Abstract: Neural activity with slow dynamics, such as persistent and ramping activity, maintains representations that bridge past and future events, often over many seconds. Persistent activity in frontal cortex is a correlate of working memory and motor planning. Network models can produce persistent activity, but the positive feedback that is critical for slow dynamics causes severe sensitivity to perturbations.

Consistently, under some conditions cortical activity can be highly sensitive to perturbations. We are using electrophysiology, cellular imaging, quantified optogenetic perturbations and mouse behavior to probe robustness of persistent activity related to planning movements.

In mice performing a delayed response task premotor cortex neurons show movement-specific ramping activity during the delay epoch, which is causally related to specific future movements. This preparatory activity is remarkably robust to large-scale unilateral optogenetic

perturbations: detailed dynamics that drive specific future movements are quickly and selectively restored by the network. Our data imply that perturbations in one part of the network are corrected by information from other parts. More generally, preparatory activity is distributed in a redundant manner across weakly coupled modules. These are the same principles used to build robustness into engineered control systems. Our studies therefore provide an example of consilience between neuroscience and engineering. Network models incorporating these principles show robustness that is consistent with our data.