Mechanisms of perceptual learning: Prolonged intermixed preexposure reduces the effectiveness of the unique and the common elements.

Abstract

In three experiments, rats were given intermixed or blocked preexposure to two similar compound stimuli, AX and BX. In Experiment 1, following preexposure, animals were given appetitive conditioning training with the compound AX. A subsequent generalization test showed better discrimination between AX and BX in the group given intermixed than in the one given blocked preexposure. Experiments 2 and 3 assessed the nature of the learning mechanisms underlying this instance of the perceptual learning effect. Experiment 2 assessed the associability of the common and unique elements (X and A); animals in the group given intermixed preexposure showed poorer conditioning with both the X and the A elements than those given blocked preexposure. Experiment 3 further assessed the perceptual effectiveness of the distinctive element A using a superimposition test (the capacity of A to interfere with the conditioned response commanded by an independent conditioned stimulus). The results showed, in line with the outcome of Experiment 2, that the unique element A is more salient following blocked than intermixed preexposure. These results are discussed by reference to current theories of perceptual learning.

The opportunity to compare similar stimuli can reduce the effectiveness of features they hold in common.

“…It should be acknowledged that, for the most part, the learning mechanisms considered so far would apply as readily to procedures in which stimulus presentations are widely spaced as to those used here (and also those of Artigas & Prados, 2017; Ballesta et al 2021; and Mondragón & Murphy, 2010. Although no evidence has been found that the preexposure schedule determines the properties of the X element when stimulus presentations are widely spaced (see for a review Hall, 2020), this does not necessarily imply that the mechanisms described above are ineffective.…”

“…This is a slightly different proposal from that just discussed according to theories (like that of Pearce & Hall, 1980) that make a distinction between the associability of a stimulus and its salience—the former governing the rate of learning whereas the latter also determines the vigor of the response. An account in terms of salience change comes from the analysis offered by Ballesta et al (2021; see also Artigas & Prados, 2014, 2017). They proposed that during intermixed preexposure, X would form inconsistent and weak associations with the unique elements, which ultimately results in separate representations for each element (A, B, and X), whereas blocked exposure would establish stronger associations in each block of trials, generating configural-like representations (AX, BX).…”

“…They were kept on a 12-hr light/dark cycle that began at 8:00 a.m. This sample size has proven adequate in the past to detect changes in common element properties following rapid pre-exposure to compound stimuli with similar procedures and designs (e.g., Ballesta et al, 2021; Mondragón and Murphy, 2010). The sample size in these studies was, as in our case, eight subjects per group.…”

“…The experiments by Mondragón and Murphy (2010) and by Ballesta et al (2021) tested the properties of the X stimulus after intermixed or blocked preexposure by using it as the CS in a simple excitatory conditioning paradigm. The present Experiment 3A (see Table 1) used the same procedure.…”

Special issue on recent advances in perceptual learning.

“…Rodriguez et al (2021) present empirical data from human perceptual learning tasks directly investigating the hypothesis that salience to distinctive elements of a stimulus compound is enhanced following an intermixed preexposure procedure. However, using a rodent perceptual learning paradigm, Ballesta et al (2021) provide evidence suggesting that the salience of both distinctive and common elements of two similar stimulus compounds is reduced following prolonged intermixed preexposure. George and Haddon (2021) also present data from a novel rodent perceptual learning task suggesting that sometimes preexposure to stimuli, in their case differing along the frequency continuum, can diminish the rate of subsequent discrimination learning between the two.…”