|
Metacognition and (Artificial) Consciousness| title | Metacognition and (Artificial) Consciousness |
|---|
| start_date | 2025/11/10 |
|---|
| schedule | 15h15 |
|---|
| online | no |
|---|
| location_info | Room B10 |
|---|
| summary | will describe recent progress on understanding the computational and neural basis of human metacognition, and explore the application of these findings to theories of consciousness in biological and artificial systems. I first outline evidence for a hierarchical framework for metacognition in which self-performance is tracked over multiple interacting timescales, allowing people to construct and update internal models of their skills and abilities in a range of domains. I then motivate a connection between (local) metacognition and conscious experience, on both evolutionary and theoretical grounds. This view proposes that a conscious percept is one that it is tagged as a reliable (confident) estimate of the environment for use in planning and decision-making. I’ll recap various lines of evidence in support of this view – including how conscious perception unfolds during virtual maze navigation and how judgments of reality (vs. imagination) are formed. I will end by exploring how a functionalist approach to consciousness and metacognition helps us better understand and characterise so-called indicators of consciousness in artificial systems. |
|---|
| responsibles | Allen |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2025/10/30 08:20 UTC |
| |
|