Date: Feb. 26, 2020 3:00pm – 4:30pm

Location: Dunton Tower: Room 2203

Speaker: Brendan Conway-Smith

Title: Clarifying Metacognition through Computational Modeling

Abstract:

Metacognition is often described as cognition acting on itself. Metacognition is known to improve memory, reasoning, emotional regulation, and motor skills.

First, I discuss the two main barriers to understanding metacognition as within the research literature: its high abstraction and disputed terminology. Second, I introduce a method to overcome these barriers by means of computational modeling. I present my method of grounding metacognition within a computational cognitive architecture. This allows us to first define the base units of cognition, and how they come to act on themselves. These well-defined computational units are then built upward to form increasing complex cognitive processes, to arrive at metacognition. This computational view of metacognition is then connected back to the research literature and built into working models within ACT-R.

I conclude with an overview of how a computational methodology can help us better understand how cognition operates, acts on, and improves itself.