Effects of input correlations on neuronal and network activity

Abstract

We are interested in how input correlations influence the dynamics of neurons and networks. Temporal correlation of input spike trains into neurons can arise from the sharing of inputs or from projecting neurons which are correlated in their spiking activity. Spike correlation between neurons can also emerge from network interaction, such as recurrent projections. Neuronal activity in different parts of the brain has been found to be modulated in terms of their firing rate and correlation during cognitive and behavioral tasks. In the first part, I will talk about the effect of intrinsic heterogeneity on the activity of a population of model neurons (Yim et al., 2013). In the high input regime, the sum of synaptic inputs to a neuron can be approximated by a fluctuating input noise. By rescaling the dynamical equation, we derive mathematical relations between multiple neuronal parameters and a fluctuating input noise. To this end, common input to heterogeneous neurons is conceived as an identical noise with neuron- specific mean and variance. The output firing rate of a neuron is largely shaped by the mean level of the noise, whereas the distributed values of the variance give rise to different degrees of imprecise spiking. As a consequence, the neuronal output rates can differ considerably, and their relative spike timing becomes desynchronized. Our theory can explain experimental findings from in vitro recordings (Padmanabhan and Urban, 2010). The second part is about the significance of input correlations for signal representation in the striatum (Yim et al., 2011). Cortico-striatal projection neurons outnumber striatal neurons by a factor 10, with each striatal neuron receiving massive synaptic inputs from the cortex. However, anatomical evidence suggests that cortico-striatal projections are structured such that neighboring striatal neurons are unlikely to share their inputs. To understand the functional consequences of input correlations on the signal representation of cortical activity in the striatum, we simulated a large-scale network model of striatum and cortico-striatal projections. We suggest that for the network architecture of the striatum, there is a preferred cortico-striatal input configuration for optimal signal representation, setting the stage for action selection presumably implemented in the basal ganglia.