Proceedings of the Annual Meeting of the Cognitive Science Society

To better understand the underlying mechanisms of individual and ensemble perception in naturalistic scenes, we compared three bayesian resource-rational models on experimental data (from 27 healthy adults): the 'Individual Encoding Model' (IEM), a variant of the summation model; the 'Ensemble Encoding Model' (EEM), related to the automatic averaging model; and the 'Task Adapted Encoding Model' (TAEM), a flexible combination of both models that adapts to task demands. In the experiment, participants encoded and reproduced either an individual object position or an ensemble position (group centroid) in a 3D-rendered scene using a computer mouse. In both tasks, we manipulated set size (3, 6, 10 objects) and presentation time (50, 100, 800 ms). The EEM and TAEM generally explained the human behavioral data best. We conclude that, in naturalistic scenes, the choice between individual versus ensemble perception is likely driven by the more compact scene representation of the ensemble model.