Printer Friendly

Expanding the explanatory base of behavior analysis via modern connectionism: selectionism as a common explanatory core.

Abstract

Selection is an important component of Darwin's functional explanation of the origin and extinction of species. It is an equally important component of Skinner's functional explanation of the origin and extinction of behavior. Darwin's functional explanation was marginalized within biology on the basis that it was an incomplete explanation because it lacked any plausible proximal causal mechanism for how variation was instantiated and for how natural selection could operate on this variation. Population genetics completed Darwin's evolutionary explanation by providing the required proximal causal mechanism information. Skinner's functional explanation of behavior has been marginalized within psychology for the same general reason that Darwin's theory was marginalized within biology. No proximal causal mechanisms are available to explain behavioral variation and how contingent consequences can selectively reinforce or strengthen target behaviors. Arguments that the experimental analysis of behavior can proceed without this information are correct in the same way that Darwin could continue his research in the absence of population genetics. However, history demonstrates that marginalization will remain until proximal causal information is provided. Parallel Distributed Processing Connectionist Neural Networks provide the requisite proximal causal explanations. This article demonstrates how this explanatory approach is fully compatible with the experimental analysis of behavior. Expansion of the explanatory basis of behavior analysis could potentially promote it within psychology to the same degree that population genetics promoted evolution within biology.

**********

This article notes that applied behavior analysis and its selectionist orientation have been marginalized but that selectionism is gaining acceptance at an exponential rate in the form of Parallel Distributed Processing (PDP) connectionism (10) which is theoretically consistent with radical behaviorism because it is a superset of radical behaviorism. It therefore follows that applied behavior analysis may be able to expand its appeal by extending its explanatory base to include the new selectionist explanations used by PDP connectionism.

The first section of this article summarizes the selectionist approach that underlies the experimental analysis of behavior and its behavior therapeutic applications. This is done to emphasize the importance and value of this explanatory approach and to demonstrate my commitment to this perspective so that there can be no doubt as to my sincerity in this matter. Another reason for beginning with this section is that PDP connectionism supports selectionism.

The second section of this article notes that explanation based on selection outside of PDP connectionism has become severely marginalized within psychology. A future consequence of this trend, if left unchecked, is that fewer and fewer proponents of applied behavior analysis will have less and less impact on science, clinical practice, and education. Representation and influence in professional societies will continue to wane. It is time to act in new more effective ways before extinction fully occurs.

The third portion of this article proposes a "recovery plan" based on PDP connectionism. The main reasons are: 1) this form of connectionism fully embraces selectionist explanation, 2) it constitutes a superset of behaviorism that is compatible with mainstream psychology, and 3) its growth is exponential. Empirical support for these three claims is provided by the work of John Donahoe, a "card-carrying" behaviorist.

SELECTION OF BEHAVIOR

The word operant means to operate on the physical and/or social environment. Operants refer to theoretical distributions that describe slight variations in how the same behavior recurs in the same situation. We shall refer to them as response variations to highlight explanation by selection. For example, a rat does not consistently press a bar in exactly the same way. Slight variations in force, duration, and body position characterize each bar press. Some response variants effectively depress the lever; others do not. Contingent food access alters the distribution of operant response variants. The frequency of ineffective response variants will decrease whereas the frequency of effective response variants will increase. We therefore say that the environment selects behavior. We can equivalently assert that consequences shape behavior.

Extensive experimental evidence clearly demonstrates that contingent consequences alter the shape of operant response variant distributions. Removing response variants from the low end of a distribution causes the mean of the remaining response variants to be higher than the original mean. Removing response variants at the high end of a distribution causes the mean of the remaining response variants to be lower than the original mean. Stated otherwise, selectively deleting response variants from either end of an operant distribution modifies the remainder of the response variant distribution in predictable ways, which modifies the organism's future behavior. This scientific principle or law has been empirically supported by laboratory evidence derived from many species and over various responses. Supporting evidence is too extensive and well known to be reviewed here.

It is descriptively accurate to say that the effective response variants are reinforced by contingent access to food as long as reinforced only means to make stronger, more forceful and/or of longer duration or more probable. It is equally correct to say that the ineffective response variants are extinguished as long as extinguished means only that the organism emits them less frequently if at all. The devastating distortions that currently surround this terminology have altered my vocabulary variations to where I now speak almost exclusively of selecting response variants.

The selection of behavior is a truly general process and pertains to all aspects of what people do; including relationships with their pets and other people. A well-known psychologist related the following informative story. His routine upon arriving home from work began by petting his dog who eagerly greeted him. At some point he became increasingly preoccupied with work and gradually paid less and less attention to his dog upon arriving home. This behavior change on his part meant that an increasingly large proportion of his dog's greeting response variants became ineffective in that they no longer set the occasion for being petted. This behavioral change on the part of the pet owner steadily modified the dog's greeting response distribution. The modal greeting response variant gradually shifted and was shaped toward more aggressive behavior. The pet owner first noticed these behavior changes when his dog greeted him by jumping up on him and licking his face. Initially he was puzzled and wondered what had "gotten into his dog"; why had he changed so much. The dog's greeting behavior generalized to visitors and thereby became more problematic. This psychologist's understanding of behavioral selection enabled him to see how his behavior had shaped his dog's greeting response variants. This insight led to an effective treatment. He greeted his dog before his dog greeted him thereby drastically reducing what the dog had to do to get petted. That behavioral change on his part edited the dog's response variant distribution in the opposite way. The distribution of the dog greeting response variants soon returned to what it once was and all was well ever after. This dog story would work just as well if it were about two people. Relationships entail mutual control and influence, whether based on positive or negative consequences. We continually modify response variants about many facets of each other's behavior whether we intend on doing so or not. Sometimes this process produces behavior disorder that leads to professional referral. People are frequently unable to improve their situation because they do not understand the selection process whereby distributions of response variants are shaped. Formulating behavior disorder in selectionist terms is a special skill that all behavior therapists should have. Bringing this knowledge to psychologists in general and clinical psychologists in particular is one important contribution that behaviorists have long been trying to make.

Cognitive and trait explanations have not clarified the functional relationship between consequences and the shape of response variant distributions because they do not address these issues. Rather, they engage explanation on a structural basis. They look inside the organism for personality structures such as traits and when such are found they are not related to distributions of response variants. Instead, trait proponents presume an unexplained relationship between internal psychological states and the behaviors to be explained; they show no interest nor make any effort to explain how psychological states are related to distributions of response variants. These structural explanations impede selectionist explanations because they take the explanatory search away from response variation distributions and the contingent consequences that reshape them.

Selectionist explanations are developmental explanations because they are based on the life history of the organism. The selection of behavior frequently entails the behavior of other people and often occurs in social contexts, which makes selection relevant to social, family, community, and group psychology. The selection of behavior changes the organism physically in ways that biologists and neuroscientists can now measure which makes behavioral selection relevant to and compatible with neuroscience. Hence, many developmental, social, and biological changes can be explained on the basis of behavioral selection. A science of behavior based on selection provides general scientific explanations that stand on their own. For example, neuroscience explanations will only add relevant details about how the process of behavioral selection takes place but will not alter the fact that behavioral selection occurs in the way that the experimental analysis of behavior has established. Modern inquiry within biology complements and supports selectionist explanations. Modern inquiry within psychology occurs mainly in opposition to selectionist explanations.

THE MARGINALIZATION AND POSSIBLE EXTINCTION OF BEHAVIOR ANALYSIS

Selectionist explanation dominates biological theory. Darwin's evolution by variation and natural selection has been hailed as perhaps the most important and revolutionary idea in all of science. Skinner's explanation of behavioral selection is directly parallel with Darwin's explanation in that both depend upon variation and selection (Skinner, 1963, 1966, 1975, 1981, 1984a, 1984b; Smith 1983, 1984; Tryon, 1993). Selectionist explanation of behavior is both parsimonious and superbly supported by data collected under well-controlled laboratory conditions for various behaviors in multiple species including human behavior. Many useful clinical studies and demonstrations of therapeutic value could be cited that are based on selectionism. One would therefore think that explanation by selection should be highly regarded and commonly practiced in psychology generally and clinical psychology in particular.

It is therefore surprising that selection is a decidedly unpopular and marginalized explanatory approach within psychology. When discussed at all, it is usually pitted against a preferred cognitive explanatory approach. This sentiment is reflected in professional affiliations. Figures recently obtained from the American Psychological Association show that of its 155,000 current members only 650, less than half of one percent, support the experimental analysis of behavior by membership in Division 25. Only a small minority of Association for the Advancement of Behavior Therapy (AABT) members identify themselves with the experimental analysis of behavior. The large majority of psychologists and AABT members endorse a cognitive-behavioral perspective while many others claim a purely cognitive orientation. This was not always the case. The selectionist approach once was much better represented in these and related professional organizations. We may therefore reasonably conclude that selectionism is on the decline in psychology. The graying of the American professorate implies accelerated retirement of selectionists as well as cognitivists. However, the differential rates with which selectionists and cognitivists are replacing themselves professionally indicates that cognitivists will predominate among new hires thereby further eroding and marginalizing the selectionist position. We can anticipate that fewer and fewer courses will teach selectionism thereby educating fewer and fewer students about this orientation resulting in fewer and fewer selectionist job candidates and less interest in hiring them. Left unchecked, these developments will decimate membership in professional organizations to the point where revenues will no longer support journal publication. It is alarming but not altogether incorrect to say that explanation by selection has been marginalized to the point of where its extinction is within view. Current verbal response variants by advocates of selectionism may be technically correct and highly articulate but the effects of their rhetoric seem either to be ineffective or to be further alienating the majority of psychologists. What is wrong here? Perhaps a functional analysis of this behavior is in order.

We need look no further than the history of biology and its initial response to Darwin's theory of evolution by variation and selection to find a major reason why selection is not an acceptable or satisfactory explanation to many psychologists (cf. Donahoe, 1997). One could not possibly tell from the current fanfare surrounding Darwin that most biologists initially rejected his views for more than 75 years. See Bowler (1983), Catania (1978, 1987), Donahoe, Burgos, and Palmer (1993), Mayer (1982), and Tryon (1993) for further details. Darwin presented a functional analysis that interrelated biological variation and selection by natural consequences. Opposition to his functional analysis was based on his lack of plausible proximal causal mechanisms for how variation was instantiated and for how consequences altered that variation because genetics was an unknown science at that time. His functional explanation was therefore deemed partial and insufficient even though he spent the greater portion of his adult life compiling voluminous and meticulous support for his views. No amount of empirical support for his functional theory could have persuaded other scientists because such data did not address the absence of plausible proximal causal, explanatory, mechanisms for how variation and selection exerted their effects. No amount of confirmation that his functional analysis was correct could explain why it was correct. Biologists rejected Darwin's functional theory of evolution for more than 75 years until the field of population genetics satisfactorily explained both issues. Addressing the genetic basis of variation and selection augmented rather than diminished Darwin's contributions. Darwin would likely be an obscure figure in the history of science today if plausible proximal explanatory mechanisms had not been found for his functional theory. I submit that psychologists are withholding general acceptance of selectionist explanations of behavior for similar reasons. It follows logically that marginalization of selectionist explanations within psychology will continue until satisfactory explanations of these matters have are presented. Donahoe (1997) concurs. It also follows that Skinner may be promoted within psychology as Darwin was within biology by extending explanation to plausible proximal causal mechanisms underlying behavioral selection.

A second reason why I believe that psychologists have not accepted behavioral selection is that current explanations by selection are not psychological because they do not engage psychological processes of perception, learning, and/or memory, etc. Put otherwise, the absence of psychological processes from plausible proximal causal mechanisms driving behavioral variation and selection precludes such explanations of behavior from being psychological. Making no contact with any psychological process has not proved to be a very good marketing strategy for convincing psychologists to endorse behavioral selection (11).

Isolationism is a third factor; corollary to the second factor that inhibits the acceptance of selectionist explanation of behavior. Wilson (1998) observed that the natural sciences are integrated. He coined the term consilience to refer to the integration of theory and method across the natural sciences. The experimental analysis of behavior presently shows little evidence of consilience. Behaviorists shun attempts to provide plausible proximal causal explanations in neuroscience terms as much as they resist explanations in psychological terms. Behaviorists do not have plausible proximal causal mechanisms that they prefer to psychological or neuroscience explanations; they simply oppose all attempts to completely explain behavioral selection by addressing intermediary causal steps; by addressing the 0 in the S-O-R model. At least three motivations can be identified for this position. One motive is that a complete explanation is not necessary because one can conduct behavioral experiments and perform behavior therapy in the absence of a complete scientific understanding of behavioral selection. A second motive is that all considerations of plausible proximal causal mechanisms get in the way of completing a functional analysis and therefore are to be opposed as part of the problem rather than part of the solution. A third motive is that all proximal causal explanations are necessarily circular and therefore unacceptable.

All three motives are misguided. With regard to the first motive, just as it was possible for Darwin to conduct many studies in the absence of any plausible proximal mechanisms to explain his functional analyses, so it is possible for behaviorists to conduct experiments and provide therapy based on functional analysis without any further causal understanding of behavioral selection. However, behaviorists should also expect to be marginalized by the scientific community as completely as Darwin initially was because they cannot fully explain why their functional relationships work as they do. Assertions that a complete understanding of behavioral selection is beyond the scope of behavioral analysis are unacceptable. Darwin's functional theory of evolution was never accepted on its own terms. Its marginalization prevailed for more than 75 years and would have continued indefinitely had explanation not proceeded past a functional analysis. Skinner's functional theory is currently marginalized and will remain so until a more complete understanding of behavioral selection is provided. Arguments that the experimental analysis of behavior stands on its own may be motivated by an attempt to justify this branch of science but such behavior has had the unintended consequence of shunning all other scientists on the basis that their work makes no contribution to understanding behavioral selection. Such ruptures of consilience continue to isolate the experimental analysis of behavior within the general scientific community. Acceptance of evolutionary theory was conditional upon consilience with population genetics because it provided the necessary explanatory extensions to account for how behavioral variation was instantiated and how the process of natural selection could work as Darwin described. Darwin had to "go it alone" because the field of genetics had not yet been developed. Skinner and his intellectual descendents have chosen to go it alone. Isolation and marginalization are the clear consequences in both cases.

With regard to the second motive, Darwin's functional analysis in terms of variation and selection was advanced not retarded by genetic mechanisms. It is therefore not true that consideration of proximal causal mechanisms always interferes with a functional analysis. Citing instances where attention to proximal causal mechanisms has interrupted completion of a functional analysis is not proof that all efforts to understand the proximal causal processes associated with behavioral selection will do so.

With regard to the third motive, the formerly circularly defined hypothetical construct of learning has been replaced by an objective neuroscience analysis of underlying biological events (see below). Further progress in this arena is the objective of cognitive neuroscience. Not incorporating neuroscience findings has not proved to be a very good marketing strategy for convincing scientists in general to endorse behavioral selection.

All three of these reasons for rejecting behavioral selection must be addressed, in my opinion, if the selectionist understanding of behavior is to become widely endorsed by psychologists. Some proximal causal mechanism for how response variant distributions are instantiated and how they are reshaped by contingent consequences that engage the interests of psychologists, neuroscientists, and other natural scientists must be found or endorsed if the experimental analysis of behavior hopes to maintain the interest of more than a few people. PDP connectionism provides an opportunity to do this in a way that does not repudiate fundamental theory construction values held by behavior analysts. Donahoe and Dorsel (1997) and their contributor's concur with this view.

PDP CONNECTIONISM NEURAL NETWORK LEARNING THEORY

Connectionism exists in many forms. I refer exclusively to the parallel-distributed processing (PDP) version of connectionism. This form of connectionism is opposed to the same black box and arrow symbol manipulating cognitive psychology that Skinner and others have long criticized. PDP connectionists note that symbol manipulating cognitive psychologists do not explain how cognition works or how it influences behavior. Ascribing functional properties to cognitive constructs such as schemas is insufficient because no explanation of how schemas accomplish their functions is provided. Their theoretical assertions are functional statements absent any plausible proximal causal explanation for how these functions are implemented (12). On the other hand, symbol manipulating cognitive psychologists criticize PDP connectionism as a form of neo-Behaviorism (cf. Elman et al., 1996, pp. 103-104).

The sections below sketch essential points regarding PDP connectionism. Those who wish to learn more should consult introductory texts to this field (e.g., Elman et al., 1996; McLeod, Plunkett & Rolls, 1998; O'Reilly, & Munakata, 2000; Thagard, 2000).

Learning and Reinforcement

Learning has long been factually, empirically, defined as "... a more or less permanent change in behavior which occurs as a result of practice" (Kimble, 1961, p. 2). This approach defines learning as an inferred variable, a hypothetical construct, characterized in terms of behavior change. It therefore cannot be used to explain behavior change. To argue: 1) behavior has changed, 2) this behavior change is evidence that learning occurred, 3) This behavior change occurred because of learning is logically defective because it entails circular reasoning. Observation is followed by inference, which is used to explain the observation. Observed behavior change is being used to both infer learning and to explain itself. The concept of conditioning is considered to be a subset of learning that is associated with contingent environmental consequence and therefore entails the same circular reasoning. So is the concept of reinforcement (see below).

There have been many theoretical definitions of learning. We confine ourselves to Hebb (1949) who speculated that learning alters synaptic properties. Hebb (1949, p. 62) conjectured "When the axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased" (Levine, 199 1, pp. 16-18; Wasserman, 1989, pp. 212-214). This approach once suffered from the "and then something magical happens about here" limitation. Skinner (1938, p. 421) criticized such speculations as involving the Conceptual Nervous System (CNS) rather than the Central Nervous System. However, neuroscience has since confirmed Hebb's conjecture and demonstrated through direct measurement under controlled laboratory conditions that learning entails synaptic change (see Function alters Structure section below). Reinforcement occurs because synaptic properties are changed by certain contingent consequences. These findings are sufficiently dependable that one can now safely and correctly infer specific biological changes on the basis of the specific behavioral changes that have previously been taken as evidence of learning. Defining learning in biological terms: 1) infuses it with physical reality and removes it from the realm of hypothetical speculation and 2) solves the circularity problem.

Hebb's learning principle is called the Hebbian learning rule when implemented mathematically in PDP simulations. Another learning rule, the Delta learning rule, is mathematically equal to the Rescorla-Wagner model of classical conditioning (Levine, 1991, pp. 58-60). This similarity reflects fundamental agreement between behaviorism and PDP connectionism about how learning occurs.

The singularly most important fact here for present purposes is the role played by selection in the learning process. Consider a network of neurons interconnected with synapses. Contingent consequences selectively strengthen, reinforce, particular pathways by creating synaptic Long-term Potentiation (LTP) or selectively weaken specific pathways by creating synaptic Long-term Depression (LTD) (See Function Alters Structure section). Synapses connecting pairs of neurons are strengthened or not depending upon their activation levels, which depends on the sum of the stimuli received across their many dendrites and the sigmoidal activation function that triggers depolarization. These are all deterministic processes that begin with sensory input and are modified by contingent consequences. Neural Darwinism is a term that has been used to characterize these processes (cf. Levine, 199 1, pp. 239-24 1). These selectionist processes are driven by consequences rather than by theorist design. PDP connectionism provides a selectionist account of how structure and function interact in a cumulative way to produce behavioral and psychological development. It is an important theory construction advance that PDP connectionism can support such an integrated causal explanation.

It is not generally acknowledged that the concept of reinforcement is also circularly defined. A reinforcer is functionally identified by the increase in frequency, intensity, duration, or probability of behavior associated with its onset contingent upon the emission of a target behavior. But then the increased frequency, intensity, duration, or probability of the target behavior is explained in terms of reinforcement. Woods' (1974) taxonomy of instrumental conditioning provides another way to see this. He defines Reward Conditioning in terms of the contingent onset of a positive reinforcer and notes that this results in the strengthening of behavior. However, a positive reinforcer is identified by its ability to strengthen behavior when presented contingently. Explaining an increase in the frequency, intensity, duration, or probability of behavior via reinforcement is circular. This explanatory practice has been accepted despite its flawed circular logic because investigators and therapists can manipulate reinforcers once they have been identified. However, it remains the case that the observation, behavior change, used to identify a reinforcer makes explaining the behavior change in terms of reinforcement circular. The problem is parallel with the concept of learning and other behaviorally defined constructs. In general, functional theories cannot escape circular definitions because they cannot go beyond a functional description of behavior. They cannot explain why or how functional relationships occur; they can only document that they occur. One cannot ask the legitimate and reasonable question, "why is a reinforcer reinforcing"; or equivalently, "why do some stimuli alter behavior when presented contingently and others do not?" Citing any other functional relationship at the behavior-analytic level cannot adequately answer such questions. Functional relationships among other than behavioral (e.g., biological) processes and/or structural reasons must be cited in order to provide plausible proximal causal mechanisms for the documented functional behavioral-analytic relationships. Functional relationships among biological variables pertinent to synaptic function have already been provided to explain learning. They also explain why reinforcers are reinforcing. Functional theories are always incomplete because they provide incomplete explanations due to self-imposed restrictions on their explanatory base. We now consider structural explanations.

Structure Determines Function

I use the term structure to refer to the physical structure of the brain, including the status of the trillion or so synapses, at any particular instant in time and could extend this definition to include the entire organism if necessary. Response to the very next stimulus is completely determined by said structure, which reflects the cumulative reinforcement history of the organism, its genetic endowment, and all developmental factors leading to its current state at a particular instant in time. This configuration can be confidently said to determine the very next response because all variables with physical existence have been included up to that point in time leaving no room for any other variable to confound this explanation. PDP connectionism primarily characterizes this state in terms of a) neural architecture, and b) the activation (excitatory or inhibitory) status of all synapses entailed by this architecture.

Neural architecture refers to the physical interconnections of all neurons. Many brain structures are layered and impulses travel from one layer to another. Impulses also travel from one structure to another and back again. Both excitatory and inhibitory pathways are present. Some simplification of this enormously complex situation is presently required in order to advance inquiry. Parallel Distributed Processing Connectionist Neural Networks (PDP-CNNs) therefore presently entail a few layers of nodes that are frequently called neurons because their functioning is limited to important basic neural properties including some form of dendritic summation and a nonlinear activation function that determines if the neuron fires or not. Stimulus inputs are presented to the first layer in the form of a distributed representation that can be at any of several levels of abstraction ranging from global characteristics down to individual receptors. Neurons that fire propagate their impulses to subsequent layers. Neural architectures the implement recurrent networks feed impulses back to prior layers. Behavior is indicated by the status of output (response) neurons. These networks yield specific results based on the complex interaction of first principles. This level of mechanism specificity goes well beyond the black box and arrow models that characterize most of cognitive psychology.

An extensive literature now supports the view that brain damage alters behavior (Kalat, 2001; Kolb & Wbishaw, 1996). Sach's (1985) colorful clinical description of patients with neurological deficits indicates that the brain is highly modularized and that damage to any module modifies behavior; often in bizzare ways.

Neural architecture is strongly influenced by our genetic material (DNA). Learning entails association and NMDA receptors have been shown to influence how well associations are formed. NMDA receptors are composed of NR2A and NR2B subunits. NR2B subunits stay open longer than NR2A subunits enabling them to better detect when two neurons are concurrently active and setting the occasion for the biochemical cascade that defines learning. Staying open longer allows more calcium ions to pass which makes the NR2B response more vigorous than the NR2A response. NR2B units predominate in young organisms while NR2A units predominate in old organisms. The superior functional properties of NR2B subunits promote better learning in young than old organisms.

Normal genetic expression entails a shift from NR2B to NR2A units over the organism's lifetime. Tang et al. (1999) created transgenic mice by linking a copy of the mouse NR2B gene to a promoter that is only active in the forebrain (bippocampus and amygdala) and injected it into fertilized mouse eggs. These additional genes, along with the mouse's own NR2B producing gene, produced an over expression of NR2B subunits in the forebrain of the resulting mice. These genetically altered mice excelled on six behavioral tests of learning compared to control subjects. These tasks included a novel-object recognition task, a retention test, contextual and cued fear conditioning tasks, a fear-extinction task, and a hidden-platform water maze. Genetic alteration was the independent variable with learning and memory measures as dependent variables. The resulting experimental evidence of the causal influence of genetics on behavior directly addresses the interactive issue posed by Hayes (1998) and avoids the problem of biologism raised by Hayes (1998). See Tryon (2000b) for further details.

Without PDP connectionism there would be no understanding of how NMDA receptor properties could influence behavior. Only by understanding how multilayered networks give rise to behavior can one see how changing synaptic properties by modifying NMDA receptors causes behavioral changes. The PDP connectionist demand for mechanism answers drives the field of behavior genetics well beyond summary percentage of variance accounted for statements and pointless heredity vs. environment debates. Alternatively, selectionist explanations that impact NMDA subunit distributions are persuasive. PDP Connectionism advances our understanding of behavior genetics by explaining how the physical changes wrought by genetics influences the emergence of learning, memory, and behavior.

Function Alters Structure

Skinner rejected what he called the metaphor of storage and maintained that experience simply changes the organism and how it behaves. Frey (1997), Kalat (2001, pp. 385-388), and Kandel (1989, 1991), Rolls and Treves (1998, pp. 322-325), and Spitzer (1993, pp. 42-5 1) present empirical evidence showing that learning entails changes in synaptic function called Long-term Potentiation (LTP) and Long-term Depression (LTD) (Kalat, 2001; Kolb, & Wbishaw, 1996; Singer, 1997). These changes include genetic expression through RNA transcription of DNA segments that result in new protein synthesis. Some structural changes also occur. Reinforcement and all other forms of learning entail these basic processes.

The PDP connectionist approach to learning is phylogenetically general (Tryon, 1995b). The same vocabulary and set of basic concepts apply equally well to humans as to animals. Laboratory research with animals that supports PDP connectionism is accepted much more readily than is laboratory research on behavior analysis. Gluck and Bower (1988) place the divorce between human and animal learning in psychology at around 1968. PDP connectionism provides a contemporary platform for making animal research relevant to human behavior.

Theory Construction Values

This section attempts to clarify the important theoretical values that behaviorism and PDP connectionism share. This is important because behavior analytic proponents rightfully do not want to broaden their explanatory base by compromising important theory construction values (cf. Tryon, 1995a, 1996).

Explanation by selection., Discussion above and below this section articulates the role of selection in PDP connectionist neural networks. Donahoe (1997), Donahoe, Burgos, and Palmer (1993), and Donahoe and Palmer (1989) further document this point.

Absence of homunculus. Skinner (1977, 1989) criticized explanations based on an inner process or being that made decisions. He correctly observed that such explanations merely redescribe behavior in psychological terms that beg the question of how the homunculus arrived at the decision cited as the causal explanation for the observed behavior. To say that I behaved in some way because I chose to do so explains nothing until the reasons for my choice have been identified. PDP connectionism does not appeal to homunculi. Choice is an emergent network property that is computed iteratively from first principles. Computer simulations are conducted to implement and document every step in a sequence resulting in choice. Each neuron knows only what it learns from its neighbors yet the network settles into a final state that is associated with one response or another. Choice is a holistic emergent property of network function; not the action of a homunculus.

No rules. Skinner (1977) maintained, "Rules are widely used as mental surrogates of behavior ..." (p. 8). To explain behavior as an instance of rule following begs the explanatory question until one explains how the behavior from which the rule is inferred was shaped. Asserting that behavior is rule governed begs the explanatory question of how this is possible. How does reading, verbalizing, or thinking about a rule lead to behavior? Rule learning suffers from the same circular logic that once characterized learning. Rules are inferred from behavioral regularities and these regularities are then explained as instances of rule following. PDP connectionism does not explain behavior in terms of rule following. Moreover, Allan (1993) reported that human contingency judgments are better accounted for by connectionist models than by rule-based models.

No copy theory of perception. Skinner (1977, 1989) properly objected to the copy theory of perception because it begs the question of how the copy is perceived inside the organism. PDP-CNN investigators such as Finkel and Sajda (1992, 1994) have simulated the perception of visual illusions using an artificial 128 x 128 pixel "retina7 and a ten layer neural network. No copy theory of perception was used.

Experience changes the organism. We have already discussed above how contingent consequences set the occasion for a biochemical cascade that alters synaptic properties that change behavior.

Learning. This term is defined by biochemical and physical changes that have been independently and objectively observed and quantified under laboratory conditions. Learning is no longer defined in a circular way.

Determinism. Synaptic changes are deterministically produced by contingent consequences. Neural processing leading to emergent behavior is frequently modeled in a deterministic way.

EMPIRICAL EVIDENCE

The work of John W. Donahoe provides clear evidence of the fundamental compatibility of PDP connectionism and the experimental analysis of behavior. His writings (Donahoe, 1991, 1997; Donahoe, Burgos, & Palmer, 1993; Donahoe & Palmer, 1989; Donahoe, Palmer & Burgos, 1997a; 1997b) and his web page [http://www.umass.edu/neuro/faculty/donahoe.html] detail a biobehavioral approach to the experimental analysis of behavior that embraces PDP connectionism and neural network simulations in order to better understand the role of selection in reinforcement and stimulus control. His leadership in this field is exemplary and points to a viable future.

NEURAL NETWORK LEARNING THEORY

I coined the term Neural Network Learning Theory to refer to Parallel Distributed Processing Connectionist Neural Networks for four reasons. First, learning, which implies memory or else it cannot be cumulative, is arguably the primary psychological principle in that so much of development is influenced by what we learn. PDP connectionism makes clear that learning is involved in all that we do; e.g., perception is learned, memories are learned, etc ... I have argued elsewhere that behavior therapy should be defined in terms of applied learning theory (Tryon, 2000a). I want to emphasize that PDP connectionism greatly informs us about the learning process and the biological mechanisms that support it.

Second, PDP connectionism occurs in a network context. While it is possible to discuss learning in terms of network nodes, these nodes have been given important properties of real neurons. Neuroscience has informed and continues to inform these learning models. I chose to recognize these interdisciplinary contributions by referring directly to neural networks. I limit this reference to the PDP versions of neural network models, as other forms of connectionism are not as compatible with behaviorism as is this form.

Third, I did not want to substitute one ism for another. I did not want to appear to be against behaviorism by being for connectionism. I now realize that PDP connectionism is a superset of behaviorism in that the former fully contains the latter. This is the logical basis for arguing that embracing the broader explanatory base associated with PDP connectionism does not require any compromise of behaviorist principle. This is a "no cost" explanatory extension.

Fourth, concerns parsimony. NNLT entails four words whereas PDP-CNN entails six.

RELATIONAL FRAME THEORY

Hayes, Barnes-Holmes and Roche (2001) present Relational Frame Theory (RFT) as a post-Skinnerian account of human language and cognition that they would like to see supplant contemporary psychological approaches to cognition and language. By post-Skinnerian they mean more than a theory that is presented after Skinner's death. They mean a theory that goes beyond the explanatory bounds Skinner would permit. They view this extension as desirable and if successful would provide an alternative to the PDP connectionist approach discussed above. The next few paragraphs identify several fatal flaws in this proposal.

Equivalence classes are fundamental to RFT but they are acknowledged to be a behavioral Varado in that conditional discriminations are not expected to either reverse or combine (Hayes, et al., 2001, p. 18). Training to select stimulus B given stimulus A should not also produce selecting stimulus A given stimulus B but that is what occurs with stimulus equivalence. Stimulus equivalence contradicts behavioral theory and cannot be used to support or extend it. Hayes et al. (2001, p. 19) noted the relevance of equivalence classes to language but did not recognize their obligation to explain the phenomena of stimulus equivalence prior to speculating on how it helps explain human cognition and language. Citing an unexplained behavioral paradox does not provide an adequate explanation of anything.

Hayes et al. (2001, p. 21) referred to "derived relational responding" as the kernel of their behavioral analysis of language and cognition. These authors further asserted, "Relational Frame Theory embraces the simple idea that deriving stimulus relations is learned behavior" (p. 22). Learning in this context is circularly defined and therefore cannot be used as a behavioral explanation. Observing instances of relational responding occasions inferences about learning which, are used to explain relational framing. The author's state, "Verbal behavior is the action of framing events relationally, and verbal stimuli are stimuli that have their effects because they participate in relational frames" (p. 144). Deriving stimulus relations, relational frames, and relational responding are inferences based on behavior and cannot be used to explain anything else without engaging in the same circular reasoning and hypothetical constructs that cognitive psychologists are criticized for using. To say that verbal stimuli have their effects because they participate in relational frames is circular because relational framing was inferred from the verbal behaviors used to certify the presence of the relational frame. Using circularly defined hypothetical constructs to explain cognition and language truly constitutes a post-Skinnerian theory but not in the positive way intended by the authors.

Absent the above criticisms, RFT would be a partial explanation at best because it remains a functional analysis. No plausible proximal causal mechanisms are provided to explain the many presumed processes entailed in learning relational frames and how they change behavior. It remains a black box and arrow model with new names for the boxes (13).

Hayes et al. (2001) promote RFT as a self-contained theory. This might be a virtue if mature sciences were all independent of each other. On the contrary, Wilson's (1998) book entitled Consilience is subtitled The Unity of Knowledge because mature sciences are interrelated and form an interwoven tapestry of knowledge. RFT's avoidance of interlevel theory stands in opposition to the community of natural science and therefore constitutes an impediment to greater acceptance of the experimental analysis of behavior within the larger scientific community.

PDP-Connectionism on the other hand is extraordinarily consilient. Its fundamental compatibility with radical behaviorism, neuroscience, and cognitive psychology provides a heuristic link to the mature sciences that will advance the experimental analysis of behavior. Donahoe (1997) further articulates the benefits of PDP-connectionism for the experimental analysis of behavior in his chapter entitled "The Necessity of Neural Networks" (emphasis added).

CONCLUSIONS

The first section of this article summarized the selectionist approach that underlies the experimental analysis of behavior and its behavior therapeutic applications in order to emphasize the importance and value of this explanatory approach and to demonstrate my commitment to this perspective so that my subsequent comments could not be dismissed as a someone who was not sympathetic to the experimental analysis of behavior. I have personally witnessed the marginalization of behavior analysis for over 30 years since starting graduate school in 1966. 1 believe that PDP connectionism provides a way to reverse this trend because: 1) it entails explanation by selection rather than design, 2) it entails a superset of behaviorism that is consilient with both contemporary cognitive psychology and neuroscience, and 3) its growth is exponential. I am not the first person to promote PDP connectionism within the experimental analysis of behavior (cf. Donahoe's work). I believe that the strong substantive and historical parallels between Darwin and Skinner clearly point to the necessity for a consilient extension of the explanatory base underlying behavioral selection and predict a bright future for the experimental analysis of behavior only if these historical lessons are heeded.

The fundamental theoretical similarity that unites both PDP connectionism and radical behaviorism is that they are both selectionist explanations. Donahoe and Palmer (1989) reviewed Rumelhart and McClelland's (1986) seminal work on PDP connectionism for readers of the Journal of the Experimental Analysis of Behavior. They stated "These (PDP connectionist) models are fundamentally different from typical models of cognitive psychology in that they are selectionist rather than essentialist in flavor. That is, the functionality of connections among the units is the result of selection by the environment rather than design by the theorist" (p. 399). Donahoe and Palmer emphasized the fundamental theoretical compatibility of radical behaviorism and PDP connectionism. They wrote, "It is clear, then, that adaptive networks simulate complex behavior through a selection process (i.e., "learning") and that the selection process is a function of the consequences scheduled for the output of the network. In behavior-analytic terms, complex environment-behavior relations in adaptive networks are the product of selection by reinforcement" (p. 404). Latter on they wrote "The accounts of differential conditioning provided by adaptive networks and experimental-analytic findings are strikingly and persuasively congenial" (p. 408).

The subtitle of this article "Selectionism as a Common Explanatory Core" emphasizes that behaviorism and PDP connectionism share explanation by selection. Succinctly stated, Darwin replaced creationism with phylogenetic selection and Skinner similarly replaced what he described as creationist cognitive psychology with ontogenetic selection. PDP connectionism extends selection to neuroscience in a way that encompasses both Darwin's and Skinner's explanations. PDP connectionism is a superset of behaviorism that includes both Skinner and Darwin.

Explanation

But what are the benefits of an expanded explanatory base? Why should we make such an effort? One benefit is that the resulting theoretical synthesis is a major scientific achievement on its own; one that helps repair the corrosive disunity that continues to characterize psychology (cf. Staats, 1983). Explanation is a fundamental goal of science. Disciplines such as astronomy rely entirely on explanation and prediction because they cannot control astrophysical phenomena. A related benefit is to promote consilience with neuroscience in particular and other natural sciences more generally for that is the hallmark of a mature science (cf. Wilson, 1998).

These benefits may not motivate psychologists who neither value explanation beyond functional analysis nor see themselves in interdisciplinary terms or who do not wish to learn about a field they chose to forgo in favor of psychology. For those readers who are motivated more by practical than scientific values, a final "bottom line" benefit for endorsing PDP connectionism is that it effectively promotes explanation by selection among psychologists and thereby reverse the marginalization of applied behavior analysis. PDP connectionism is a major marketing strategy for making selectionism relevant to mainstream psychology in a way that classic radical behaviorism has not.

Consilience

The extraordinary interdisciplinary and consilient nature of PDP connectionism is very reassuring and exciting. How could so many intelligent people in so many disciplines be wrong about the perceived benefits of this approach? Psychologists, all branches of neuroscience, mathematicians, physicists, engineers, and philosophers are working together in this field. Participating in such a large intellectual consortium is stimulating and supportive. It feels good to be part of an expanding and vibrant future especially when one has observed the progressive marginalization of their core discipline over the past 30+ years.

On the other hand, interdisciplinary study is daunting. I find that I cannot understand many articles written by mathematicians and physicists because they presume doctoral study in these areas that I have not had. I read their introductions and conclusions with interest and rely on my colleagues in those disciplines to evaluate the adequacy of the technical merits. One need not work in all areas of PDP connectionism in order to endorse it. Any degree of participation is welcome. Those who do not have and/or cannot develop the requisite skills can support others who wish to expand their explanatory base. I have learned to accept my limitations and make contributions where I can. I am pleased that my Bidirectional Memory Model of PTSD satisfies all published standards for what a complete theory must accomplish including making novel predictions (cf. Tryon, 1998, 1999). 1 am also pleased with the theoretical integration that this interdisciplinary perspective affords (cf. Tryon, 1993,1995a,1995b,1996,2000a,2000b).

REFERENCES

Allan, I. G. (1993). Human contingency judgments: Rule based or associative? Psychological Bulletin, 114, 435-448.

Bower, G. H., & Hilgard, E. R. (1981). Theories of learning (5th ed). Englewood Cliffs, NJ: Prentice-Hall.

Bowler, P. J. (1983). The eclipse of Darwinism: Anti-Darwinian evolution theories in the decades around 1900. Baltimore: The Johns Hopkins University Press.

Catania, A. C. (1978). The psychology of learning: Some lessons from the Darwinian revolution. Annals of the New York Academy of Sciences, 309, 18-28.

Catania, A. C. (1987). Some Darwinian lessons for behavior analysis: A review of Bowler's The Eclipse of Darwinism. Journal of the Experimental Analysis of Behavior, 47, 249-257.

Donahoe, J. W. (1991). The selectionist approach to verbal behavior: Potential contributions of neuropsychology and connectionism. In L. J. Hayes, and P. N. Chase (Eds.) Dialogues on verbal behavior: The first international institute on verbal relations. Reno: NV: Context Press.

Donahoe, J. W. (1997). The necessity of neural networks. In G. E. Stelmach & P. A. Vroon (Series Eds.) & J. W. Donahoe & V. P. Dorsel (Vol. Eds.), Advances in psychology: Vol. 121, Neural-network models of cognition: Biobehavioral foundations (pp. 1-19). Amsterdam: Elsevier.

Donahoe, J. W., Burgos, J. E., & Palmer, D. C. (1993). A selectionist approach to reinforcement. Journal of the Experimental Analysis of Behavior, 60, 17-40.

Donahoe, J. W., & Dorsel, V. P. (Eds.) (1997). Neural-network models of cognition: Biobehavioral foundations. In G. E. Stelmach & P. A. Vroon (Eds.) Advances in psychology (vol 121). Amsterdam, Elsevier.

Donahoe, J. W., & Palmer, D. C. (1989). The interpretation of complex human behavior: Some reactions to parallel distributed processing, edited by J. L. McClelland, D. E. Rumelhart, and the PDP Research Group. Journal of the Experimental Analysis of Behavior, 51, 399-416.

Donahoe, J. W., Palmer, D. C., & Burgos, J. E. (1997a). The S-R issue: Its status in behavior analysis and in Donahoe and Palmer's learning and complex behavior. Journal of the Experimental Analysis of Behavior, 67, 193-211.

Donahoe, J. W., Palmer, D. C., & Burgos, J. E. (1997b). The unit of selection: What do reinforcers reinforce? Journal of the Experimental Analysis of Behavior, 67, 259-273.

Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: The MIT Press.

Finkel, L. H., & Sajda, P. (1994). Constructing visual perception. American Scientist, 82, 224-237.

Frey, U. (1997). Cellular mechanisms of long-term potentiation: Late maintenance. In G. E. Stelmach & P. A. Vroon (Series Eds.) & J. W. Donahoe & V. P. Dorsel (Vol. Eds.), Advances in psychology: Vol. 121, Neural-network models of cognition: Biobehavioral foundations (pp. 105-128). Amsterdam: Elsevier.

Gluck, M. A., & Bower, G. H. (1988). From conditioning to category learning: an adaptive network model. Journal of Experimental Psychology: General, 117, 227-247.

Hayes, S. C. (1998). Resisting biologism. The Behavior Therapist, 21, 95-97.

Hayes, S. C., Barnes-Holmes, D., & Roche, B. (2001). Relational frame theory: A post-Skinnerian account of human language and cognition. New York: Kluwer.

Hebb, D. O. (1949). The organization of behavior. New York: Wiley.

Kalat, J. W. (2001). Biological psychology (7th ed.). Belmont, CA: Wadsworth/Thomson Learning.

Kandel, E. R. (1989). Genes, nerve cells, and the remembrance of things past. Journal of Neuropsychiatry, 1, 103-125.

Kandel, E. R. (1991). Cellular mechanisms of learning and the biological basis of individuality. In E. R. Kandel, J. H. Schwartz, & T. M. Jessell (Eds.). Principles of neural science (pp. 1009-1031). Norwalk, CT: Appleton & Lange.

Kimble, G. A. (1961). Hilgard and Marquis' conditioning and learning (2nd ed.). New York: Appleton-Century-Crofts.

Kolb, B., & Whishaw, I. Q. (1996). Human neuropsychology (4th ed.). San Francisco: W. H. Freeman.

Levine, D. S. (1991). Introduction to neural & cognitive modeling. Hillsdale, NJ: Erlbaum.

Mayr, E. (1982). The growth of biological thought: Diversity, evolution, and inheritance. Cambridge, MA: Harvard University Press.

McLeod, P., Plunkett, K., & Rolls, E. T. (1998). Introduction to connectionist modelling of cognitive processes. Oxford: Oxford University Press.

O'Reilly, R. C., & Munakata, Y. (2000). Computational explorations in cognitive neurosicence: Understanding the mind by simulating the brain. Cambridge, MA: The MIT Press.

Plaud, J. J., & Eifert, G. H. (Eds.). (1998). From behavior theory to behavior therapy. Boston: Allyn & Bacon.

Rolls, E. T., & Treves, A. (1998). Neural networks and brain function. Oxford: Oxford University Press.

Rumelhart, D. E., & McClelland, J. L. (Eds.). (1986). Parallel distributed processing. Vols. 1 & 2. Cambridge MA: The MIT Press.

Sach, O. (1985). The man who mistook his wife for a hat. New York: Summit Books.

Singer, W. (1997). Development and plasticity of neocortical processing architectures. In G. E. Stelmach & P. A. Vroon (Series Eds.) & J. W. Donahoe & V. P. Dorsel (Vol. Eds.), Advances in psychology: Vol. 121, Neural-network models of cognition: Biobehavioral foundations (pp. 142-159). Amsterdam: Elsevier.

Skinner, B. F. (1938). Behavior of organisms. New York: Appleton-Century-Crofts.

Skinner, B. F. (1963). Operant behavior. American Psychologist, 18, 503-515.

Skinner, B. F. (1966). The phylogeny and ontogeny of behavior. Science, 153, 1205-1213.

Skinner, B. F. (1974). About behaviorism. New York: Vintage Books.

Skinner, B. F. (1975). The shaping of phylogenetic behavior. The Journal of the Experimental Analysis of Behavior, 24, 117-120.

Skinner, B. F. (1977). Why I am not a cognitive psychologist. Behaviorism, 5, 1-10.

Skinner, B. F. (1981). Selection by consequences. Science, 213, 501-504.

Skinner, B. F. (1984a). Selection by consequences. The Behavioral and Brain Sciences, 7, 477-510.

Skinner, B. F. (1984b). The evolution of behavior. Journal of the Experimental Analysis of Behavior, 41, 217-221.

Skinner, B. F. (1989). The origins of cognitive thought. American Psychologist, 44, 13-18.

Smith, T. L. (1983). Skinner's environmentalism: The analogy with natural selection. Behaviorism, 11, 133-153.

Smith, T. L. (1984). Biology as allegory: A review of Elliott Sober's The Nature of Selection. Journal of The Experimental Analysis of Behavior, 46, 105-112.

Spitzer, M. (1999). The mind within the net: Models of learning, thinking, and acting. Cambridge, MA: The MIT Press.

Staats, A. W. (1983). Psychology's crisis of disunity: Philosophy and method for a unified science. (pp. 120-121) New York: Praeger Publishers.

Tang, Y. P., Shimizu, E., Duba, G. R., Hampon, C., Kerchner, G. A., Zhuo, M., Liu, G., & Tsien, J. Z. (1999). Genetic enhancement of learning and memory in mice. Nature, 401, 63-69.

Thagard, P. (2000). Coherence in thought and action. Cambridge, MA: The MIT Press.

Thorndike, E. L. (1898). Animal intelligence: An experimental study of the associative processes in animals. Psychological Review Monograph Supplement, 2 (No. 8).

Thorndike, E. L. (1933). A proof of the law of effect. Science, 77, 173-175.

Tryon, W. W. (1993). Neural Networks: II. Unified Learning Theory and Behavioral Psychotherapy. Clinical Psychology Review, 13, 353-371.

Tryon, W. W. (1995a). Resolving the cognitive behavioral controversy. The Behavior Therapist, 18, 83-86.

Tryon, W. W. (1995b). Synthesizing animal and human research via neural network learning theory. Journal of Behavior Therapy and Experimental Psychiatry, 26, 303-312.

Tryon, W. W. (1996). Yes--Neural network learning theory can resolve the behavioral-cognitive controversy. The Behavior Therapist, 19, 70, 72-73.

Tryon, W. W. (1998). A neural network explanation of posttraumatic stress disorder. Journal of Anxiety Disorders, 12, 373-385.

Tryon, W. W. (1999). A bidirectional associative memory explanation of posttraumatic stress disorder. Clinical Psychology Review, 19, 789-818.

Tryon, W. W. (2000a). Behavior therapy as applied learning theory. The Behavior Therapist, 23, 131-134.

Tryon, W. W. (2000b). Neural network learning theory integrates behavior therapy and behavior genetics. The Behavior Therapist, 23, 3-4,6, 8-9, 19-20, 22.

Wasserman, P. D. (1989). Neural computing: Theory and practice. NY: Van Nostrand-Reinhold.

Wilson, E.O. (1998). Consilience: The unity of knowledge. New York: Knopf.

Woods, P. J. (1974). A taxonomy of instrumental conditioning. American Psychologist, 29, 584-597.

(10) E. L. Thorndike's (1898) theory of animal intelligence was called connectionism because he described teamed associations between stimuli and responses as bonds or connections. Making or breaking habits strengthened or weakened connecting bonds between sense impressions and actions. His spread of effect papers (Thorndike, 1933) described the beginnings of a network theory. He argued that reinforcement automatically influenced the connection at hand but it also acted on temporarily adjacent connections occurring just before or just after the time when reinforcement was given.

(11) Behaviorists do not have preferred plausible proximal causal mechanisms that they feet are superior to psychological and/or biological mechanisms but rather they reject all efforts as closing the explanatory gap. Hence, behaviorists can only presume rather than explain the ability of contingent consequences to reshape response variant distributions. The question of why reinforcers work as they do remains unaddressed and unanswered. Behaviorists suppose that understanding how selection works within the organism is not their responsibility and they refuse to devote time and resources to these questions. Assertions that neuroscience can only confirm functional relationships established by behavior analysis implies openness to biological explanations as compatible with and extending functional analytic explanations but resistance remains high to including contemporary neuroscience into behavioral analytic explanations. Even the need or desirability of extending functional analytic explanations in any way remains unacknowledged.

(12) It is notable that PDP connectionism criticizes symbol manipulating cognitive psychology on the same grounds that symbol manipulating cognitive psychology criticizes radical behaviorism; namely that their explanations entirely entail functional statements that lack plausible proximal causal mechanisms to account for why and how these functions occur as they do.

(13) No personal criticism is intended by this seemingly harsh remark. We clearly have different views and I could not find a more delicate way of presenting the truth as I see it.

Warren W. Tryon

Fordham University
COPYRIGHT 2002 Behavior Analyst Online
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Tryon, Warren W.
Publication:The Behavior Analyst Today
Date:Jan 1, 2002
Words:9201
Previous Article:Different effects of individual and small group monetary incentives on high performance.
Next Article:Creation of an Applied Behavior Analysis Council for Accreditation (ABACA).
Topics:


Related Articles
Pessimism linked to poor health.
Explanatory and predictive roles of inner causes: a reply to Overskeid.
Explanatory Style as a Predictor of College Performance In Students with Physical Disabilities.
Explanatory styles: tips for baseball & softball coaches.
Introductory Econometrics: A Modern Approach, 3d ed.
Accentuate the positive: the relationship between positive explanatory style and academic achievement of prospective elementary teachers.
Comparative politics; the quest for theory and explanation.
Multilevel analysis for applied research; it's just regression!
On differences between explanatory and pragmatic clinical trials.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters