Comparing elemental and configural associative theories in human causal learning: a case for attention

In two causal learning experiments with human participants, the authors compared various associative theories that assumed either elemental (unique cue, modified unique cue, replaced elements model, and Harris' model) or configural processing of stimuli (Pearce's theory and a modification of it). The authors used modified patterning problems initially suggested by Redhead and Pearce (1995). Predictions for all theories were generated by computer simulations. Both configural theories and the unique cue approach failed to account for the observations. The replaced elements model was able to account for part of the data, but only if the replacement parameters could vary across discrimination problems. The Harris model and the modified unique cue approach, assuming that the salience of stimuli decreases with an increasing number of stimuli in a compound, successfully accounted for all of our data. This success implies that attentional factors should be explicitly taken into account in associative learning theory.     

Associative learning and elemental representation: II. Generalization and discrimination

This paper follows on from an earlier companion paper (McLaren & Mackintosh, 2000), in which we further developed the elemental associative theory put forward in McLaren, Kaye, and Mackintosh (1989). Here, we begin by explicating the idea that stimuli can be represented as patterns of activation distributed across a set of units and that different stimuli activate partially overlapping sets (the degree of overlap being proportional to the similarity of the stimuli). A consequence of this view is that the overall level of activity of some of the units representing a stimulus may be dependent on the nature of the other stimuli present at the same time. This allows an elemental analysis in which provision for the representation of configurations of stimuli is made. A selective review of studies of generalization and discrimination learning, including peak shift, transfer along a continuum, configural discrimination, and summation, suggests that the principles embodied in this class of theory deserve careful consideration and will form part of any successful model of associative learning in humans or animals. There are some phenomena that require an elemental/associative explanation.     

Generalization decrement in human contingency learning

An association between a cue and an outcome will generalize to a similar novel cue to some extent, but not completely. Learning theorists refer to the discrepancy between responding elicited by the original cue and the novel cue as a generalization decrement. Two experiments used a contingency learning task with human participants to compare the size of a generalization decrement between configurations of cues that were altered by adding or subtracting compositional elements. The results suggest that adding elements to a configuration can produce a generalization decrement, but removing elements produces a more robust generalization decrement. Furthermore, the generalization decrement caused by adding elements was not likely to be caused by competing orienting responses. The results are used to contrast Pearce's (1987, 1994) and Wagner's (2003) models of stimulus generalization.     

Generalization decrement in human contingency learning

An association between a cue and an outcome will generalize to a similar novel cue to some extent, but not completely. Learning theorists refer to the discrepancy between responding elicited by the original cue and the novel cue as a generalization decrement. Two experiments used a contingency learning task with human participants to compare the size of a generalization decrement between configurations of cues that were altered by adding or subtracting compositional elements. The results suggest that adding elements to a configuration can produce a generalization decrement, but removing elements produces a more robust generalization decrement. Furthermore, the generalization decrement caused by adding elements was not likely to be caused by competing orienting responses. The results are used to contrast Pearce's (1987, 1994) and Wagner's (2003) models of stimulus generalization.     

Generalization gradients in human predictive learning: Effects of discrimination training and within-subjects testing

Generalization gradients have been investigated widely in animal conditioning experiments, but much less so in human predictive learning tasks. Here, we apply the experimental design of a recent study on conditioned fear generalization in humans (Lissek et al., 2008) to a predictive learning task, and examine the effects of a number of relevant procedural parameters drawn from the generalization literature in animal conditioning. Experiment 1 shows that prior discrimination learning and steady-state testing procedures sharpen the gradient; Experiment 2 shows that within-subjects testing of the range of generalization stimuli also sharpens the gradient. In addition, Experiment 2 shows that, in case of very flat initial generalization, an orderly gradient can reveal itself through differential rates of extinction learning. Finally, Experiment 2 also evidenced an orderly gradient of generalization-of-extinction. These results suggest that discrimination processes have an important effect on the generalization of predictive learning in humans, and highlight behavioral analogies between animal conditioning and human predictive learning.     

Overshadowed cues have reduced ability to retroactively interfere with other cues

The present series of experiments explores the interaction between retroactive interference and cue competition in human contingency learning. The results of two experiments show that a cue that has been exposed to a cue competition treatment (overshadowing) loses part of its ability to retroactively interfere with responding to a different cue that was paired with the same outcome. These results pose problems for associative models of contingency learning and are also difficult to explain in terms of current theories of causal reasoning. Additionally, it is proposed that in light of the interaction between interference and cue competition, interference could be used as an indirect measure for the study of cue competition effects.     

Conceptual similarity promotes generalization of higher order fear learning

We tested the hypothesis that conceptual similarity promotes generalization of conditioned fear. Using a sensory preconditioning procedure, three groups of subjects learned an association between two cues that were conceptually similar, unrelated, or mismatched. Next, one of the cues was paired with a shock. The other cue was then reintroduced to test for fear generalization, as measured by the skin conductance response. Results showed enhanced fear generalization that correlated with trait anxiety levels in the group that learned an association between conceptually similar stimuli. These findings suggest that conceptual representations of conditional stimuli influence human fear learning processes.     

Elemental and configural processing of novel cues in deterministic and probabilistic tasks

Three experiments examined whether transfer of past learning depends on how well the original discrimination is learned. To vary terminal learning levels, cues were either perfect (deterministic) or imperfect (probabilistic) predictors of a trial’s outcome. Participants in Experiment 1 acquired a configural, an elemental or a control discrimination. Tests for generalization showed that past learning influenced the processing of new compounds formed from elements of the original discrimination, especially so when the original discrimination was deterministic. Similar results were found in Experiment 2 when the test stimuli were elements derived from novel compounds presented after the original discrimination was acquired. Experiment 3 used filler trials to equate learning in the probabilistic and deterministic tasks, and demonstrated that final levels of learning, rather than task per se, was the critical variable mediating transfer. Implications for rule-based and associative theories are discussed.     

Learned changes in stimulus representations (a personal history)

Almost 40 years ago I began what turned out to be a programme of research on the way in which experience can change the effectiveness of the events used as stimuli in standard associative learning procedures. In this personal history I will describe my early (failed) attempts to find evidence for the acquired distinctiveness of cues, and my conclusion that experience tends to reduce, not enhance the associability of stimuli. I then go on to describe my attempts to square this conclusion with the stubborn empirical fact that, in some circumstances, pretraining with (or preexposure to) stimuli, can facilitate subsequent discrimination between them. I describe experiments (conducted mostly with rats as the subjects) showing how some of these effects can be explained in associative terms. Others, however, seemed to demand an explanation in terms of a new learning process that modulates the effective salience of stimuli. I go on to describe attempts to specify the nature of this process, and (bringing the story up to date) to describe recent experiments investigating the effects of salience modulation in human perceptual learning.     

Cue selection in verbal discrimination learning of retarded subjects.

A series of experiments was conducted to examine cue function in trigram verbal discrimination learning by retarded subjects. The issue was to determine the factors that control attention in this type of learning situation. The two variables of chief interest were trigram meaningfulness and reinforcement history. The major general conclusion was that retarded subjects employ complex cue selection strategies in solving a verbal discrimination involving compound verbal stimuli. Major findings were as follows: (a) Retarded subjects exhibit a response bias in favor of words over nonsense trigrams; (b) relatively little active cue selection based on meaningfulness was observed; (c) a compound stimulus discrimination was more difficult to learn than a discrimination involving single verbal stimuli; (d) compound and single stimuli may be processed differently in original discrimination learning; (e) cue position probably controls attention in compound discrimination learning; and (f) a frequency theory of verbal discrimination learning is supported by these data.     

Different parameters support generalization and discrimination learning in Drosophila at the flight simulator

We have used a genetically tractable model system, the fruit fly Drosophila melanogaster to study the interdependence between sensory processing and associative processing on learning performance. We investigated the influence of variations in the physical and predictive properties of color stimuli in several different operant-conditioning procedures on the subsequent learning performance. These procedures included context and stimulus generalization as well as color, compound, and conditional discrimination (colors and patterns). A surprisingly complex dependence of the learning performance on the colors' physical and predictive properties emerged, which was clarified by taking into account the fly-subjective perception of the color stimuli. Based on estimates of the stimuli's color and brightness values, we propose that the different tasks are supported by different parameters of the color stimuli; generalization occurs only if the chromaticity is sufficiently similar, whereas discrimination learning relies on brightness differences.     

A neural model of hippocampal–striatal interactions in associative learning and transfer generalization in various neurological and psychiatric patients

Building on our previous neurocomputational models of basal ganglia and hippocampal region function (and their modulation by dopamine and acetylcholine, respectively), we show here how an integration of these models can inform our understanding of the interaction between the basal ganglia and hippocampal region in associative learning and transfer generalization across various patient populations. As a common test bed for exploring interactions between these brain regions and neuromodulators, we focus on the acquired equivalence task, an associative learning paradigm in which stimuli that have been associated with the same outcome acquire a functional similarity such that subsequent generalization between these stimuli increases. This task has been used to test cognitive dysfunction in various patient populations with damages to the hippocampal region and basal ganglia, including studies of patients with Parkinson’s disease (PD), schizophrenia, basal forebrain amnesia, and hippocampal atrophy. Simulation results show that damage to the hippocampal region—as in patients with hippocampal atrophy (HA), hypoxia, mild Alzheimer’s (AD), or schizophrenia—leads to intact associative learning but impaired transfer generalization performance. Moreover, the model demonstrates how PD and anterior communicating artery (ACoA) aneurysm—two very different brain disorders that affect different neural mechanisms—can have similar effects on acquired equivalence performance. In particular, the model shows that simulating a loss of dopamine function in the basal ganglia module (as in PD) leads to slow acquisition learning but intact transfer generalization. Similarly, the model shows that simulating the loss of acetylcholine in the hippocampal region (as in ACoA aneurysm) also results in slower acquisition learning. We argue from this that changes in associative learning of stimulus–action pathways (in the basal ganglia) or changes in the learning of stimulus representations (in the hippocampal region) can have similar functional effects.     

The fate of redundant cues during blocking and a simple discrimination

In each of three experiments animals received blocking, A+ AX+, in which food was always presented after one stimulus, A, that was occasionally accompanied by another stimulus, X. They also received a simple discrimination, AX+ BX-, in which the presence and absence of food was signaled by two compounds that contained one unique cue, A or B, and one common cue, X. In each of these designs, X can be said to be redundant relative to A as a signal for food. Test trials at the end of training revealed that responding during X was stronger after blocking than after the simple discrimination. These results contradict predictions from theories of learning that assume changes in associative strength of a stimulus are determined by a global error term based on the outcome predicted by all the stimuli that are present for a conditioning trial. The results are interpreted, instead, by assuming either that animals store a memory of every trial to which they have been exposed, or that learning is governed by an error term based on the significance of individual stimuli.     

Inescapable shock and attention to internal versus external cues in a water discrimination escape task

In these experiments we examined discrimination learning in a water escape task following exposure to escapable, yoked inescapable, or no electric shock. Inescapable shock did not have an effect on swim speeds in any of the experiments. Inescapable shock interfered with the acquisition of a position (left-right) discrimination when an irrelevant brightness cue (black and white stimuli) was present. However, inescapable shock did not affect the acquisition of the position discrimination when the irrelevant brightness cue was removed. Inescapably shocked subjects showed facilitated learning relative to escapably shocked and nonshocked subjects when the brightness cue was included as a relevant cue. These data may resolve discrepancies between studies that did, and did not, find inescapable shock to interfere with the acquisition of discriminations. Moreover, they point to attentional processes as one locus of the cognitive changes produced by inescapable shock and suggest that exposure to inescapable shock biases attention away from "internal" response-related cues toward "external" cues.     

Inhibition of return for faces

Inhibition of return (IOR) refers to slower reaction times when a target appears unpredictably in the same location as a preceding cue, rather than in a different location. In the present study, frontal images of human faces were presented intact as face configurations, were rearranged to produce scrambled-face configurations, or were pixilated and randomized to produce nonface configurations. In an orienting paradigm designed to elicit IOR, face and scrambled-face stimuli were used as cues (Experiment 1), as targets (Experiment 2), or along with pixilated nonface stimuli as both cues and targets (Experiment 3). The magnitude of IOR for subsequent localization targets was unaffected by cue configuration. Likewise, the magnitude of IOR was unaffected by target configuration. These results suggest that IOR is a "blind" mechanism that is unaffected by the mere occurrence of biologically relevant cue and target stimuli.     

The hide-and-seek of retrospective revaluation: recovery from blocking is context dependent in human causal learning

In a typical blocking procedure, pairings of a compound consisting of 2 stimuli, A and X, with the outcome are preceded by pairings of only A with the outcome (i.e., A+ then AX+). This procedure is known to diminish responding to the target cue (X) relative to a control group that does not receive the preceding training with blocking cue A. We report 2 experiments that investigated the effect of extinguishing a blocking cue on responding to the target cue in a human causal learning paradigm (i.e., A+ and AX+ training followed by A- training). The results indicate that extinguishing a blocking cue increases conditioned responding to the target cue. Moreover, this increase appears to be context dependent, such that increased responding to the target is limited to the context in which extinction of the blocking cue took place. We discuss these findings in the light of associative and propositional learning theories.     

Asymmetries in cue competition in forward and backward blocking designs: Further evidence for causal model theory

A hallmark feature of elemental associative learning theories is that multiple cues compete for associative strength when presented with an outcome. Cue competition effects have been observed in humans, both in forward and in backward blocking procedures (e.g., Shanks, 1985) and are often interpreted as evidence for an associative account of human causal learning (e.g., Shanks & Dickinson, 1987). Waldmann and Holyoak (1992), however, demonstrated that cue competition only occurs in predictive, and not diagnostic, learning paradigms. While unexplainable from an associative perspective, this asymmetry readily follows from structural considerations of causal model theory. In this paper, we show that causal models determine the extent of cue competition not only in forward but also in backward blocking designs. Implications for associative and inferential accounts of causal learning are discussed.     

Peak shift revisited: a test of alternative interpretations

In Experiment 1, 2 groups of human subjects were trained to respond to 1 of 2 light intensity stimuli, S2 or S4, and then were tested for generalization with a randomized series of increasing values from S1 to S11. Both groups, including the group trained to respond to dimmer value, showed peak shifts to a brighter more centrally located test stimulus. In Experiment 2, which used line angle stimuli, both the size of the difference between S+ and S- and the range of test stimuli that extended beyond S+ were varied. The larger the S(+)-S- separation and the larger the range, the greater was the peak shift obtained. In Experiment 3, training involved an S- (line angle) surrounded by 2 S+ values with testing symmetrical about the training values and covering either a narrow or a wide range. The wide range produced greater peak shifts in both directions from S-. All 3 experiments support an adaptation-level interpretation of intradimensional discrimination learning and generalization test performance in human subjects. Related work with animals suggests the presence of similar processes.     

Causal Learning: Association Versus Computation

Causal learning enables humans and other animals not only to predict important events or outcomes, but also to control their occurrence in the service of needs and desires. Computational theories assume that causal judgments are based on an estimate of the contingency between a causal cue and an outcome. However, human causal learning exhibits many of the characteristics of the associative learning processes thought to underlie animal conditioning. One problem for associative theory arises from the finding that judgments of the causal power of a cue can be revalued retrospectively after learning episodes when that cue is not present. However, if retrieved representations of cues can support learning, retrospective revaluation is anticipated by modified versions of standard associative theories.     

A dissociation between causal judgment and outcome recall

It has been suggested that causal learning in humans is similar to Pavlovian conditioning in animals. According to this view, judgments of cause reflect the degree to which an association exists between the cause and the effect. Inferential accounts, by contrast, suggest that causal judgments are reasoning based rather than associative in nature. We used a direct measure of associative strength, identification of the outcome with which a cause was paired (cued recall), to see whether associative strength translated directly into causal ratings. Causal compounds AB+ and CD+ were intermixed with A+ and C- training. Cued-recall performance was better for cue B than for cue D; thus, associative strength was inherited by cue B from the strongly associated cue A (augmentation). However, the reverse was observed on the causal judgment measure: Cue B was judged to be less causal than D (cue competition). These results support an inferential over an associative account of causal judgments.