Purchase this article with an account.
Rogelio Luna Almeida, Megan P. Roussy, Adam Sachs, Stefan Treue, Julio C. Martinez-Trujillo; Neuronal ensembles of primate Lateral Prefrontal Cortex encode spatial working memory in different frames of reference. Journal of Vision 2020;20(11):1753. doi: https://doi.org/10.1167/jov.20.11.1753.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Single neurons of the Lateral Prefrontal Cortex (LPFC) of macaques can encode spatial working memory (WM) signals. However, it is not fully understood whether ensembles of potentially interconnected neurons encode locations based on different reference frames during spatial WM. We trained two rhesus monkeys on an Oculomotor-Delayed Response task that allowed us to dissociate the memorized spatial locations between a screen-centered (spatio-topic) and a retino-centered (retino-topic) reference frames. The monkeys fixated a dot that appeared at one of 16 positions on the stimulation screen. Then a cue stimulus appeared at a different position for 1000ms. The animals kept looking at the fixation dot for another 1000ms (WM delay) and upon its removal they made a saccade to the memorized target location. Liquid reward was delivered for correct responses. We recorded the extra-cellular activity of single- and multi-units by implanting multi-electrode arrays dorsally (dLPFC) and ventrally (vLPFC) from the principal sulcus (areas 8A and 9/46), respectively. We computed the average firing rate during the WM delay for each unit and trained a linear classifier to decode the visual quadrant that included the memorized cue position in on every trial. Next, we constructed neuronal ensembles of different sizes (i. e., n = 2, 3… 88) and decoded the memorized quadrant from each ensemble. We found that best decoding accuracy for individual dLPFC and vLPFC neurons equaled 45% in the retinotopic reference frame and 40% in the spatiotopic. In contrast, best decoding accuracies yielded by ensembles of dLPFC and vLPFC cells equaled 69% and 79% in the retinotopic reference frame, respectively, and 55% and 52% in the spatiotopic reference frame. Interestingly, best decoding accuracies corresponded to ensembles sizes from ~10 to 20 neurons. Neuronal ensembles of LFPC could encode locations during spatial WM in both reference frames by using a population-based dynamic.
This PDF is available to Subscribers Only