Proceedings of the Annual Meeting of the Cognitive Science Society

Previous research suggests a capacity limit in dynamic mental imagery: people simulate the motion of only one object at a time (Balaban & Ullman, 2024). This raises an obvious question: how do people imagine interactions between multiple moving objects? Here, we used a novel paradigm to examine people's estimates of the timing and location of collision events that happened either between (1) two imagined moving entities, or (2) one imagined moving entity and a static patch. We propose and evaluate two mental simulation models: In the Parallel model, both entities are advanced simultaneously until collision. In the Switch model, the mind alternates between entities, updating one for several steps before switching to the other, and back again. We found that the Switch model better predicted participants' collision time estimates in the two-entity condition relative to the one-entity condition, whereas the Parallel model failed to capture this difference (Experiment 1 and Experiment 3). Moreover, the Switch model predicted systematic biases in reported collision positions that the Parallel model did not (Experiment 2 and Experiment 4). Together, these findings suggest that the mental simulation of interactions between multiple dynamic objects is better described by a Switch model, in which the mind alternates between entities, rather than simulating them in parallel.