Contextual conditional discriminations

Three experiments used a discriminated operant procedure to study conditional discrimination learning in rats. The first experiment showed that rats were capable of learning a biconditional discrimination in which two contexts served as conditional cues signalling the reinforcement contingencies associated with two discriminative stimuli. The discrimination was learned equally well when one discriminative stimulus signalled food, the other its absence, and when one stimulus signalled food, the other extinction plus mild footshock.

In Experiment 2 it was shown that prior training on such a conditional discrimination enhanced the subsequent context specificity of simple conditioning relative to control groups of animals for whom the prior training had not been conditional. Experiment 3 showed that a reversal of the significance of one pair of discriminative stimuli produced no spontaneous reversal in performance to a second, target, pair.

The pattern of results is best accounted for by an analysis of contextual conditional discrimination learning in terms of stimulus configurations and offers no support for the notion that rats may learn a general conditional rule or set.     

Redistributing control between the elements of a compound stimulus

Twenty-eight pigeons were given discrimination training. Four groups were run in which for 0, 2, 4 and 9 sessions of training colour was an irrelevant cue and tone was relevant. Generalization tests were given in extinction; the training colour (C₁) and another colour (C₂) were presented with tone (T) and noise (N). Responding to these four stimulus combinations, TC₁, TC₂, NC₁ and NC₂, allowed the assessment of control by colour and tone. A reduction of control by colour and an increase in control by tone were found to be positively related to the amount of training with colour irrelevant. Tests showed a strong inverse relationship between control by colour and tone.     

Combination of drive and incentive

This experiment was designed to investigate the combination of drive and incentive as determinants of performance. Nine groups of rats were trained to press a bar under three levels of food deprivation (12, 24 and 36 hr.) and three incentive conditions (1,2 and 3 pellets). Response strength was estimated by counting the number of responses with a latency of 1 sec. or less during five 20 trial sessions. The results indicated that response strength increased with hours of deprivation and with amount of food reward. Significant interactions between sessions and drive, and sessions and incentive provided additional support for the multiplicative combination of H (habit) and D (drive) and H and K (incentive). The lack of significant interaction between D and K was interpreted as supporting the hypothesis that D and K combine additively rather than multiplicatively.     

Overtraining, transfer to proprioceptive control and position reversal

If rats are overtrained on a visual discrimination they generally learn the reversal of the discrimination faster than non-overtrained rats. In position discriminations, however, this effect does not generally hold—indeed several investigators have found overtraining to retard position reversal. One of the important differences between the two types of problem seems to be the presence of irrelevant cues in visual discriminations, and their absence in position discriminations. It is suggested that a second important feature of position discriminations is that overtraining usually causes control of the maze running habit to be transferred from external to proprioceptive stimuli, and that successful reversal cannot normally occur until external control is re-established. In two experiments a study by Krechevsky and Honzik (1932) is repeated with certain modifications, with results that support this hypothesis; and a third experiment provides direct evidence of transfer to proprioceptive control in a T-maze. It is shown that this analysis will explain the apparently conflicting results of all recent position reversal experiments.     

One-trial overshadowing

Two experiments on conditioned suppression in rats examined the extent to which conditioning to a tone was overshadowed by the presence of a light. When animals received four conditioning trials, a light of moderate intensity significantly overshadowed the tone. On the first trial of conditioning, however, the light had no effect on conditioning to the tone, unless (Experiment I) its intensity was increased, or (Experiment II) it occurred in closer temporal contiguity to reinforcement than the tone. The results establish that overshadowing can occur on the first trial of conditioning, contrary to the predictions of several theories, but suggest that this may be for different reasons from those operating to produce overshadowing over a series of conditioning trials.     

Blocking and overshadowing between intra-maze and extra-maze cues: A test of the independence of locale and guidance learning

Rats were trained on an elevated maze where the rewarded alternative was defined either in terms of intra-maze cues (rubber or sandpaper flooring on rewarded and unrewarded arms, regardless of their position) or in terms of extra-maze cues (the correct arm always pointed toward a particular corner of the room, and was sometimes covered with rubber and sometimes with sandpaper), or where both sets of cues were simultaneously relevant. In Experiment 1 rats pretrained with either intra-maze or extra-maze cues alone relevant learned less about the other set of cues than non-pretrained control groups, when, in a second phase of the experiment, both sets of cues were simultaneously relevant. Experiment 2 confirmed that intra-maze cues could block extra-maze cues, and ruled out one alternative explanation of the results of Experiment 1. Experiment 3 showed that extra-maze cues overshadowed intra-maze cues, but that there was no reciprocal overshadowing of extra-maze by intra-maze cues. This was despite the fact that animals learned the intra-maze discrimination significantly faster than the extra-maze discrimination. Experiment 3 also suggested that rats did not solve the extra-maze discrimination by learning to approach or avoid specific extra-maze cues, but rather by locating the correct arm by reference to the entire set of extra-maze cues. The results suggest that locale or place learning and cue or guidance learning, in O'Keefe and Nadel's (1978) terminology, interact with one another in much the same way as does learning about any pair of stimuli in a Pavlovian conditioning experiment.     

Bigger than a breadbox? Attention to distractors may not enhance negative priming

In several recent studies, P. A. MacDonald and colleagues (e.g., P. A. MacDonald & S. Joordens, 2000) reported unusually large negative priming effects and claimed that attention to distractors, counter to expectations, served to enhance the magnitude of the effect. In 3 experiments using their novel comparative judgment task, negative priming was assessed using both a control condition based on that of P. A. MacDonald, S. Joordens, and K. N. Seergobin (1999) and one in which control probe items exactly matched those on ignored repetition trials. In MacDonald et al.'s unmatched control condition, participants were faster than on ignored repetition trials, but this difference was reduced or absent when control items were matched. This result led to the conclusion that the apparently large negative priming effect reported by MacDonald and colleagues may be an artifact arising because judgments for a subset of their control trials were relatively easier than for ignored repetition trials.     

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.     

Enduring consequences of experimentally induced biases in interpretation

Previous research has demonstrated that it is possible to induce biases in the interpretation of ambiguous text passages by training. Participants consistently trained to interpret emotionally ambiguous passages in either a negative or positive direction show training-congruent effects when presented with new ambiguous material. These training effects are demonstrated by participants' subsequent recognition ratings for disambiguating sentences, which represent both possible meanings of the novel ambiguous test passages. In this series of experiments, we investigated the durability of these training effects over time spans of up to 1 day and found them to be robust. The findings encourage us to believe that induced biases may serve as a useful analogue to those observed clinically.     

Overshadowing of a stimulus-reinforcer association by an instrumental response

In two experiments, rats were first exposed to pairings of a clicker and food; they were subsequently, in order to measure the effectiveness of the clicker as a conditioned reinforcer, given the opportunity to press a lever which turned the clicker on. For one group of animals the food originally delivered in the presence of the clicker had been contingent on their performance of an instrumental response (running in a running wheel); for a second the contingency between clicker and food had been purely classical. Although the actual correlation between clicker and food was identical for the two groups, the clicker was a less effective conditioned reinforcer for the first group than for the second. In a third experiment, all animals were initially required to run to obtain food in the presence of the clicker, but one group received additional trials on which food was delivered contingent on running in the absence of the clicker. This group showed less tendency to lever press for the clicker than a second group that had received free food on trials when the clicker was not presented. The results of all three experiments suggest that conditioning to the clicker could be overshadowed if the occurrence of food was more reliably predicted by the execution of an instrumental running response; they thus support the view that instrumental conditioning depends on the establishment of an association between response and reinforcer similar to the association between stimulus and reinforcer underlying classical conditioning.