Human Cognition Talk: Simulation and symbols: a framework for physical common sense

As humans we are experts at understanding and interacting with our physical world: we can judge whether a stack of dishes will topple, pour coffee from a new carafe without spilling, and figure out whether a box is empty or filled with heavy books by how it sits on the sofa. We find these capabilities so simple we call them common sense, yet accomplishing these tasks requires coordination between perception, reasoning, and planning in order to predict future outcomes, make inferences about unobserved factors, and select appropriate actions. The cognitive systems that underlie this type of physical common sense have long been debated, with one theory suggesting that they arise from mental simulations over internal world models, while another theory suggests that they result from manipulating abstract symbols to reason about physical outcomes. Here I propose that people use both of these cognitive systems for physical common sense, combining or trading-off between them to efficiently solve day-to-day problems. I will focus on one line of research in which I first demonstrate that we do have multiple systems for conceptualizing physics, and then propose and test a framework for how we combine those systems. I will end by discussing other ongoing and future work to describe how simulation and symbols combine to support physical common sense more generally.

Event Type: 
Berkeley Way West
Wednesday, January 25, 2023
Event Sponsor: 
Psychology, Department of
Event Speakers: 
Kevin Smith