Analysis of the role of associative inhibition in perceptual learning by means of the same-different task.

Abstract

In Experiment 1a, participants were exposed, over a series of trials, to separate presentations of 2 similar checkerboard stimuli, AX and BX (where X represents a common background). In one group, AX and BX were presented on alternating trials (intermixed), in another, they were presented in separate blocks of trials (blocked). The intermixed group performed to a higher standard than the blocked group on a same-different test. A superiority of intermixed over blocked exposure was also evident in a within-subject design (Experiment 1b) and when the test required discrimination between a preexposed stimulus and the background (e.g., AX vs. X), even if the background changed between preexposure and test (AY vs. Y) (Experiment 2). In Experiment 3, the intermixed/blocked effect was observed when, in preexposure, stimulus presentations were alternated with the background alone (e.g., AX/X). This suggests that the perceptual learning effect is not the consequence of inhibitory associations between unique features but to increased salience of those features. Experiment 4 confirmed this finding and also ruled out an account of the effect in terms of trial spacing.

Varieties of perceptual learning

“…Indeed, as Gibson herself argued, the most effective schedule should be one in which the two stimuli are presented simultaneously. This prediction was confirmed by Mundy et al (2007) in an experiment on face recognition by people, and also by Mitchell et al (2008) in an experiment in which people were required to discriminate between two very similar checkerboard patterns. The result is important, since it is clearly inconsistent with the inhibitory account of McLaren and Mackintosh: If AX and BX occur together on the same trial, A and B cannot predict each other's absence.…”

“…From the first, it was always apparent that differential latent inhibition of common and unique features could not be the only mechanism underlying unsupervised perceptual learning (Mackintosh, Kaye, & Bennett, 1991;McLaren, Kaye, & Mackintosh, 1989). One experimental finding that has been taken as reinforcing this conclusion was first reported by Honey, Bateson, and Horn (1994) in an experiment with domestic chicks but has since been replicated in numerous other studies, in experiments on taste aversion in rats (Bennett & Mackintosh, 1999;Symonds & Hall, 1995) and on spatial learning in rats (Prados, Artigas, & Sansa, 2007) and in a variety of different experiments with people (Mitchell et al, 2008;Mundy, Dwyer, & Honey, 2006). The magnitude of any perceptual-learning effect depends not only on the amount of exposure to two or more stimuli, but also on the way that exposure is scheduled.…”

“…There is, in fact, a lower level explanation that appeals to the phenomenon of habituation. As Honey et al (1994) and Mundy et al (2007) noted (see also Mitchell et al, 2008), intermixed and blocked presentations of AX and BX can equate the total number of times…”

Perceptual learning with complex visual stimuli is based on location, rather than content, of discriminating features.

“…However, because the unique features remained in the same place for within-and between-pair tests, if participants had simply learnt to look to the locations where the unique features of intermixed stimuli appeared, then the success of between-pair discriminations can be explained without recourse to changes in feature salience. Similarly, Mitchell, Kadib, et al (2008), report that after exposure to AX/BX discrimination was equivalently good for AX/X as it was for BY/Y (i.e. exposure effects generalised to a new common background -Experiment 2).…”

“…In particular, a series of studies (e.g. Lavis & Mitchell, 2006;Mitchell, Kadib, Nash, Lavis, & Hall, 2008;Wang & Mitchell, 2011) used checkerboard stimuli (See Figure 1 for examples) that were created by taking a 20 × 20 grid of multicoloured squares (these were the common features: X) and then adding, to a particular place on the background, features made of blocks of 4-6 squares, consisting one or two colours (the unique features: A/B. The exact details of both the unique and common features differed slightly between experiments).…”