Dynamics of heading and choice-related signals in areas PIVC, VIP and MSTd

FERNANDEZ LEON, J.A.; ROSENBERG, A.; CHEN, A.; ZAIDEL, A.; DEANGELIS, D.; ANGELAKI, D.

Chicago

Conferencia; 2015 SfN Neuroscience; 2015

Society for Neuroscience

Recent studies reveal that multiple cortical areas in nonhuman primates represent heading. However, little is known about the dynamics by which heading and choice are represented in neural activity. To disassociate the time varying contributions of these signals to neural population activity, we trained macaque monkeys to report the direction of self-motion relative to straight ahead in either visual or vestibular heading discrimination tasks. Using electrophysiological techniques, we quantified the contributions of heading and choice to neural responses in two areas that exhibit multimodal (visualvestibular) heading signals (dorsal medial superior temporal, MSTd; ventral intraparietal, VIP) and in an area (parietoinsularvestibular cortex, PIVC) with only vestibular activity. To separate headingandchoicerelated components of each cell?s response dynamics, we computed partialcorrelations between spike counts, heading, and choice, and then squared to calculate the accounted variance between neural activity and heading/choice. By averaging the squared partial correlation coefficients across all cells within an area, we could estimate population dynamics explained by heading and choice. Vestibular heading signals were similar in PIVC and MSTd, and slightly weaker in VIP. Choice-related activity was strongest in VIP, with MSTd showing the weakest choice activity. Similar observations were found for visual heading signals in VIP and MSTd. Targeted dimensionality reduction techniques were used to further examine how much variance in the population responses could be explained by the velocity and acceleration components, as well as choice. For vestibular signals, the time course of the targeted dimensions revealed that cell activities followed the dynamics of the velocity and acceleration components of the heading signals. The largest contribution of velocity signals was in MSTd, with weaker contributions in PIVC and VIP. Acceleration signals were strongest in PIVC and MSTd, and weakest in VIP. Corroborating the partial correlation analysis, choicerelated activity was strongest in VIP and weakest in MSTd. Similar results for the visual condition were found except that the acceleration component was not observed in MSTd and VIP (PIVC does not show visual responses). We conclude that visual-vestibular signals are differentially represented across areas, with the strongest vestibular acceleration in PIVC, visual-vestibular components of velocity in MSTd and choice in VIP. Our dimensionality reduction analyses support this observation because the population activities of these areas occupy different regions of the state space.