Object Permanence (OP) is the understanding that objects continue to exist when not directly observable. To date, this ability has proven difficult to build into AI systems, with Deep Reinforcement Learning (DRL) systems performing significantly worse than human children. Here, DRL Agents, PPO and Dreamer-v3 were tested against a number of comparators (Human children, random agents and hard coded Heuristic agents) on 3 object permanence tasks (OP) and a range of control tasks. As expected, the children performed well across all tasks, while performance of the DRL agents was mixed. Overall the pattern of performance across OP and control tasks did not suggest that any agent tested except children showed evidence of robust OP.

Authors:

Konstantinos Voudouris: University of Cambridge; Jason Darwin Liu: University of Cambridge; Natasza Siwinska: University of Cambridge; Wout Schellaert: Universitat Politècnica de València; Lucy G Cheke: University of Cambridge