Partitioning the Perception of Physical and Social Events Within a Unified Psychological Space

AbstractHumans demonstrate remarkable abilities to perceive physical and social events based on very limited information (e.g., movements of a few simple geometric shapes). However, the computational mechanisms underlying intuitive physics and social perception remain unclear. In an effort to identify the key computational components, we propose a unified psychological space that reveals the partition between the perception of physical events involving inanimate objects and the perception of social events involving human interactions with other agents. This unified space consists of two prominent dimensions: an intuitive sense of whether physical laws are obeyed or violated; and an impression of whether an agent possesses intentions, as inferred from movements. We adopt a physics engine and a deep reinforcement learning model to synthesize a rich set of motion patterns. In two experiments, human judgments were used to demonstrate that the constructed psychological space successfully partitions human perception of physical versus social events.


Return to previous page