Multiplexed allocentric and egocentric signals in the primate frontal eye fields during a cue-conflict saccade task

Abstract

Allocentric (landmark-centered) and egocentric spatial cues are optimally integrated for visuomotor behavior (Byrne et al. J. Neurophysiol. 2010; Fiehler et al. Front. Hum. Neurosci. 2014), but the neural underpinnings of such integration are unknown. We hypothesized that this occurs at an early cortical level, such that frontal cortex output would code an integrated, multiplexed motor command. To test this, we recorded 173 frontal eye field (FEF) neurons in two Rhesus macaques trained on a cue-conflict saccade task, where a visual landmark shifted during the memory interval between seeing and acquiring a target. Visual and motor response fields preferentially coded target position (T) and future gaze (G) in eye coordinates, respectively (Sajad et al. Cereb. Cortex 2015), but gaze shifted 37 % toward a virtual target (T’) fixed to the landmark (Li et al. J. Vis. 2010). To test how the FEF coded this shift, we determined the best fits of spatially tuned response fields along an allocentric spatial continuum (T-T’) and plotted these as a function of their best fits along an egocentric (T-G) continuum. A slope/bias of zero (found in control visual responses) would suggest no allocentric influence, an upward bias would indicate uncorrelated multiplexing, and a positive slope would indicate integrated multiplexing. The population of neurons with motor responses (n = 116) had a significant slope (0.35 ± 0.08, R2 = 0.13) and no significant bias. Segregated visuomotor cells and motor-only cells had significant slopes (0.34 ± 0.13, R2 = 0.1; 0.45 ± 0.13, R2 = 0.22 respectively), and significantly opposed biases (0.13 ± 0.12 and −0.18 ± 0.13 respectively). These results suggest that allocentric and egocentric signals balance and converge at the level of frontal cortex output, such that the FEF motor burst provides an integrated estimate of optimal behavior to the subcortical structures for gaze control.

Acknowledgement: Canadian Institutes for Health Research, Canada Research Chair Program, Canada First Research Excellence Fund