Achievement

Trainee Isaac Caruthers' research aims to identify the neural codes that represent distinct objects in natural auditory scenes. He records the activity of many individual neurons in the auditory cortex of the awake rat as it is exposed to various stimuli, including vocalizations. Recording from multiple neurons simultaneously, he has access to how the correlations between neurons may be incorporated into the neural code. For example, he trained a support vector machine to distinguish between different vocalizations based only on the responses of a set of simultaneously recorded neurons. He then compared the accuracy of this classifier when given the original data to the accuracy when given the responses to different trials from different neurons. The fact that the accuracy of the classifier decreases when the response trials are shuffled indicates that the correlations in the neural responses may be tuned to prevent noise in the cells from impeding the classification of vocalizations.