Strategies for processing orbitofrontal cortex neuronal activity for the prediction of rat response in an odor-cued two-alternative choice discrimination task with regression models

Abstract

The orbitofrontal cortex (OFC) has long been implicated in decision making, including in a study on rats performing odor-cued two-alternative choice discrimination task, in which some of recorded neurons were found to have been selective for making decision about the location of the goal. The present paper seeks to use neuronal activity data to train a machine learning model to predict decision in this task, comparing performance on different strategies for processing spike trains. The results support the idea that information on the decision made is coded in communication between multiple neurons, and show how performance on the model can be improved with carefully selected input.