Summation effects in human learning: evidence from patterning discriminations in goal-tracking

Participants in two human goal-tracking experiments were simultaneously trained with negative patterning (NP) and positive patterning (PP) discriminations (A+, B+, AB–, C–, D–, CD+). Both elemental and configural models of associative learning predict a PP advantage, such that NP is solved less readily than PP. However, elemental models like the unique cue approach additionally predict responding in AB– trials to be initially stronger than that in A+ and B+ trials due to summation of associative strength. Both experiments revealed a PP advantage and a strong summation effect in AB– trials in the first half of the experiments, irrespective of whether the same US was used for both discriminations (Experiment 1) or two different USs (Experiment 2). We discuss that the correct predictions of the unique cue approach are based on its assumptions of non-normalized and context-independent stimulus processing rather than elemental processing per se.     

Conditions favoring superconditioning of irrelevant conditioned stimuli

Three appetitive conditioning experiments with rats found partial learning of complex XA+, XB+, XAB- (+ stands for reinforced; - stands for unreinforced) negative patterning discriminations with intermixed A+ and B+ trials (Experiment 1). AB+ trials (Experiment 2), and A+, B+, and AB+ trials (Experiment 3). In all experiments, differential responding emerged more slowly during the learning of the negative patterning discriminations than during learning of the XA+, XB+, XC- control discriminations. Additionally, the negative patterning groups responded more to X than to a separately reinforced Y on unreinforced test trials: thus, X derived superexcitatory properties. This pattern was reversed in the control groups. Results are consistent with theories that allow for different activation patterns when elements are combined.     

Effects of overtraining on reversal learning by rats in concurrent and single discriminations

Two experiments examined the whole-partial effect of overtraining in concurrent discriminations and assessed the effect against single discrimination training in rats. In Experiment 1 overtraining facilitated reversal in Group W, in which rats were given concurrent training on two simultaneous discrimination tasks (A and B) in original learning before both tasks were reversed. By contrast, overtraining delayed reversal in Group P, in which animals were given the same training as in Group W in original learning, but only Task A was reversed. After overtraining, Group W reversed more rapidly than Group P. Without overtraining, Group P reversed more rapidly than both Group W and Group C, in which animals were trained only on Task A before this task was reversed. Experiment 2 investigated the effects of overtraining on the reversal of a successive discrimination in Groups W, P, and C. In addition, animals in a further group, Group S, received the same concurrent training as Group W before Task B was omitted and Task A reversed. Overtraining facilitated reversal in Groups W, C, and S but delayed it in Group P. After overtraining, Group W reversed more rapidly than Groups P, S, and C. Both Groups S and C also reversed more rapidly than did Group P. Without overtraining, Group P reversed more rapidly than Groups W, S, and C. There were no significant differences in reversal performance among Groups W, S, and C. These findings indicate that the effects of overtraining on reversal learning in concurrent discriminations differ from those observed in single discriminations.     

Comparing positive and negative patterning in human learning

In human causal learning, positive patterning (PP) and negative patterning (NP) discriminations are often acquired at roughly the same rate, whereas PP is learned faster than NP in most experiments with nonhuman animals. One likely reason for this discrepancy is that most causal learning scenarios encourage participants to treat the presentation and omission of the relevant outcome as two events of comparable significance and likelihood. To investigate this, the current experiments compared PP and NP using a predictive learning paradigm based on a mock gambling task. In Experiment 1, one outcome (winning) was made more salient by being less frequent than the alternative outcome (losing). Under these circumstances, PP was learned faster than NP. In Experiment 2, subjects learned two PP and two NP discriminations, one involved win versus no change outcomes, the other involved lose versus no change outcomes. The subjects learned PP faster than NP, but only when discriminating win from no change. We argue that a difference in difficulty between PP and NP relies on a difference in the salience of the outcomes, consistent with the predictions of a relatively simple model of associative learning.     

Comparing positive and negative patterning in human learning

In human causal learning, positive patterning (PP) and negative patterning (NP) discriminations are often acquired at roughly the same rate, whereas PP is learned faster than NP in most experiments with nonhuman animals. One likely reason for this discrepancy is that most causal learning scenarios encourage participants to treat the presentation and omission of the relevant outcome as two events of comparable significance and likelihood. To investigate this, the current experiments compared PP and NP using a predictive learning paradigm based on a mock gambling task. In Experiment 1, one outcome (winning) was made more salient by being less frequent than the alternative outcome (losing). Under these circumstances, PP was learned faster than NP. In Experiment 2, subjects learned two PP and two NP discriminations, one involved win versus no change outcomes, the other involved lose versus no change outcomes. The subjects learned PP faster than NP, but only when discriminating win from no change. We argue that a difference in difficulty between PP and NP relies on a difference in the salience of the outcomes, consistent with the predictions of a relatively simple model of associative learning.     

Acquisition of superexcitatory properties by an irrelevant background stimulus

Six appetitive conditioning experiments with rats demonstrated that an irrelevant X accompanying a negative patterning discrimination (XA+, XB+, XAB-) acquires extraordinarily high levels of conditioned excitation. Responding to X was similar to that evoked by 2 excitors in combination (Experiment 1) and was greater than responding to a separately reinforced Y (Experiments 2-5). Superexcitatory properties were not acquired by X in the nonpatterning discriminations of Experiments 2-4. Experiment 5 found that A and B, if anything, were weakly excitatory. Making them more strongly excitatory after conditioning did not interfere with retention of the original discrimination (Experiment 6). Results support a counterintuitive prediction of associative theories that, under carefully arranged conditions, irrelevant stimuli may acquire superexcitatory properties.     

Overtraining, extinction and shift learning in a concurrent discrimination in rats

Two experiments examined the influence of overtraining on the reversal of a concurrent discrimination. After rats are trained to criterion on two different discriminations in the same apparatus, the reversal of one of these proceeds more rapidly than when both are reversed. If the reversal is conducted after overtraining on the original discriminations, then the opposite pattern of results is observed. That is, learning about the reversal of both discriminations is more rapid than when only a single discrimination is reversed. Experiment 1 replicated this effect and suggested that it is not caused by differences in the rate of extinction during reversal learning. In order to test a cue-association account for these findings, Experiment 2 examined the effect of exchanging the negative stimuli of a concurrent discrimination. This manipulation had a disruptive influence on performance, but only when subjects were not overtrained on the original discrimination.     

Occasion-setting training renders stimuli more similar: Acquired equivalence between the targets of feature-positive discriminations

In two experiments rats received training on two concurrent appetitive feature-positive discriminations. A preliminary test in Experiment 1 confirmed previous demonstrations of the transfer of occasion-setting properties—the feature from one of these discriminations was better able to facilitate responding to the occasion-set target CS from the second discrimination than to a control stimulus that had not been the subject of occasion-setting. The source of this transfer was investigated in a second phase of training, and in Experiment 2. In both experiments one of the occasion-set CSs was paired with food, and generalization of appetitive conditioned responding from this stimulus to the second occasion-set CS, and to a control cue, was examined. There was more generalization from the first occasion-set CS to the second CS that had also been occasion-set than to the control cue. This is taken as evidence that occasion-set CSs are rendered more similar as a result of their common training history. The implications of these findings for explaining transfer of occasion setting are discussed.     

Transfer of Learning Between Concurrent and Matching (or Non-matching)-to-Sample Discriminations in Rats

Three experiments examined transfer of learning between a concurrent discrimination and a matching (or non-matching)-to-sample discrimination in rats. In Experiment 1, rats were trained to criterion (group NOT) or were overtrained (group OT) on two concurrent discriminations. Subsequently, group OT learned a matching (or non-matching) task more rapidly than did group NOT. In Experiment 2, rats were initially given matching (or nonmatching) tasks and then given whole or half reversal with these tasks. Group whole reversed faster than group half. In Experiment 3, two groups of rats were trained on matching (or non-matching) tasks, and then given concurrent discrimination training, followed by either whole or half reversal training (groups matching and non-matching). Another group (group control) received a pseudo-discrimination followed by the same training in Phases 2 and 3 as groups matching and non-matching. In groups matching and non-matching, rats learned the whole reversal more rapidly than the half reversal. But the opposite result was observed in group control. These findings suggest that transfer effects reported in Experiments 1 and 2 are governed by the same mechanism for the formation of associations between stimuli.     

Overshadowing and associability change

Three appetitive Pavlovian conditioning experiments with rats examined the associability of stimuli A and B that had a history of compound conditioning (AB+), relative to stimuli X and Y that had a history of conditioning in isolation (X+, Y+). Following this training, Experiment 1 revealed that conditioned responding was higher to X and Y than to A and B (overshadowing). In a subsequent AY+, AX-, BY- test discrimination, the AY/BY discrimination was solved more readily than the AY/AX discrimination. In Experiment 2, following AB+, X+, Y+ training, A and Y were presented as a compound and signaled the availability of reinforcement upon the performance of an instrumental response. Test trials in which A and Y were presented alone, and in extinction, revealed that A acquired greater control of instrumental responding than Y. Experiment 3 revealed that following AB+, X+, Y+ training, A and B served as more effective discriminative stimuli for instrumental responding than X and Y. Overall, these results imply that the associability of stimuli conditioned in compound is higher than stimuli conditioned in isolation. These results are discussed in terms of attentional theories of associative learning.     

The influence of associability changes in negative patterning and other discriminations

Normal rats showed faster learning of a serial negative patterning (NP) discrimination (X+, A+, X-->A-) than of a comparable feature negative (FN) discrimination (A+, X-->A-). This advantage was absent in rats with lesions of the amygdala central nucleus. Earlier data indicated that this brain lesion interferes with surprise-induced increases in attention specified by the Pearce-Hall model (J. M. Pearce & G. Hall, 1980). In the NP task, but not the FN task, omission of the reinforcer after X on X-->A- trials was surprising. A variation of the NP task (NPX), in which X was reinforced on both X+ and X-->A- trials, was learned more rapidly than the NP task. Lesioned rats were unimpaired in learning the NPX task. Evaluation of the lesion effects and the results of posttraining transfer tests suggested that the NP advantage involved attentional processes, whereas the NPX advantage was based on the acquisition of inhibitory control by aspects of excitation conditioned to X.     

Elemental and Configural Processes in Patterning Discrimination Learning

Three experiments used appetitive conditioning with rats to examine the involvement of elemental and configural processes in positive and negative patterning discriminations. The first experiment demonstrated that negative and, to some extent, positive patterning discriminations were learned more rapidly when these discriminations consisted of stimulus elements that had previously been the reinforced as opposed to the non-reinforced elements of a simple discrimination. Experiment 2 revealed an excitatory summation effect during the early phase of negative patterning learning that depended upon discrimination pretraining. The final experiment demonstrated faster discrimination learning between the compound and the less salient, rather than the more salient, element of an instrumental patterning task. The present set of results were interpreted as reflecting the possibility, consistent with connectionist theory, that internal representations of the conditioned stimuli change over the course of a patterning discrimination.     

A role for stimulus generalization in conditional discrimination learning

In each of three experiments on discrimination learning by rats, whether or not a 10-sec target stimulus was followed by food was determined by the nature of a 2-min background stimulus that accompanied it. A conditional discrimination was employed in Experiment 1 such that background A indicated food would follow one target but not the other, whereas this relationship between the targets and food was reversed in the presence of background B. Experiment 2 employed two feature-positive discriminations. Subsequent test trials revealed that the background for one discrimination was able to enhance responding during the target for the other discrimination. Experiment 3 employed a feature-positive and a feature-negative discrimination prior to test trials in which each target was presented separately during a compound of both background stimuli. The compound enhanced responding to the target from the feature-positive discrimination and reduced it to the target from the feature-negative discrimination. We suggest that to accommodate all these findings, the best explanation is provided by a configural model of Pavlovian conditioning.     

Negative patterning is easier than a biconditional discrimination

Two groups of rats were trained for 50 days on different discriminations in a magazine approach paradigm. One group was trained with a negative patterning schedule and a positive patterning schedule concurrently: they received intermixed trials of A+, B+, AB-, C-, D-, CD+ (A, B, C, and D are four distinct stimuli; the plus sign denotes reinforcement with food, and the minus sign denotes nonreinforcement). The second group of rats was trained with the same four stimuli arranged as compounds and reinforced according to the biconditional schedule AB+, CD+, AC-, and BD-. The first group learned the positive patterning schedule much more quickly than the negative patterning schedule, but they learned the negative patterning schedule more effectively than the second group learned the biconditional schedule. The authors discuss the implications of these findings for models of stimulus representation.     

Derived control by compound and single stimuli in a matching-to-sample task in children

A matching-to-sample procedure was used to investigate whether 9-year-old children would demonstrate the emergence of a derived compound-sample conditional discrimination following training in four interrelated single-sample conditional discriminations and vice versa, as adults did in previous studies. In Experiment 1, three out of three children demonstrated the emergence of a compound-sample conditional discrimination following training in four single-sample conditional discriminations. In Experiment 2, two out of three children acquired a compound-sample conditional discrimination and they demonstrated the emergence of four single-sample conditional discriminations; one of them did so only after being exposed to a remediation training and testing procedure. Training variables that facilitated discrimination emergence in both directions are discussed. In general, results showed that the sophisticated learning skills that are supposedly possessed by adults are not required to demonstrate the two types of derived control under study.     

Analysis of the unique cue in configural discriminations

Four experiments used an autoshaping procedure with pigeons to investigate the basis of configural discriminations. The elements of both a negative patterning (A+, B+, AB-) and a conditional discrimination (AC+, BD+, AD-, BC-) were paired, in a second-order procedure with two new key lights, X and Y. Responding was then tested to X and Y presented in compound with each other and with A and B. The pattern of responding to compounds containing X and Y was like the pattern of responding to compounds containing their associates, A and B. This suggests that A and B can be replaced by their associates without disrupting responding to their compounds. Because X and Y are physically different from A and B, this in turn suggests that any unique cue controlling responding to their compounds does not depend on the physical presence of the component stimuli. Instead the unique stimulus appears to arise from the joint activation of memory representations.     

Solving XOR

Three experiments examined the way in which exclusive-or (XOR) problems are solved by rats. All rats first received food-rewarded positive and negative patterning problems with two stimulus sets: either A+, B+, AB- and C-, D-, CD+, or A-, B-, AB + and C+, D +, and CD-. Subsequently, rats received revaluation trials in which A was paired with shock and C was not, prior to generalization test trials with B, D, AB, and CD (Experiments 1 & 2); or received A-->shock trials prior to tests with B and CD (Experiment 3). There was greater generalized fear to B than to either D (Experiments 1 & 2) or AB (Experiment 2) and CD (Experiments 2 & 3). These results are inconsistent with configural, connectionist models, but are consistent with an alternative connectionist model that can represent the logical structure of XOR problems.     

Transfer and Counterconditioning of Conditional Control in the Rabbit Nictitating Membrane Response

Two experiments using the rabbit nictitating membrane response investigated whether training in one conditional discrimination (A X+, B X-), enabled the feature cues (A and B) to modulate responding to another CS (Y) trained as a target stimulus in a second conditional discrimination (C Y+, D Y-). There was near-complete transfer of the feature cue's conditional control, indicating that the feature cue's ability to modulate responding is not based on an association specific to the training target. Experiment 2 also revealed that the role of a stimulus to act as a conditional cue is affected by its ability to act as a simple conditioned excitor or inhibitor. Following initial acquisition of two conditional discriminations, two feature cues were reinforced in a pattern consistent with the initial conditional discrimination (A +, B-), whereas the other two feature cues were reinforced in the reverse pattern to that of the original conditional discrimination (C-, D +). Subsequent tests revealed that the reversed training of the feature cues interfered with the original conditional discriminations. The results are consistent with theories that the feature cue gains an association with a representation of the emotional attributes of the US, which acts to modulate responding to the target stimulus through a diffuse change in motivational level. However, hierarchical theories of conditional discriminations that assume a lack of CS-specificity may also be able to explain the findings.     

Emergent simple discrimination in children: Transfer of stimulus control under non-reinforced conditions

Previous research on humans suggests that simple discriminations may emerge if both stimuli, B1 and B2, are compounded with the stimuli of a previously trained discrimination, A1 (S+) and A2 (S-), and responding to the compounds, B1A1 and B2A2, is reinforced. Two questions were addressed. First, do simple discriminations also emerge if (a) only one stimulus, B1, is compounded with a training stimulus, A1 (S+) or A2 (S-); or with both training stimuli, A1 (S+) and A2 (S-); and (b) neither B1 nor B2 is compounded with the training stimuli? Second, do simple discriminations emerge if reinforcement in the presence of the AB compounds is withheld? Normal preschool children served as subjects. The study consisted of six experiments. Transfer occurred in all experiments regardless of whether both test stimuli, one test stimulus, or none of the test stimuli were compounded with the training stimuli under non-reinforced conditions. The results can be described by the following rules: Respond to any stimulus that includes a component of a “correct” stimulus of a previous discrimination. Otherwise, respond away from the stimulus that incorporates a component from an “incorrect” stimulus of a previous discrimination. The implications of data for sensory pre-conditioning and language-based accounts are discussed.     

Probing the Time Course of Nonlinear Discriminations during Human Electrodermal Conditioning

Animal-based theories of Pavlovian conditioning propose that patterning discriminations are solved using unique cues or immediate configuring. Recent studies with humans, however, provided evidence that in positive and negative patterning two different rules are utilized. The present experiment was designed to provide further support for this proposal by tracking the time course of the allocation of cognitive resources. One group was trained in a positive patterning schedule (A−, B−, AB+) and a second in a negative patterning schedule (A+, B+, AB−). Electrodermal responses and secondary task probe reaction time were measured. In negative patterning, reaction times were slower during reinforced stimuli than during nonreinforced stimuli at both probe positions while there were no differences in positive patterning. These results support the assumption that negative patterning is solved using a rule that is more complex and requires more resources than does the rule employed to solve positive patterning.