Abstract
Conceptual and perceptual similarity lead to recognition errors with verbal material, objects, faces, and scenes: the more similar the items, the more confusion in memory. Surprisingly, whether perceptual similarity between independent scenes alone affects scene representation and confusion in memory has never been investigated. Here, we used a continuous perceptual space to evaluate this range of similarity between independent pictures, and explored whether scene recognition is affected by perceptual similarity. In Experiment 1, we measured the degree of perceptual confusion between 200 targets and associated decoys, orthogonally matching the targets on visual layout and conceptual gist. In Experiments 2 and 3, we used the resulting confusability index to assess whether perceptual similarity affects observers' performance on a 2-AFC (Exp.2) and an old/new (Exp.3) tests. Results of Pearson correlation tests addressing the relationship between the false alarm rates and confusability index showed that the rate of false alarms was not correlated to the perceptual similarity for the 2-AFC test. Does the confusability index fail to capture perceptual similarity between scenes? It does not appear to be the case, as the false alarm rate of the single recognition test was significantly correlated to the confusability index. These results strikingly contrast with the well-known effect of conceptual similarity effect, namely recognition errors increase with increased conceptual similarity. Here, on the contrary, results suggest that scene representation actually seems detailed enough in memory to support accurate discrimination of the target scenes among distractors, whatever distractors are very similar to the target scenes or not.