Seleziona la tua lingua

Thesis Project Form

Title (tentative): Simple neuron based models for behavior and decision making

Thesis advisor(s): Massobrio Paolo, Paul Tiesinga E-mail:
Address: Via All'Opera Pia, 13 - 16145 Genova Phone: (+39) 010 33 52761
Description

Motivation and application domain
In order to make a decision you have to assign a value to the different
alternatives. This process has been studied in non human primates by
presenting options with a different probability of reward and a
different value of reward, while recording the activity of neurons in
multiple decision-related areas.

General objectives and main activities
Two types of models have been proposed,
one based on reinforcement learning (RL) to account for the behavioral
choice, others to account for the behavior in terms of neural activity
produced by recurrent neural networks. The goal for this project is to
link variables from RL models, such as value and choice probability, to
activity from neural models and, if possible, link the change to
plasticity processes.
In this project simple reinforcement learning models will either be
implemented in matlab/python, however to use existing code to optimize
the behavior of recurrent neural networks.

Training Objectives (technical/analytical tools, experimental methodologies)
The student will develop computational skills by
implementing and run computer models, optimize their performance using
optimization tools in for instance tensorflow. In addition, analytic and
statistical skills are necessary to analyze the neural activity in the
model and relate it to behavioral choices and variables such as value.
It is expected that some simple analytical calculations will be
necessary to help understand the simulation results.

Place(s) where the thesis work will be carried out: Radboud University

Additional information

Pre-requisite abilities/skills: Computational Neuroscience, Python and Matlab Programming

Curriculum: Neuroengineering

Maximum number of students: 1

Financial support/scholarship: Erasmus +