One of the central challenges in the cognitive and brain sciences is understanding how organisms dynamically perceive and behave in response to other individuals. From classical theories in social cognition, cybernetics, and developmental psychology to contemporary debates concerning the mirror system, theory of mind, and game theory, we find a common question: How do organisms coordinate behavior through reciprocal interactions? Detecting social contingency—the capacity to sense and respond differentially to animate versus inanimate entities—is a crucial skill for survival and reproduction, scaffolding more complex processes such as cooperation, competition, and collective behavior. In order to investigate and define the minimal neural and behavioral mechanisms underlying this capacity, we employ a minimal “perceptual crossing” task in which both human participants and embodied artificial agents must identify their respective partners within a perceptually ambiguous environment.
(A) The task takes place in a 1-dimensional where two agents face each other. Each agent can sense the other's (A), of the other's avatar (S), and a fixed object (F). (B) Each agent has (cyan) that can send to all N neurons (black). The neurons in are interconnected, including self-connections (not depicted). The output from one neuron drives the left motor and another neuron drives the right motor (magenta).
The code that produces these simulations is publicly available at the following Github repository: see here.