Synaptic properties can dominate cortical output computations

Cortical output - information leaving the cortex through axons projecting to subcortical targets - mainly originates from pyramidal tract (PT) neurons. The determinants of spike generation in PT neurons are complex, as PT neurons receive a broad variety of local and long-range input, have extensive dendritic trees and express a plethora of active conductances, causing highly nonlinear dynamics. The interplay between these components and their importance for the overall computation PT cells perform remains poorly understood.

Here, we characterize a regime of spatiotemporal excitatory and inhibitory synaptic input patterns, for which the computation is mainly determined by synaptic properties. For such input, biophysical and morphological variability only have a minimal effect on the response. Furthermore, the computation becomes analytically tractable. We show, that PT neurons are exposed to such spatiotemporal synaptic input patterns during the onset of sensory stimuli in vivo.

This enables the PT population to respond to the stimulus robustly and with high temporal accuracy, despite the large biophysical and morphological variability between individual cells. Furthermore, we show, that in vivo responses can be accurately predicted utilizing an analytical model. This can be a starting point to investigate which computations underly cortical output and - therefore - which information is encoded in the so-far enigmatic responses of the PT neuron population.