UC San Diego

Memory and spatial navigation involve complex neural processes that depend on accurate and coordinated firing of neurons within the hippocampus and medial entorhinal cortex (mEC). Neurons within these regions are tuned to various features of the environment, including space and context, and incorrect firing is hypothesized to lead to catastrophic errors in our ability to generate new, stable, and coherent memories. There are numerous methods in which the brain might establish and maintain reliable firing, and here, we examined different network mechanisms that determine spike timing. The first property that significantly influences the firing of neurons is the network connectivity of cells within and between brain regions. To understand how networks are organized within the mEC, we employed rabies virus-mediated retrograde trading and determined that projections within the superficial layers of the mEC are organized as cortical columns, parallel to the radial axis of the brain. Next, with information on how these local projections are organized, we performed transient, local perturbations of these circuits and examined whether the precise spatial firing of neurons within the mEC is dependent on these local circuits. While we found small errors in the spatial firing of grid cells following each perturbation, these errors were brief and not cumulative, suggesting robust mechanisms for afferent inputs to maintain firing.

In addition to the connectivity between neurons, a key factor coordinating spike timing, especially between brain regions, are global oscillations that act as an external pacemaker. Within the context of the mEC and hippocampus, these are called theta oscillations, and most neurons within these brain regions are rhythmically paced to theta oscillations. To test how these oscillations might coordinate spike timing, we artificially accelerated the frequency of theta oscillations and determined that neurons within the hippocampus respond to the stimulation by shifting their oscillation frequencies to match the stimulation frequency. Finally, we provide an example of how a functional readout of coordinated spiking between neurons involving both the above described mechanisms – a local network dependent rate code, and theta oscillation dependent temporal code – may be used on a population level to generate complex internal representations.