Abstract
When presented with a moving visual stimulus, neurons in macaque area MT encode its motion direction. When the stimulus disappears and its direction is held in working memory, these neurons no longer encode motion direction. In contrast, neurons in the dorsolateral prefrontal cortex (dlPFC) encode motion direction both during the stimulus presentation and during working memory maintenance in the absence of visual input. One question that has not yet been systematically addressed is whether motion direction representations in dlPFC are stronger when the stimulus remains visually available or when it disappears and its direction is remembered. To examine this, we recorded the activity of 155 dlPFC neurons from two macaques while they performed two alternative conditions of a match-to-sample task requiring the comparison between the motion directions of a sample stimulus and a subsequent test stimulus. In the memory condition, the sample was presented for 1 s and, after a delay, followed by the test. In the no-memory condition, the sample remained visible until and during the test presentation, and therefore working memory was not required. For each neuron, we quantified sample direction representation strength using ROC analysis and compared it between the delay period of the memory condition and the equivalent period of the no-memory condition. We found that in approximately half of the direction-selective neurons, representation strength was higher when the sample remained visually available than when it was remembered. Interestingly, in the remaining half, representations were stronger when the sample direction was remembered than when the sample remained present. Our results show not only that dlPFC neurons can encode visual representations in the absence of sensory input, but also that in many neurons, the strength of these representations is in fact reduced in the presence of sensory input, when working memory is not required.
Meeting abstract presented at VSS 2012