B.3.6.4.1. Salience - Relevance integration

The main issue of the present research was an hypothetical prediction issued from the biased competition theory (Desimone & Duncan, 1995, Duncan et al., 1997). The idea was that, if endogenous and exogenous attentional processes took effects, as predicted, in the perceptual system, then one might expect a stronger integration when they concerned the same (vs. a different) perceptual dimension. First, RTs were faster when a target was salient and relevant (a case of signal redundancy) than when only salient or only relevant. This redundancy benefit appeared for both similarity condition, but, as expected, was stronger for the similar feature condition. This confirmed that the combination between salience and relevance effects could be greater when they concerned a similar feature. Furthermore, the Miller's (1982) inequality evidenced that this was a genuine integration, and not merely a statistical facilitation following a "horse-race" between two independent processes. This test proved that the two signals (endogenous and exogenous) were actually genuinely combined at some locus. These results were highly consistent with the neurophysiological data reported in introduction, as they suggested that both salience and relevance effects could concur to bias the competition in the perceptual system. This represents a genuine integration, which should take place before the "attentional selection".