This project will examine the neuronal mechanisms at work in a task of decision making performed by monkeys.
In particular, it focuses on the transformation of neuronal representations stimulus information in motor response.
It aims to uncover neuronal mechanisms at work and interpret in a model-based fashion electrophysiological recordings.
The developed framework will be applied to laminar recordings of neuronal activity in the pre-motor and motor cortex, which take part in the movement planning and execution.
The project comprises of two parts.
The first step will focus on identifying neuronal representations of stimuli and actions, in view of the existing literature with e.g. rate modulation and spike synchrony [Churchland 2012, Shahidi 2019]. The idea is then to develop and compare detection methods for correlated spiking patterns, getting inspiration from existing work like Elephant’s SPADE (https://elephant.readthedocs.io/en/latest/reference/spade.html). Correlated activity patterns can in fact be formalized using several (and distinct) assumptions about the underlying spiking process, like stationarity, which is interesting to compare in terms of statistical power for detection and in terms of relevance for empirical spike trains.
The second step is to use dynamic network models that incorporate anatomical data (e.g. cortical layers) and fit them to the spiking data. This will formalize how a neuronal network implements the input-output (stimulus-response) function, aiming to uncover the neuronal computations occuring during the anmimal task. In particular, we will focus on the estimation of the effective connectivity between neurons, which describes how neurons drive each other in the network. This model-based approach is key to test the interplay between the neuronal mechanisms (e.g. non-linearity of the neuronal firing), the network effective connectivity interplay and the spike patterns.