Speaker
Ryan Cubero
Description
How a neuron responds to complex stimuli, behaviors, and tasks can
encompass a wide range of time scales. Understanding how information
is represented in these responses across multiple temporal resolutions
then requires measures that go beyond imposing symmetry constraints
on the neuron’s tuning curves. In this study, we propose a non-
parametric, model-free indicator – which we call multiscale relevance
(MSR) – to quantify the dynamical variability of neural spiking across
multiple time scales. This allows us to select relevant neurons using
only the time stamps of the
spiking activity without resorting to any a priori external covariate or
any specific symmetries in the neurons’ tuning curves. This fully
featureless selection is done by identifying neurons that have broad
and non-trivial distribution of spike frequencies across a broad range of
time scales. When applied to neural data from the medial entorhinal
cortex, and from the thalamic and post-subicular regions of freely-
behaving rodents, we found that neurons having low MSR tend to have
low mutual information and low firing sparsity across the external
correlates that are believed to be
encoded by the region of the brain where the recordings were made. In
addition, neurons with high MSR contain significant information on
spatial navigation and allow to decode spatial position or head direction
as efficiently as those neurons whose responses have high mutual
information with the covariate being decoded. With these results, we
propose that the MSR can be used as an unsupervised method to rank
and select information-rich neurons from a heterogeneous population
without the need to appeal to any a priori external covariate.
Primary author
Ryan Cubero
