Rough Sets: Theoretical Aspects of Reasoning about Data
Rough Sets: Theoretical Aspects of Reasoning about Data
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
Humans can easily recognize complex objects even if values of their attributes are imprecise and often inconsistent. It is not clear how the brain processes uncertain visual information. We have tested electrophysiological activity of the visual cortex (area V4), which is responsible for shape classifications. We formulate a theory in which different visual stimuli are described through their attributes and placed into a decision table, together with the neural responses to them, which are treated as decision attributes. We assume that the brain interprets sensory input as bottom-up information which is related to hypotheses, while top-down information is related to predictions. We have divided neuronal responses into three categories: (a) Category 0 - cell response is below 20 spikes/s, which indicates that the hypothesis is rejected, (b) Category 1 - cell activity is higher than 20 spikes/s, which implies that the hypothesis is accepted, 3. Category 2 - cell response is above 40 spikes/s, which means that the hypothesis and prediction are valid. By comparing responses of different cells we have found equivalent concept classes. However, many different cells show inconsistency between their decision rules, which may suggest that parallel different decision logics may be implemented in the brain.