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.
Planned Secondments
-
to develop SNN for hardware implementation
-
decoding mechanisms for sensory feedback in prosthetic devices
Enrolments (in Doctoral degree/s)
Scuola Internazionale Superiore di Studi Avanzati
Supervisors
S. Panzeri, M. Diamond, E. Chicca, S. Micera, F. Petrini
Tags
COMP
PRO
ROB
TECH