B.1.5.3. Discussion

First, this experiment replicated the results of the previous experiments relative to salience effects in the random condition, which were highly similar in the three experiments. The large item was processed faster than the medium, which was processed faster than the small one. This remained true for the first quintiles, but failed to reach significance for the longest quintiles, contrary to Experiments 1 and 2. In the relevance conditions, the relevant feature was cued block-wise, probably combining relevance with perceptual priming effects. These results presented a similar, but in some ways exaggerated, pattern, relative to Experiment 2.

Perceptual priming seemed to increase the size effects between the two irrelevant items, in both (large- and small) relevance conditions. For the irrelevant items, the RTs were inversely related to the similarity between the target and the relevant item. Indeed, the medium item was processed much faster than the small item in the large-relevant and than the large item in the small-relevant condition (see Figure 3). These effects were stronger than in Experiment 2 (where they were significant only in large-relevant condition). Indeed, in the large-relevant condition, the difference between small and medium items increased from 38 ms, CI_.95 [9, 67] (all CIs, i.e. Confidence Intervals, are at 95%) in trial-wise cueing experiment to 142 ms, CI_.95 [87, 197] in block-wise cueing experiment. In the small-relevant condition, the difference between large and medium items increased from 30 ms, CI_.95 [-11, 70] in trial-wise experiment to 157 ms, CI_.95 [85, 230] in block-wise experiment.

Thus, perceptual priming seemed to affect primarily the processing of the "not cued" items. These increased size effects could not be attributed to direct priming effects. Indeed, in both cued conditions, the two irrelevant items had the same probability to be the target. Moreover, the effects were reversed from large- to small-relevant condition. This pattern of results rather suggested that perceptual priming effects benefited to the targets according their similarity with the prime (i.e. the most frequent target).

In the present experiments, all the items were very similar, as they had an identical form, and varied only by size. Biederman and Cooper (1992) showed that perceptual priming effects could persist across size variations. Therefore, similarity-based perceptual priming could likely have occurred in the present experiments, accounting at least partly for the differences between RTs in Experiment 2 and 3. For instance, in the small-relevant condition, the small item was more often the target than both other squares. Thus, it benefited certainly from priming effects, for being selected more often and longer than any other item throughout the block of trials. Then, the medium item might have benefited from priming effects more than the large item, owing to this higher similarity with the most frequent target (the small square, in this example).