Neurons in the frontal eye field (FEF) and the lateral intraparietal area (LIP)
are believed to play a role in the planning and execution of saccades. It has
been suggested that neurons in these areas represent a plan for an upcoming saccade
vector. Our aim is to elucidate how FEF and LIP neurons store and modify
Attractor network models of spatial memory predict that excitatory connections between similarly tuned neurons are capable of driving an ensemble of neurons in the absence of external input. In such models, stored information drifts or degrades the longer it is maintained. A similar effect is also seen in the variability memory-guided saccades made by monkeys the longer they are asked to remember. The neural correlates of this behavior have not been clearly identified. We have measured the responses of FEF neurons in a delayed saccade task with delays of different lengths and have examined the direction tuning curves as a function of time in these neurons. Preliminary data indicates that tuning width of some FEF neurons broadens over the delay, consistent with a degradation of spatial memories over time in FEF.
FEF neurons demonstrate persistent activity that is thought to be critical for maintaining a spatial memory "on line" however it is unclear to what extent this area contributes to memory-guided behavior. We have performed subthreshold intracortical microstimulation (ICMS) during a delayed saccade task to perturb saccade plans represented in FEF. Stimulation at sites in FEF caused a systematic deviation of saccade endpoints in a direction antiparallel to evoked saccades elicited via suprathreshold stimulation at the site. Interestingly, this result suggests microstimulation introduces an artificial signal indicating that the eyes have moved (most likely a corollary discharge signal), causing the animal to update its saccade plan to compensate for the fictive movement. Control experiments have shown that stimulation effects only occur when stimulation is applied during a delay (memory) period and that effects are reduced when the animal is trained to ignore intervening eye movements during the delay, which are both consistent with a stimulation effect on a corollary discharge pathway.
We performed similar experiments in LIP and observed rare and inconsitent effects. Significant deviations from control were observed in only 8 of 43 LIP sites in two animals. In the population of LIP sites where significant memory activity was recorded (n=17), we observed a small but significant deviation in a direction away from the RF, similar to our FEF finding, that was also strongest for adjacent target directions. No effects were observed in the population of LIP sites
where only visual responses were recorded.
Delay Period Microstimulation in the Frontal Eye Fields Updates Spatial Memories
White RL, Snyder LH.
Soc Neuro Abstr, 2004.
Delay period microstimulation perturbs memory-guided saccade behavior.
White RL, Snyder LH.
COSYNE Abstr, 2004.
During a gaze shift, the retinotopic representation of a target location that is fixed in the world (world-fixed reference frame) must be updated, while the representation of a target fixed relative to the center of gaze (gaze-fixed) must remain constant. To investigate how such computations might be performed, we trained a recurrent neural network to store and update a spatial location based on a gaze perturbation signal, and to do so flexibly based on a contextual cue. We found the network utilized an eye-centered coding scheme that correctly updated for world-fixed targets, but suppressed updating for gaze-fixed targets. Output was less precise for world-fixed targets. Furthermore, the network strongly preferred gaze velocity to gaze position signals for updating, and gaze position gain fields were not present when velocity signals were available. This implies that gain fields may not be obligatorily involved in spatial updating.
A Neural Network Model of Flexible Spatial Updating
White RL, Snyder LH
J Neurophysiol 91, 1608-1619 (2004)