|
Adaptive gain control during human decision-making| old_uid | 15709 |
|---|
| title | Adaptive gain control during human decision-making |
|---|
| start_date | 2015/06/01 |
|---|
| schedule | 16h |
|---|
| online | no |
|---|
| location_info | Leolin Price Lecture Theatre |
|---|
| details | Host: Patrick Haggard |
|---|
| summary | Human performance on perceptual classification tasks is often described as approaching that of an ideal observer. However, human cognitive resources are finite, and decisions about uncertain sensory information are limited by "late" noise arising beyond the sensory stage. I will discuss the view that humans have evolved "robust" choice strategies that protect decisions against late noise, and which involve adapting the gain of information processing to focus on a subset of the feature information. Describing tasks in which observers average the feature information arising from a stream of discrete visual events (samples), I will present evidence that this gain control can be implemented both flexibly and extremely rapidly. These experiments disclose new behavioural phenomena in perceptual classification, including robust averaging, variance priming, and selective integration, and I will discuss these and their accompanying neural correlates. Finally, I will provide a computational model that accounts for these effects, and draws a link between psychophysical studies of decision formation, and cognitive accounts of the decision process that have emphasised how capacity is limited by a central bottleneck. |
|---|
| responsibles | Lawrence |
|---|
| |
|