Equilibria of the Rescorla-Wagner model

The Rescorla-Wagner model has been a leading theory of animal causal induction for nearly 30 years, and human causal induction for the past 15 years. Recent theories (especially Psychol. Rev. 104 (1997) 367) have provided alternative explanations of how people draw causal conclusions from covariational data. However, theoretical attempts to compare the Rescorla-Wagner model with more recent models have been hampered by the fact that the Rescorla-Wagner model is an algorithmic theory, while the more recent theories are all computational. This paper provides a detailed derivation of the long-run behavior of the Rescorla-Wagner model under a wide range of parameters and experimental setups, so that the model can be compared with computational theories. It also shows that the model agrees with competing theories on a wider range of cases than had previously been thought. The paper concludes by showing how recently suggested modifications of the Rescorla-Wagner model impact the long-run behavior of the model.     

From conditioning to category learning: an adaptive network model

We used adaptive network theory to extend the Rescorla-Wagner (1972) least mean squares (LMS) model of associative learning to phenomena of human learning and judgment. In three experiments subjects learned to categorize hypothetical patients with particular symptom patterns as having certain diseases. When one disease is far more likely than another, the model predicts that subjects will substantially overestimate the diagnosticity of the more valid symptom for the rare disease. The results of Experiments 1 and 2 provide clear support for this prediction in contradistinction to predictions from probability matching, exemplar retrieval, or simple prototype learning models. Experiment 3 contrasted the adaptive network model with one predicting pattern-probability matching when patients always had four symptoms (chosen from four opponent pairs) rather than the presence or absence of each of four symptoms, as in Experiment 1. The results again support the Rescorla-Wagner LMS learning rule as embedded within an adaptive network model.     

Blocking and backward blocking involve learned inattention

Four experiments examine blocking of associative learning by human participants in a disease diagnosis procedure. The results indicate that after a cue is blocked, subsequent learning about the cue is attenuated. This attenuated learning after blocking is obtained for both standard blocking and for backward blocking. Attenuated learning after blocking cannot be accounted for by theories such as the Rescorla-Wagner model that rely on lack of learning about a redundant cue, nor can it be accounted for by extensions of the Rescorla-Wagner model designed to address backward blocking that encode absent cues with negative values. The results are predicted by the hypothesis that people learn not to attend to the blocked cue.     

Teaching the Rescorla-Wagner model using STELLA-II

Although the Rescorla-Wagner model is widely taught in learning courses, most textbooks present a simplified version that does not contain all of its parameters. On the basis of Rescorla and Wagner 1972) and the Turbo PASCAL program by Lachnit, Schneider, Lipp, and Kimme! (1988), we created a simulation using STELLA-II, an icon-based program by High Performance Systems, Inc. We argue that this simulation facilitates teaching the complete Rescorla-Wagner model. Using trans-switching (Kimmel & Ray, 1978; Lachnit, 1986; Murrin & Kimmel, 1986) as an example, we demonstrate the effectiveness of the STELLA-II version of the Rescorla-Wagner model as a heuristic tool.     

A way to install conditioned modulation in the Rescorla-Wagner Axiom: The amplifier model of Pavlovian conditioning (AMP)

Although the Rescorla-Wagner axiom has been successful in explaining and predicting many phenomena in Pavlovian conditioning, it offers no explanation for the ability of stimuli to modulate conditioned responding to other stimuli (i.e., the occasion-setting effect). This article provides a mathematical model for Pavlovian conditioned modulation by adding simple rules to the Rescorla-Wagner axiom. In spite of its rudimentary stage, the proposed model can explain many of the basic findings of modulatory learning and suggests some new areas of research.     

Marmosets as a next-generation model of comparative cognition

Common marmosets have traditionally been used in fields such as comparative psychology, cognitive neuroscience, and ethology, and as animal models of human disease research because of their relatively small body size, high reproductive rate, similar breeding system characteristics to those of humans, high dependency on vocal communication, and behavioral effects from drug administration that resemble those of humans. Although the animal has proved to be quite advantageous in animal model research in comparison to rodents, and has specific ecological and evolutionary characteristics that are worthy of exploration, few attempts have been made until recently to combine both types of approach. Thus, supported by neuroscientific methodologies that have recently been developed using this species, the authors suggest that the evolutionary origins of cognitive and social functions common to or differentiated in primates, could be uncovered through research that explores the behavioral and neural bases of cognition in the marmosets.     

Self-starvation in the rat: running versus eating

Rats given the combination of unrestricted access to an activity wheel and restricted access to food can lose weight to the extent that they will die unless removed from these conditions. Although this has been known for forty years, why this happens has remained unclear. The phenomenon is paradoxical in that one might expect such rats to eat more as their weight decreases, but in fact they eat less than resting controls. This lecture first examines some of the factors than influence whether self-starvation will occur, such as age, time of food access, type of food and ambient temperature. It then compares competing explanations such as circadian adaptation, thermo-regulation and food aversion learning. As so often in psychology, it turns out that self-starvation results from a combination of many separate factors. The general implications of this research are examined, including whether it provides a useful animal model for human anorexia nervosa.     

Cerebellar Substrates for Error Correction in Motor Conditioning

The authors evaluate a mapping of Rescorla and Wagner's (1972) behavioral model of classical conditioning onto the cerebellar substrates for motor reflex learning and illustrate how the limitations of the Rescorla-Wagner model are just as useful as its successes for guiding the development of new psychobiological theories of learning. They postulate that the inhibitory pathway that returns conditioned response information from the cerebellar interpositus nucleus back to the inferior olive is the neural basis for the error correction learning proposed by Rescorla and Wagner (Gluck, Myers, & Thompson, 1994; Thompson, 1986). The authors' cerebellar model expects that behavioral processes described by the Rescorla-Wagner model will be localized within the cerebellum and related brain stem structures, whereas behavioral processes beyond the scope of the Rescorla-Wagner model will depend on extracerebellar structures such as the hippocampus and related cortical regions. Simulations presented here support both implications. Several novel implications of the authors' cerebellar error-correcting model are described including a recent empirical study by Kim, Krupa, and Thompson (1998), who verified that suppressing the putative error correction pathway should interfere with the Kamin (1969) blocking effect, a behavioral manifestation of error correction learning. The authors also discuss the model's implications for understanding the limits of cerebellar contributions to associative learning and how this informs our understanding of hippocampal function in conditioning. This leads to a more integrative view of the neural substrates of conditioning in which the authors' real-time circuit-level model of the cerebellum can be viewed as a generalization of the long-term memory module of Gluck and Myers' (1993) trial-level theory of cerebellar-hippocampal interaction in motor conditioning.     

Genetics and educational psychology

Background: Molecular genetics, one of the most energetic and exciting areas of science, is slowly but surely coming to educational psychology. Aims: We review recent molecular genetic research on learning disabilities as a sign of things to come in educational psychology. We also consider some misconceptions about genetics that have slowed the acceptance of genetics in educational psychology. Samples: Diverse samples of children with learning disabilities have been studied, primarily in the UK and the USA. Methods: Linkage analysis can detect genes that have large effects on learning disabilities. Association analysis can detect genes of much smaller effect size, which is important because common disorders such as learning disabilities are likely to be influenced by many genes as well as by many environmental factors. Results: For reading disability, replicated linkages have been identified on chromosomes 6, 15 and 18. A gene responsible for a rare type of language impairment has recently been identified. For common language impairment, linkages on chromosomes 16 and 19 have recently been reported. More than 200 genetic disorders, most extremely rare, include mental retardation among their symptoms, and chromosomal abnormalities are a major cause of mental retardation. Conclusions: Although finding specific genes associated with learning disabilities is unlikely to have much of a direct application for teachers in the classroom, such findings will have far‐reaching implications for diagnosis, treatment and prevention of learning disabilities and for research in educational psychology. Educational psychology has been slower to accept evidence for the importance of genetics than other areas of psychology in part because of misconceptions about what it means to say that genetics is important for common complex disorders such as learning disabilities.     

Action unit classification using active appearance models and conditional random fields

In this paper, we investigate to what extent modern computer vision and machine learning techniques can assist social psychology research by automatically recognizing facial expressions. To this end, we develop a system that automatically recognizes the action units defined in the facial action coding system (FACS). The system uses a sophisticated deformable template, which is known as the active appearance model, to model the appearance of faces. The model is used to identify the location of facial feature points, as well as to extract features from the face that are indicative of the action unit states. The detection of the presence of action units is performed by a time series classification model, the linear-chain conditional random field. We evaluate the performance of our system in experiments on a large data set of videos with posed and natural facial expressions. In the experiments, we compare the action units detected by our approach with annotations made by human FACS annotators. Our results show that the agreement between the system and human FACS annotators is higher than 90% and underlines the potential of modern computer vision and machine learning techniques to social psychology research. We conclude with some suggestions on how systems like ours can play an important role in research on social signals.     

Action unit classification using active appearance models and conditional random fields

In this paper, we investigate to what extent modern computer vision and machine learning techniques can assist social psychology research by automatically recognizing facial expressions. To this end, we develop a system that automatically recognizes the action units defined in the facial action coding system (FACS). The system uses a sophisticated deformable template, which is known as the active appearance model, to model the appearance of faces. The model is used to identify the location of facial feature points, as well as to extract features from the face that are indicative of the action unit states. The detection of the presence of action units is performed by a time series classification model, the linear-chain conditional random field. We evaluate the performance of our system in experiments on a large data set of videos with posed and natural facial expressions. In the experiments, we compare the action units detected by our approach with annotations made by human FACS annotators. Our results show that the agreement between the system and human FACS annotators is higher than 90% and underlines the potential of modern computer vision and machine learning techniques to social psychology research. We conclude with some suggestions on how systems like ours can play an important role in research on social signals.     

Does the type of judgement required modulate cue competition?

According to the comparator process hypothesis (Matute, Arcediano, & Miller, 1996), cue competition in the learning of between-events relationships arises if the judgement required involves a comparison between the probability of the outcome given the target cue and the probability of the outcome given the competing cue. Alternatively, other associative accounts (the Rescorla-Wagner model: Rescorla &Wagner, 1972) conceive cue competition as a learning deficit affecting the target cue-outcome association. Consequently, the comparator process hypothesis predicts that cue competition occurs in inference judgements but not in contiguity ones, for only the first type of judgement implicitly involves such a comparison. On the other hand, the Rescorla-Wagner model predicts cue competition in both inference and contiguity judgements, because it establishes no relevant role for the type of judgement in producing cue competition. In Experiments 1 and 2 we manipulated the relative validity of cues and the type of question (inference vs. contiguity) in a predictive learning task. In both experiments we found a cue competition effect, but no interaction between the relative validity of cues and the type of question, suggesting that the Rescorla-Wagner theory suffices to explain cue competition.     

How the challenge of explaining learning influenced the origins and development of John B. Watson's behaviorism

Before he invented behaviorism, John B. Watson considered learning one of the most important topics in psychology. Watson conducted excellent empirical research on animal learning. He developed behaviorism in part to promote research and elevate the status of learning in psychology. Watson was much less successful in the adequacy and originality of the mechanisms he proposed to explain learning. By assimilating the method of classical conditioning and adopting Pavlov's theory of stimulus substitution, Watson linked behaviorism with a new method that could compete with both Titchener's method of introspection and Freud's methods of psychoanalysis. Watson's interest in explaining psychopathology led to the discovery of conditioned emotional responses and a behavioristic explanation for the learning of phobic behavior. Watson established learning as a central topic for basic research and application in American psychology.     

A review of recent developments in research and theories on human contingency learning

Over the past 20 years, human contingency learning has resurfaced as an important topic within experimental psychology. This renewed interest was sparked mainly by the proposal that associative models of Pavlovian conditioning might also apply to human contingency learning--a proposal that has led to many new empirical findings and theoretical developments. We provide a brief review of these recent developments and try to point to issues that need to be addressed in future research.     

Associative and Normative Accounts of Negative Transfer

Subjects were given a prediction task in which they had to learn that one cue, P (positive), was followed by the outcome, and a compound of P and another cue, N (negative), was not followed by the outcome. Next, N was tested in compound with a transfer cue, T, which had signalled the outcome but had never been compounded with N. Experiment 1 confirmed an important assumption of the Rescorla-Wagner model (Wagner & Rescorla, 1972) that negation of T should depend on the specific P cue compound with N being positively contingent. Experiments 2 and 3 confirmed the model's prediction that no decrement in negative transfer should be observed following postlearning devaluation of P. Unfortunately, the model did not anticipate that a large proportion of devaluation trials relative to learning trials would attenuate negative transfer (Experiment 4), nor did it predict that negative transfer would occur when P was neutral during the learning stage and was only later made positive (Experiment 3). The results can be accommodated by the Rescorla-Wagner model if one assumes that absent cues have their associative strengths altered.     

Backward Blocking and Recovery from Overshadowing in Human Causal Judgement: The Role of Within-compound Associations

We replicated and extended a project by Dickinson and Burke (1996) that concerned human causal judgement. In a medical diagnostic setting, college students' ratings of the causal efficacy of target cues showed retrospective revaluation: relative to a proper control condition, ratings of target cues both increased ('recovery from overshadowing') and decreased ('backward blocking') during a second stage of training in which competing cues, but not target cues, were presented. These changes in causal judgements were exhibited only by subjects who had learned which target and competing cues were paired with one another during the first stage of training. These results cannot be explained by the Rescorla-Wagner (1972) model of associative learning, but they can be explained by the revised model of Van Hamme and Wasserman (1994); the revised model assigns non-zero salience to nonpresented target stimuli whose memories or representations are retrieved by competing stimuli that had previously been paired with those target stimuli.     

Comparative cognition in the 1930s

According to the received view of the history of psychology, behaviorism so dominated psychology prior to the 1960s that there was little research in animal cognition. A review of the research on animal cognition during the 1930s reveals a rich literature dealing with such topics as insight, reasoning, tool use, delay problems, oddity learning, abstraction, spatial cognition, and problem solving, among others. Material on “higher processes” or a related topic was prominent in the textbooks of the period. Tracing academic lineages reveals such teachers as Harvey Carr, Robert M. Yerkes, and Edward C. Tolman as sources of this interest. The alleged hegemony of strict behavioristic psychology, interpreted as excluding research on animal cognition, requires revision. Some possible reasons for this neglect are suggested.     

Learning: Association or Computation? Introduction to a Special Section

The fundamental nature of learning is a central problem in psychology. Traditionally, psychologists have assumed that learning must involve the formation of associations. Early last century, the pioneering work of Pavlov on conditioned learning in animals seemed to put this assumption beyond doubt. More recently, many psychologists came to believe that a different kind of process must underlie complex learning, such as language learning in humans, and that this process must be described as computational rather than associative. Whether complex human learning is associative or computational continues to be a subject of intense research. The articles in this Special Section turn this debate on its head by asking whether simple animal learning is associative or computational. Surprisingly, the question is still very much open, and excitingly, it appears quite tractable.     

Human science and counselling psychology: Closing the gap between research and practice

In this article, the case is made that human science-or ‘qualitative research’ as it is now popularly known–promises to be a better way of closing the gap between research and practice in counselling psychology than has been provided by the natural science approach to the discipline as instantiated in the Boulder model of the scientist-practitioner. Like the practice of counselling psychology, human science focuses on subjectivity and stresses the achievement of an understanding as opposed to the demonstration of truth; it stresses collaboration with participants rather than a subject-object dualism; and it emphasizes holism in contrast with fragmentation. Some practicalities entailed in the conduct of human science in counselling psychology are discussed.     

ALTSim: A MATLAB simulator for current associative learning theories

ALTSim is a MATLAB-based simulator of several associative learning models, including Pearce’s configural model, the extended configural model, the Rescorla-Wagner model, the unique cue hypothesis, the modified unique cue hypothesis, the replaced elements model, and Harris’s elemental model. It allows for specifying all relevant parameters, as well as exact stimulus sequences by graphical user interfaces. It is an easy-to-use tool that facilitates evaluating and comparing the featured associative learning models. ALTSim is available free of charge from www.staff.uni-marburg.de/~lachnit/ALTSim/.