|
Optimal neuronal decoding predicts Human perceptual decisions| old_uid | 367 |
|---|
| title | Optimal neuronal decoding predicts Human perceptual decisions |
|---|
| start_date | 2005/12/13 |
|---|
| schedule | 16h |
|---|
| online | no |
|---|
| summary | Processing of sensory information about our environment relies on the concerted activity of thousands of single neurons in the brain. Different models propose how this information is read-out and combined from the complex patterns of neuronal activity to support perceptual decision mechanisms. Here is outlined a new approach, a model which decodes neuronal population responses to provide (1) exact predictions of perceived sensory stimuli and (2) the accuracy, or sensory threshold, associated with subjective perception. The model is based solely on Fisher Information as the upper theoretical limit of available information about a sensory stimulus. It is postulated that actual decisions about a perceived stimulus can be based on a comparison of Fisher Information from pools of neurons encoding either of two alternatives in a discrimination task. The model is applied to two visual features, orientation and motion discrimination, in a paradigm requiring the discrimination of motion/orientation in the presence of distracting motion/orientation. Specifically, the model predicts the pattern of perceptual misjudgments also known as motion/orientation repulsion, induced by the task irrelevant signals. In addition, it allows to predict the accuracy of discrimination performance. These results show that perceptual misjudgments and threshold elevations during motion/orientation discrimination are based on neuronal populations encoding information in the theoretical optimal way. Moreover, our approach allows to reverse engineer the tuning widths of human motion and orientation detectors. |
|---|
| oncancel | Attention : changement d’horaire |
|---|
| responsibles | Gutkin |
|---|
| |
|