that were the subject matter of physics.  His operationalism was based on the assumption that as long as he didn't posit any mental entities that were responsible for behavior, he wasn't committing himself to any theory. But of course he was commiting himself to the theory that each observation was independent of all of the others- i.e. atomism. All of the logical Positivists-Carnap, Schlick, Russell-- eventually rejected operationalism for precisely this reason. Operationalism is self-contradictory because it cannot work unless observations are theory-independent, and the claim that observations are theory-independent is itself a theoretical claim.


[8] There were, of course, people sitting out in the universe before baseball was invented.  But none of those people were pitchers or first basemen until they started playing baseball. I cannot see any way that any of those people can have any characteristics that are not functionally dependent on their relationships to other conceptual structures of some sort, and can make no sense out of the concept of "bare particulars."

[9] Lakoff rarely mentions Chomsky by name, but that is because Chomsky's methods and assumptions are so widely accepted that they have become virtually synonymous with modern linguistics. Lakoff rightly concludes that most of the problems he has discovered with Chomskian linguistics stems from Chomsky' s reliance on principles that are accepted by almost everyone in modern science. Consequently he usually aims most of his criticisms at a set of principles he calls objectivism, rather than at Chomsky in particular. An interesting aside is that objectivism was also the name that Ayn Rand gave to her philosophy, and although Lakoff is clearly unaware of this, most of the principles that he critiques under this name are the same principles that Rand feels should be defended at all costs.