"A push-pull model of prefrontal cortex during a sequential discrimination task"
Sequential discrimination tasks are widely used in psychophysical studies. In a typical such task, a subject is presented with a first stimulus (f1), and then, after a delay of a few seconds, with a second stimulus (f2), after which the subject must make a decision based on a comparison of the two (f2 > f1?). Sequential discrimination thus requires at least three components: loading a stimulus with a particular value (f1) into the working memory system, storing that value over a few seconds, and then comparing the memory of f1 to f2 when the second stimulus f2 is presented. For a somatosensory variant of this task, the neurophysiological pathway has been largely identified. In particular, neurons that participate in all three components of the task have been found in the prefrontal cortex of macaques [Romo et al., 1999; Brody et al., unpublished observations]. Neurons that exhibit activity as a function of the first stimulus (f1) fall into two classes: those with firing rate proportional to f1 (positive tuning) and those with firing rate proportional to -f1 (negative tuning). Based on these findings, we set up a simple neural network model for the prefrontal cortex. The network consists of oppositely tuned neurons that form a line attractor and are therefore able to store the value of a single stimulus (such as f1). We show how the input of neurons from somatosensory cortex S2 manipulates the attractor dynamics of the network model so that stimulus f1 can be loaded or that stimulus f2 can be compared to the memory of f1. The model agrees with key aspects of the electrophysiological evidence and is able to carry out all three components of the task (i.e., loading, storing, comparing) within a single, integrated, framework.