Physical Problem-Solving Across Timescales

Explore a fascinating lecture on the computational and cognitive foundations of flexible physical problem-solving across multiple timescales of experience. Delve into research combining perspectives from cognitive science, machine learning, and robotics to understand how humans efficiently use and repurpose objects as tools. Discover the importance of structured knowledge, learning, and mental simulation in explaining human flexibility and efficiency in physical problem-solving. Examine how actions must be structured in relational, object-oriented ways and how intuitive physics plays a critical role. Learn about the process of searching through and adapting actions over multiple timescales, from a few attempts within a problem to a lifetime of embodied experience. Gain insights into the universality of tool use in human cultures and its rarity among other animals, highlighting the complexity and intelligence behind these behaviors.

Kelsey Allen - Physical problem-solving across timescales - IPAM at UCLA