“School of Cognitive”
Back to Papers HomeBack to Papers of School of Cognitive
| Paper IPM / Cognitive / 11233 |
|
||||||||||
| Abstract: | |||||||||||
|
The transformation from sensory representation to sensory decision involves the formation of categories. These experiments were designed to test the idea that the decoding processes - whereby categories are extracted from a continuous dimension of representations - might adapt according to categorical boundaries imposed on the sensory system by external rules. If it is repeatedly suggested to a subject that stimuli that initially feel different in fact belong to the same category, will those stimuli begin to feel identical? To answer this question, we employed a flutter-vibration discrimination task in which feedback and rewards were biased. We used a psychophysical test to evaluate subjects sensory capacities before and after training to look for any changes in decision criterion or in sensitivity.
The Just Noticeable Difference (JND) in frequency, for vibrations applied to the finger tip, was determined using the staircase paradigm prior to training. Then, during training, subjects were asked to compare two vibrations presented to the same finger tip (one forced choice) and state whether they were ?the same? or ?different? in frequency. On each trial, subjects lost or gained points in a biased manner, as follows: the feedback was drawn from a normally distributed function with a standard deviation twice as wide as subjects' JND. If subjects' response matched the outcome of the response drawn from the distribution, they were rewarded; otherwise, they lost points on that trial. This procedure meant that stimulus pairs which were discriminable according to the JND were commonly scored as ?the same? by the computer, and subjects lost points if they reported them as discriminable. In other words, the feedback rewarded the subjects (with a certain probability) for not feeling the frequency difference between stimulus pairs. However, the subjects were not aware of the bias in feedback. After training, JND was again calculated and increased in a statistically significant manner. In other words, pairs of vibrations that, before training, could be discriminated and therefore belonged to different sensory categories, were felt by the subjects after training as being identical. Further analyses showed that the JND changes could not be explained based on a change in the subject?s decision criterion. These findings suggest that reward distribution can adaptively shift the sensory representation of stimuli. It appears that as subjects began to maximize the reward probability during training, the transformation from sensory representation to sensory decision was altered. Subjects formed new sensory categories for vibration frequency according to external rules.
Download TeX format |
|||||||||||
| back to top | |||||||||||
