B.3.6.3.1. Reaction times

An ANOVA was performed on correct RTs, with salience (neutral, salient target) and relevance (relevant, neutral) and perceptual similarity (similar, different) as within-subject factors. The distractor condition, of no interest for the current purpose, was not included in the analysis, so as to allow focusing on salience effects upon target's processing. For the sake of brevity, accuracy analyses were not presented, but they were consistent with RT results (see Figure 10, bottom compared to top panel).

First, the target was processed faster when it was cued (F(1,9) = 54.49, MSE = 1,001,615, p < .001) and when it was made salient (F(1,9) = 28.88, MSE = 273,978, p < .001). The effect of similarity was not significant (F(1,9) = 2.04, MSE = 52,073, p > .18). The interactions between feature similarity and relevance (F(1,9) = 1.30, MSE = 6,565, p > .28) and between feature similarity and salience (F(1,9) = 0.07, MSE = 212, p > .79) were not significant. The interaction between relevance and salience was not significant either (F(1,9) = 3.63, MSE = 14,653, p > .089).

Finally, the two-ways interaction of interest (relevance x salience x similarity) was significant (F(1,9) = 12.47, MSE = 46,479, p < .007). Paired comparisons (unilateral t-tests) showed that the target salience accelerated RTs, even when the target was cued. This was true for the similar salience (135 ms, S.E.: 22: t(9) = 6.12, p < .001), as well as for the different salience (45 ms, S.E.: 17: t(9) = 2.72, p < .012). However, one should note that this salience effects were much stronger in the former, same dimension, than in the different dimension condition (t(9) = 3.18, p < .006). This suggested that the advantage for redundancy (i.e. target salient and relevant vs. target only relevant or only salient) was stronger when both signals concerned the same dimension rather than a different one.

Figure 10: Performances in the visual search task, according to the salience condition (distractor salient, target salient, no salience) and to the salience-relevance perceptual similarity (similar, different), in terms of speed (top panel) and accuracy (bottom panel). Bars represent 1 standard error. See text for statistics. Sim.: Similar dimension; Diff.: Different dimension. Bars represent 1 S.-E.