An investigation of the mechanisms responsible for perceptual learning in humans

Discrimination between similar stimuli is enhanced more by intermixed pre-exposure than by blocked pre-exposure to those stimuli. The salience modulation account of this intermixed-blocked effect proposes that the unique elements of intermixed stimuli are more salient than those of blocked stimuli. The inhibition account proposes that inhibitory links between the unique elements of intermixed stimuli enhance discrimination. The current thesis evaluated the two accounts in their ability to explain this effect in humans.In Experiments 1 and 2, categorisation and same-different judgements were more accurate for intermixed than for blocked stimuli. This indicates that intermixed pre-exposure decreases generalisation and increases discriminability more than does blocked pre-exposure.In Experiments 3 5, same-different judgements were more accurate when at least one of the two stimuli was intermixed. This enhanced discrimination was not confined to two stimuli that had been directly intermixed. These results are better explained by salience modulation than by inhibition.Experiments 6 8 employed dot probe tasks, in which a grid stimulus was followed immediately by a probe. Neither intermixed nor blocked stimuli showed facilitated reaction times when the probe appeared in the location of the unique element. In Experiments 9 11 participants learned to categorise the intermixed unique elements more successfully than the blocked unique elements, but only when the unique elements were presented on a novel background during categorisation. Experiments 6 11 provide weak evidence that the intermixed unique elements are more salient than their blocked counterparts.In Experiment 12, participants were presented with the shape and location of a given unique element, and were required to select the correct colour. Performance was more accurate for intermixed than for blocked unique elements. In Experiment 13, participants learned to categorise intermixed, blocked and novel unique elements. Performance was better for intermixed than for blocked and novel unique elements, which did not differ. None of the proposed mechanisms for salience modulation anticipate these results.The intermixed-blocked effect in human perceptual learning is better explained by salience modulation than by inhibition. However, the salience modulation accounts that have been proposed received little support. An alternative account of salience modulation is considered