ESR6 - Spiking Neural Networks for Information Representation and Decoding

Using information-theoretic computational analyses of real spike trains, based on the concept of intersection information, we will determine how information about tactile stimuli is carried by neuronal populations, from peripheral to central at various levels, and how the representation at a given stage of processing is read out to give rise to progressively determining the final percept and its use for the task.

TAGS   Biological Touch   Prosthetics   Technologies for Touch

Overview

Objectives

Using information-theoretic computational analyses of real spike trains, based on the concept of intersection information, we will determine how information about tactile stimuli is carried by neuronal populations, from peripheral to central at various levels, and how the representation at a given stage of processing is read out to give rise to progressively determining the final percept and its use for the task. The metric for the biological validity of this approach (and the devices based upon it) is to use the candidate decoding algorithm in order to specify both the object being contacted by the sensory system and the subjects’ psychophysical choice. Correct decoding of the stimulus indicates that the decoding algorithms have identified information-carrying algorithms, while correct decoding of choice indicates that the algorithms have identified the same elements used by the brain to construct perception. The fellow will therefore develop psychophysical behavioral paradigms for rats and humans in parallel, with methods for fully characterizing motor strategy and sensory input. S/he will record neuronal population data sets at multiple stages of rat tactile processing pathway.

Expected Results

Description of the spike-timing based neural population codes employed for tactile coding and perceptual decisions across the brain. Mathematical extrapolation of these principles to rules to encode information in artificial sensors and to use this information in robots for performing tasks.

Secondments

  • UNIBI

    developing SNN for hardware implementation

  • EPFL and SensArs

    decoding mechanisms for sensory feedback in prosthetic devices

Supervisors

  • S. Panzeri

  • M. Diamond

  • E. Chicca

  • S. Micera

  • F. Petrini

Alejandro Pequeño Zurro

While there is still many unknowns in the research challenge of understanding general intelligence, latest trends point towards the idea that the brain is not an isolated entity and the input and processes from the environment are constantly modifying your brain. During my previous studies in the area of robotics I have been researching sensory-adaptive systems inspired by animal experiments. It was a natural step for me to transition towards computational neuroscience focus on sensory systems and give some light in how sensory processing in the brain modifies our behaviour.

Neutouch for me

I found this research project very interesting since it combines a wide approach between artificial systems, cognition and neuroscience which perfectly matches with my multidisciplinary background.

Info

  • Research Topics

    Computational Neuroscience

  • Institution

    Istituto Italiano di Tecnologia — Università degli Studi di Genova

  • Background

    B.Sc. Telecommunication, Universidad Autonoma de Madrid (2010-2015)
    M.Sc. Robotics, University of Southern Denmark (2016-2018)

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No. 813713 ).
© Neutouch Project