Past events: Aurelio Cortese. Metacognitive resources for efficient learning

What: Metacognitive resources for efficient learning

Where:  BCBL auditorium (and BCBL zoom room 2)

Who: Aurelio Cortese. PhD, Chief researcher and Deputy Department Head, Computational Neuroscience Labs, Advanced Telecommunications Research Institute International, Kyoto - Japan

When:  Wednesday, March 23rd at 2:30 PM.

How do we learn from limited experience? A central issue in reinforcement learning (RL) is the ‘curse-of-dimensionality’, which arises when a problem's degrees-of-freedom are much larger than the number of available training samples. In the brain, higher cognitive functions such as metacognition or abstraction may provide a biological solution by generating low dimensional representations on which RL can operate. In this talk I will present our recent work in which we used neuroimaging and modeling to investigate the neural and computational basis of efficient RL. We found that people can learn remarkably complex task structures non-consciously, and that metacognition appears tightly coupled to this learning ability. Furthermore, when people use an explicit (conscius) policy to select relevant information, learning is accelerated by abstractions. The prefrontal cortex is differentially involved in separate aspects of learning: dorsolateral prefrontal cortex encodes metacognitive processes and is coupled with the basal ganglia, while ventromedial prefrontal cortex engages in valuation and abstraction and is coupled with sensory cortices. I will discuss the implications of these findings and introduce a new framework for metacognitive artificial intelligence.