Computer network model predicts dependence of neocortical laminar activation patterns on form of stimulation

TitleComputer network model predicts dependence of neocortical laminar activation patterns on form of stimulation
Publication TypeConference Paper
Year of Publication2010
AuthorsLytton, WW., & Shepherd G. M. G.
Conference NameSociety for Neuroscience 2010 (SFN '10)
KeywordsSFN, Society for Neuroscience
Abstract

In rodent motor cortex, the local excitatory network is dominated by layer 2/3 (L2/3) to L5A/B pathways, implying a top-down laminar organization in which these superficial layers are in a privileged position that allows them to project activity onto the middle layers. An initial linear feedforward model of the disinhibited network corresponded with physiology, demonstrating that activation of higher layers in a disinhibited network would create widespread activation of multiple laminae while activation of deep layers would not. We have now extended this model to a spiking neuronal network model simulated in NEURON. This new model suggests that the widespread multilaminar activation previously described would be most likely to occur with relatively prolonged, random activation to the upper layers. The ongoing activation provided by such prolonged activation gave superficial layers time to recruit lower layers. Paradoxically, brief strong tetanic input demonstrated the opposite pattern. In this case, superficial stimulation caused only brief activation due to the powerful recruitment of many cells in a brief, synchronized population event, suggestive of an interictal spike. This population event caused all the cells to become refractory simultaneously, precluding further activation. By contrast, tetanic stimulation at deeper layers, those corresponding to L5, provided a slower, more diffuse activation upwards into L2/3, mediated by the weaker projections in this direction. This created conditions of relatively prolonged diffuse activation of L2/3. This upward activation then led to recurrent activation back down into the lower layers where the activity had originated, producing a prolonged widespread response across laminae. Our analysis highlights basic issues in translating a linear mapping to a spiking neural network model, and our results suggest the existence of multiple modes of activity propagation in the local excitatory network in the neocortex, which are recruited in a manner that depends on both the spatial (laminar) and temporal (synchronous or disperse) pattern of input.