Amount of training affects associatively-activated event representation
Introduction
The view that learned performances become increasingly automatized with extended training is widely held (see, for example, Allport, 1937Tolman, 1948Kimble and Perlmuter, 1970Dickinson, 1980). Although the definition of automaticity often proves troublesome, a common thread in these definitions is the notion that automatic behavior is not guided by goal expectancies, and is resistant to changes in response–reinforcer or stimulus–reinforcer relations. For example, Tolman (1948)noted that rats with extensive experience in a maze were unlikely to abandon a well-practiced route when a shorter route to the goal was made available, whereas rats with moderate amounts of training adapted rapidly. Outcomes like these led Tolman (1948)to suggest that early in training, rats formed plastic, comprehensive cognitive maps, which might be replaced by inflexible, narrow strip maps with extended training. Similarly, Dickinson (1980)suggested that extended training might be accompanied by a shift in the associative structure responsible for learned behavior, from stimulus–reinforcer or response–reinforcer associations which are sensitive to modifications in the reinforcer to more automatic, stimulus–response habits which are not. The experiments described in this article report investigations of these views in the context of the Pavlovian appetitive conditioning of rats.
Recently, there has been extensive investigation of the involvement of stimulus–reinforcer goal expectancies in the production of simple conditioned responses in Pavlovian conditioning experiments. Most modern accounts of Pavlovian conditioning presume that, as a result of conditioning, the conditioned stimulus (CS) acquires the ability to activate an internal representation of the unconditioned stimulus (US). One consequence of this activation is the production of a conditioned response (CR) that often resembles the unconditioned response (UR). Thus, the normal performance of the CR is thought to be mediated by the activation of a US representation.
This claim is supported by considerable data that show that responding to the CS is often sensitive to post-training alterations in the value of the US. For example, Holland and Straub (1979)first trained rats with Pavlovian pairings of an auditory CS with a food pellet US. Then, the food pellet US was devalued by pairing it with injections of the toxin, LiCl, in the absence of the CS. As a result of these US–LiCl pairings, the rats formed an aversion to food pellets and would not consume them in later consumption tests. More important, a test of conditioned food cup responding to the auditory CS (in the absence of the food pellet US) showed a spontaneous drop in CRs as well. These results are consistent with the view that the performance of these CRs was mediated by a representation of the US: when the value of that representation was changed after the completion of training, the nature of the Pavlovian CR changed accordingly.
Holland, 1981, Holland, 1990a, Holland, 1990bnoted another implication of the view that Pavlovian CS activate a representation of the US: a CS-activated representation of the US might mediate new learning about the US, even in the absence of that US. For example, Holland (1981), Experiment 1) first paired an auditory CS with a food pellet US. Then, the auditory CS was paired with injection of LiCl, in the absence of the food US itself. Although the CS–LiCl pairings did not alter behavior during the CS (many investigators have found it difficult to establish direct associations between auditory CSs and illness, e.g. Garcia and Koelling, 1966), in subsequent consumption tests, these rats consumed fewer food pellets than control rats that had received either the CS and US or CS and LiCl unpaired. Holland (1981)interpreted these results as indicating that a CS–activated representation of the food pellet US substituted for the food itself in the acquisition of the conditioned aversion. That is, the aversion to the food was acquired as a result of pairing of a representation of food (evoked by the CS) and the sickness-inducing toxin; no actual pairing of food and toxin occurred.
Experiments 1 and 2 of this report extended Holland’s (1981) studies by investigating the effects of varying amounts of initial auditory CS–food training on the ability of the auditory CS to substitute for the food US in the acquisition of an aversion to that food. In these experiments, the mediated learning effects observed by Holland (1981)occurred only with relatively small amounts of initial training. Using conditioning procedures identical to those of Experiment 2, Experiment 3 examined the effects of variations in the amount of initial CS–food training on the sensitivity of CR performance to post-training devaluation of the US, as in Holland and Straub’s (1979) studies. In contrast with the training effects on mediated learning about the food observed in Experiments 1 and 2, but consistent with data from other laboratories (e.g. Colwill and Rescorla, 1985), mediated performance of CRs to the CS itself was unaffected by the amount of initial CS–food training. That is, although extended training reduced the rats' ability to use a CS-activated representation of food to learn new information about that food, performance of an established CR to the CS remained dependent on the integrity of that representation.
Section snippets
Experiment 1
Experiment 1 examined the effects of variation in the number of CS–US pairings on the CS's ability to substitute for the food US in the acquisition of a food aversion. Each rat first received 64 pairings of one CS (CS1) with a distinctively flavored food US (US1) and 16 pairings of another CS (CS2) with another food US (US2). To equate experience with the two USs, the rats also received 48 pairings of a visual CS with US2. Next, half of the rats received pairings of CS1 with injection of LiCL,
Experiment 2
Experiment 2 extended the results of Experiment 1 by examining the effects of intermediate numbers of CS–US pairings, and by providing an unpaired control. In Experiment 1, 16 CS–US pairings yielded surrogate-mediated conditioning but 64 pairings did not; Experiment 2 examined the phenomenon after 0, 16, 28, and 40 pairings. Four groups of rats received 16 pairings of CS1 with US1, and either 0, 16, 28, or 40 pairings of CS2 with US2. To equalize experience with US1 and US2, the four groups of
Experiment 3
The results of Experiments 1 and 2 are in accord with the view that extended training leads to greater automatization. A simple interpretation of these results is that early in Pavlovian conditioning, the CS acquires the ability to activate a representation of the US, but loses that ability with continued training. One implication of this simple account is that other phenomena believed to be mediated by the CS's activation of a representation of the US should also be dependent on minimal
Discussion
The results of Experiments 1 and 2 join others in suggesting that a CS-activated representation of the US can serve as a substitute for that US in the establishment of new learning (see Holland, 1990a, Hall, 1996, for reviews). Casually speaking, experiencing illness while merely remembering a flavored food can be sufficient to establish an aversion to that food itself. Moreover, these results show that this ability of a CS to mediate acquisition of new learning about the US is more pronounced
Acknowledgements
This research was supported in part by grants from the National Science Foundation (USA). National Institutes of Health (USA) guidelines for the care and use of laboratory animals were followed throughout the conduct of this research. A portion of the procedures and results of Experiment 2 were described briefly by Holland (1990a)pp. 122–3).
References (23)
- et al.
Dissociable effects of anterior and posterior cingulate cortex lesions on the acquisition of a conditional visual discrimination: facilitation of early learning vs. impairment of late learning
Behav. Brain Res.
(1996) - et al.
The amygdala and emotion
Cur. Opin. Neurobiol.
(1996) - et al.
The taste reactivity test. I. Mimetic responses to gustatory stimuli in neurologically normal rats
Brain Res.
(1978) Acquisition of representation-mediated conditioned food aversions
Learn. Motiv.
(1981)Event representation in Pavlovian conditioning: image and action
Cognition
(1990)Personality: A Psychological Interpretation
(1937)- et al.
Instrumental responding remains sensitive to reinforcer devaluation after extensive training
J. Exp. Psychol. Anim. Behav. Proc.
(1985) Contemporary Animal Learning Theory
(1980)- et al.
Relation of cue to consequence in avoidance learning
Psychon. Sci.
(1966) - et al.
Neural substrate of intramodal and crossmodal memory in rhesus monkeys
Int. J. Psychol.
(1996)