Human latent inhibition: Problems with the stimulus exposure effect

Abstract

Latent inhibition (LI) is a startlingly simple effect in which preexposure of a stimulus without consequence retards subsequent responding to a stimulus-consequence relation. The effect was first demonstrated with Pavlovian conditioning in animals and was later suggested to be a marker of human psychopathology such as schizophrenia. Individual differences in LI has supported the continued use of animal models to understand human mental health. In this review, we ask whether there is sufficient evidence to support the continued application of LI from animal models to human psychopathology because of the weak evidence for LI in humans. There is considerable variability in the methods used to assess LI, sustaining different theoretical accounts of the effects observed, which differ from the accepted accounts of LI as demonstrated in animals. The review shows that although there have been many experiments testing human LI, none provide the necessary experimental controls to support the conclusion that retarded responding is caused simply by preexposure to a stimulus, as has been demonstrated with animal models. Establishing this conflict, we set out a framework for future research.

Keywords: Associative learning; Latent inhibition; Masking tasks; Priming.

Fig. 1

Data redrawn from Rescorla (1971), showing change in mean suppression ratio of Phase 2 trials for a preexposed and non-preexposed group

Fig. 2

Three panels showing 2 × 2 contingency tables displaying the four possible combinations of response − outcome information. The S-O contingency is illustrated across three panels. Panel A shows a generic overview for calculating S-O contingency. S-O contingency can be viewed as a matric of four trial events, where a, b, c, and d represent the frequencies of each S-O conjunction. Contingency is determined by the difference between the likelihood of the outcome occurring in the presence of the St [a/(a + b)] with the absence of the St [c/(c + d)]. A normative model for the S-O contingency (Allan, 1980) takes the difference between these two conditional probabilities ΔP = a/(a + b) - c/(c + d). Panel B shows the S-O contingency tables for a preexposed stimulus in Phase 1 and Phase 2 of a standard LI test. Panel C shows the S-O contingency tables Phase 1 and Phase 2 combined for a preexposed and non-preexposed stimulus