Perfect timing: Computer simulations help to disentangle different ways to drive brain cells
Click to enlarge. Caption see below.
The research focused on the role of correlations, a common measure for similarity in the temporal structure of neuronal signals. Man Yi Yim and her colleagues found, among other things, that the temporal organization of the inputs can effectively render a brain cell either highly active or totally inactive. They report their findings in the current issue of the Journal of Computational Neuroscience.
The general mechanisms described in their paper also allow to reconcile some hard-to-understand findings in recent experiments. The timing of inputs also determines the timing of outputs, in addition to the activity level. Two kinds of inputs are being considered in the work of Yim and colleagues: spike-driving (SD) and non-spike-driving (NSD) inputs. SD inputs are strong inputs that are very likely to drive a brain cell to spike, while NSD inputs are not. In computer models, various combinations of these two types of inputs could be tested and their impact on brain cells could be predicted. The “inverse problem” was also treated: Given the activity of brain cells as observed in experiments, the structure of the unobserved inputs can be reconstructed to some degree. In particular, the simulation result in the computer model for correlated NSD inputs, but uncorrelated SD inputs, is consistent with the experimental observations in one type of brain cells in the somatosensory cortex of rats. Therefore, this work predicts that neighboring brain cells possibly receive an ongoing bombardment of correlated background inputs, and occasionally uncorrelated strong inputs.
Figure Legend
Scheme of the structure of inputs to a pair of brain cells that may explain certain experimental findings in the somatosensory cortex of rats.
Original publication (open access):
Yim MY, Kumar A, Aertsen A, Rotter S (2014) Impact of correlated inputs to neurons: modeling observations from in vivo intracellular recordings, to appear in J Comput Neurosci. doi: 10.1007/s10827-014-0502-z