Vicente Vivanco Cepeda¹, Joshua B. Tenenbaum¹, Kevin A. Smith¹, Vivian C. Paulun¹; ¹Massachusetts Institute of Technology (MIT)

Imagine pouring a box of granola into a bowl. Are you considering hundreds of individual chunks or the motion of the group as a whole? Human perceptual limits suggest we cannot be representing the individuals, implying we simulate 'ensembles' of objects. If true, we would need to represent group physical properties beyond individual aggregates, similar to perceiving ensemble properties like color, size, or emotion. However, unlike previously discovered ensemble perception, physical properties such as mass are not directly observable. In two experiments we study whether people can perceive the mass of an ensemble, and if this representation is more than an aggregate of the properties of individual objects. Participants viewed video clips of marbles falling onto a cloth. On each trial, they had to choose which of two consecutive videos showed the larger mass. In Experiment 1, 40 participants completed two conditions: 1) comparing the masses of two individual marbles, and 2) comparing the average mass of groups of 25 marbles with different means. For the same mass ratios, we found people discriminate ensembles better than individuals (χ2(9)=37, p<.001). Does this imply ensemble processing beyond the sum of parts? In Experiment 2, 21 participants compared the average mass of a group of 25 marbles with different means against a similar set with five changed marbles. Crucially, immediately after they were asked to identify one of the marbles that had changed weight. Participants’ mass judgments were significantly better (66% correct) than can be expected if they were relying on individual information (58% correct given 33% correct identification of one marble, Haberman & Whitney (2012); binomial test, p<.001). Together this supports the concept of ensemble perception in intuitive physics, extending our understanding of how people represent and simulate sets of objects.

Acknowledgements: KAS and JBT were supported by National Science Foundation Science Technology Center Award CCF-1231216, and NSF grant 2121009