A. G. Baker

Systematic comparison of the effects of hippocampal and fornix-fimbria lesions on acquisition of three configural discriminations

The effects of lesions to the hippocampal system on acquisition of three different configural tasks by rats were tested. Lesions of either the hippocampus (kainic acid/colchicine) or fornix‐fimbria (radiofrequency current) were made before training. After recovery from surgery, rats were trained to discriminate between simple and compound‐configural cues that signaled the availability or nonavailability of food when a bar was pressed. When positive cues were present, one food pellet could be earned by pressing a lever after a variable time had elapsed. The trial terminated on food delivery (variable interval 15 s). This procedure eliminates some possible alternative explanations of the results of previous experiments on configural learning. Hippocampal lesions increased rates of responding and retarded acquisition of a negative patterning task (A+, B+, AB−); using a ratio measure of discrimination performance these lesions had a milder retarding effect on a biconditional discrimination (AX+, AY−, BY+, BX−), and they had no effect on a conditional context discrimination (X: A+, B−; Y: A−, B+). Fornix‐fimbria lesions did not affect acquisition of any of these tasks but increased rates of responding. The results suggest that several task parameters determine the involvement of the hippocampus in configural learning; however, all tasks tested can also be learned to some extent in the absence of an intact hippocampal system, presumably by other learning/memory systems that remain intact following surgery. The lack of effect of fornix‐fimbria lesions on any of these tasks suggests that retrohippocampal connections with other brain areas may mediate hippocampal contributions to the learning of some configural tasks. An analysis of these results and of experiments on spatial learning situations suggests that involvement of the hippocampus is a function of the degree to which correct performance depends on a knowledge of relationships among cues in a situation. Hippocampus 7:371–388, 1997. © 1997 Wiley‐Liss, Inc.

Extinction of the Context and Latent Inhibition

Abstract The hypothesis that latent inhibition could be reduced by extinguishing the experimental context, that is, exposing the rats to the context between exposure to the conditional stimulus (CS) and conditioning, was tested. All experiments used the conditioned emotional response procedure. In Experiment 1, extinction was not effective when the animals were exposed to the clicker 40 times off the baseline of responding for food and when the clicker CS was partially reinforced with shocks during the test phase. In Experiments 2 and 3, latent inhibition could be reduced by extinction if the animals were exposed to the CS 24 or 16 times on-baseline, and if continuous reinforcement was used during the test. In Experiments 4, 5, and 6, we attempted to determine which variable was responsible for the discrepant results. In Experiment 4, extinction was effective with 20 or 40 on-baseline exposures to the CS, using continuous reinforcement during the test. In Experiment 5, extinction was not effective with exposure on- or off-baseline, using 24 exposures and partial reinforcement. Finally, in Experiment 6, extinction reduced latent inhibition using continuous, but not partial, reinforcement with 40 exposures off-baseline. From these results, we concluded that Wagners model of habituation was not sufficient to account for latent inhibition and that a hybrid model, using both associative and cognitive representational processes, was necessary.

Contextual conditioning and the US preexposure effect in conditioned fear.

A series of five experiments was carried out in which fear of context caused by exposure to shocks was manipulated by signaling the shocks with a discrete stimulus, signaling the days during which shocks occurred with a session-long stimulus, or switching the context between exposure and the subsequent test. All these manipulations influenced fear of the context in the manner predicted by the Rescorla-Wagner associative model. Following this, all the rats were given conditioning trials with shock and a different discrete stimulus. All preexposure treatments produced consistent and reliable interference with conditioning with the exception of signaling the shocks with a discrete stimulus, which greatly reduced interference. These results are interpreted as being consistent both with a cognitive explanation of the US exposure effect, which claims that animals learn that shocks are unpredictable during conditioning and this knowledge retards future conditioning when they are predictable, and with an adaptation explanation, which claims that unpredictable shocks produce chronic fear and this fear through either a change in adaptation level or through emotional exhaustion renders the shocks less reinforcing during the conditioning test.

Contextual conditioning and reinstatement of extinguished instrumental responding

In three experiments we studied the relationship between contextual conditioning and the reinstatement of extinguished lever pressing that occurs when noncontingent food is introduced following extinction. In all three experiments the non-contingent food was presented off-baseline (with the response levers not present). On subsequent tests, with the response levers present, animals that had been exposed to food showed more reinstatement of lever pressing than control animals. This finding rules out alternative mechanisms for the reinstated responding that rely on the interaction of non-contingent food and responding, such as superstitious reinforcement or the discriminative after-effects of food. In addition, in each experiment we demonstrated that manipulations known to affect contextual conditioning (signalling the food in Experiment 1, context extinction in Experiment 2, and switching contexts in Experiment 3) reduced the reinstatement. These results are consistent with the claim that contextual conditioning is important in controlling instrumental conditioning and closely parallel findings concerning the reinstatement of Pavlovian responsing following extinction.

Associative and Normative Models of Causal Induction: Reacting to Versus Understanding Cause

This chapter discusses two views of causal judgment that are roughly analogous to a distinction between being able to react appropriately to causes and being able to understand them. The associationist view is identified with the British Empiricists. It claims that the judgments of cause come from certain empirical cues to causality, which includes: (1) regular succession, (2) temporal contiguity, and (3) spatial contiguity. Associations between events are strengthened when the events are contiguous and are weakened when an event occurs by itself. These models have the advantage that they are computationally simple and they impose a low memory load on the organism because experience is stored as a small number of associative strengths. They have the disadvantage that information about past events is lost in the computation. Also, these models do not have episodic memory. The second classes of models are referred to as normative models. They claim that humans and other animals compute the covariation between cause and effect and then use this information as part of a causal model or schema. The chapter reviews that a retrospective normative model makes the choice of domain in which to do the normative calculation. Associative models and those that involve causal models or schema are appropriate to overlapping but not identical domains of information processing. Simple associative ideas can be used in many situations in which contiguity is important, but in which mental models are unavailable. These can be used in situations in which associative networks are difficult to apply.

Judging the importance of constant and variable candidate causes: A test of the power PC theory.

In two causal induction experiments subjects rated the importance of pairs of candidate causes in the production of a target effect; one candidate was present on every trial (constant cause), whereas the other was present on only some trials (variable cause). The design of both experiments consisted of a factorial combination of two values of the variable causes covariation with the effect and three levels of the base rate of the effect. Judgements of the constant cause were inversely proportional to the level of covariation of the variable cause but were proportional to the base rate of the effect. The judgements were consistent with the predictions derived from the Rescorla-Wagner (1972) model of associative learning and with the predictions of the causal power theory of the probabilistic contrast model (Cheng, 1997) or “power PC theory”. However, judgements of the importance of the variable candidate cause were proportional to the base rate of the effect, a phenomenon that is in some cases anticipated by the power PC theory. An alternative associative model, Pearces (1987) similarity-based generalization model, predicts the influence of the base rate of the effect on the estimates of both the constant and the variable cause.

A role for CS-US contingency in Pavlovian conditioning

Two experiments evaluated the role of conditioned stimulus-unconditioned stimulus (CS-US) contingency in appetitive Pavlovian conditioning in rats. In both experiments, some groups received a positively contingent CS signaling an increased likelihood of the US relative to the absence of the CS. These groups were compared with control treatments in which the likelihood of the US was the same in the presence and absence of the CS. A trial marker served as a trial context. Experiment 1 found contingency sensitivity. There was a reciprocal relationship between responding to the CS and the trial marker. Experiment 2 showed that this result was not stimulus or response specific. These results are consistent with associative explanations and the idea that rats are sensitive to CS-US contingency.

Contextual conditioning during free-operant extinction: Unsignaled, signaled, and backward-signaled noncontingent food

Three experiments were designed to study the effects of contextual conditioning on the extinction of instrumental leverpressing that had been reinforced on a random-interval schedule. In Experiment 1, noncontingent food retarded extinction, but signaling food delivery, a treatment that should reduce contextual conditioning, reduced the interference. Experiment 2 replicated the results of Experiment 1 and demonstrated that if the food preceded rather than followed the signal, the retardation of extinction was not reduced but was enhanced. In Experiment 3, non-contingent leverpressing was used to directly verify that the three treatments—forward signaling, noncontingent food, and backward signaling—differentially influenced contextual conditioning. Forward signaling produced the least, and backward signaling produced the most, contextual conditioning. This monotonic relationship between contextual conditioning and interference with extinction was used as evidence to support the argument that context-food associations are important in controlling instrumental responding.

Human judgments of positive and negative causal chains

Three experiments investigated the way participants construct causal chains from experience with the individual links that make up those chains. Participants were presented with contingency information about the relationship between events A and B, as well as events B and C, using trial-by-trial presentations. The A-B and B-C contingencies could be positive, negative, or zero. Although participants had never experienced A and C together, A-C ratings were a multiplicative function of the A-B and B-C contingencies. These findings can be generated by an auto-associator using the delta rule. This explanation is also useful for understanding sensory preconditioning and second-order conditioning.

Asymptotic judgment of cause in a relative validity paradigm

We report three experiments in which we tested asymptotic and dynamic predictions of the Rescorla—Wagner (R—W) model and the asymptotic predictions of Cheng’s probabilistic contrast model (PCM) concerning judgments of causality when there are two possible causal candidates. We used a paradigm in which the presence of a causal candidate that is highly correlated with an effect influences judgments of a second, moderately correlated or uncorrelated cause. In Experiment 1, which involved a moderate outcome density, judgments of a moderately positive cause were attenuated when it was paired with either a perfect positive or perfect negative cause. This attenuation was robust over a large set of trials but was greater when the strong predictor was positive. In Experiment 2, in which there was a low overall density of outcomes, judgments of a moderately correlated positive cause were elevated when this cause was paired with a perfect negative causal candidate. This elevation was also quite robust over a large set of trials. In Experiment 3, estimates of the strength of a causal candidate that was uncorrelated with the outcome were reduced when it was paired with a perfect cause. The predictions of three theoretical models of causal judgments are considered. Both the R-W model and Cheng’s PCM accounted for some but not all aspects of the data. Pearce’s model of stimulus generalization accounts for a greater proportion of the data.