Algebraic Patterns as Ensemble Representations

AbstractObservers rapidly extract summary statistics from sets of visually presented items, like the mean size of a set of circles, or the mean expression of a set of faces. Their excellent ability to report summary statistics stands in contrast to near-chance representation of any of the individuals. Here we asked to what extent this ‘ensemble perception’ signature extends to a more abstract property: relations among elements. Participants watched ten unique animations of visually patterned objects (hereafter, ‘shapes’) colliding with each other and producing a new shape. Collisions conformed to ABA patterns, such that the result shape always matched one of the collider shapes. Recognition tests showed that participants accurately recalled the collisions they saw, but also falsely accepted foils which conformed to the ABA pattern but which were not in fact specifically seen (were rearrangements of the original shapes across collisions). On the other hand, they were much less likely to accept foils which did not conform to the pattern, but were equally distinct rearrangements (e.g., AAB). This suggests that participants represented the overall, common pattern better than the specifics of what they saw; the superior encoding of the summary relative to the individuals thus applies to summaries of relations. However, in contrast to prior findings with visual features, we did not find that recall of individual patterns was entirely at chance. Our paradigm offers a way to pursue future questions such as the pressures and motivations which might govern the trade-off between summarizing evidence vs. retaining individual experiences.

Return to previous page