23–24 Jun 2022
Virtual
Europe/Berlin timezone

Incremental Awake-NREM-REM Learning Cycles: Cognitive and Energetic Effects in a Multi-area Thalamo-Cortical Spiking Model

P1-1
23 Jun 2022, 14:10
3m
Virtual

Virtual

Poster & advertisement flash talk Main track Poster

Speaker

Mr Leonardo Tonielli (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy)

Description

INTRODUCTION/MOTIVATION
This work leverages Apical Isolation (AI) and Apical Drive (AD)[1][2] principles to induce in a model some of the favourable energetic and cognitive effects associated to NREM and REM sleep. Also, we follow the Apical Amplification (AA)[3] concept during awake learning. This way, we added REM to the brain states accessible to the thalamo-cortical spiking model[4][5] already capable of expressing realistic AWAKE and NREM brain dynamics.

METHODS
We developed a multi-area thalamo-cortical spiking model made of integrate-and-fire neurons: the thalamic layer provides perceptual input through contralateral feedforward connections to cortical areas, which gather and process such information by means of Spike-Timing-Dependent-Plasticity and Winner-Take-All Cell-Assemblies (CA) circuitry (Fig 1A).

RESULTS-AND-DISCUSSION
We found the optimal sleep stages duration is at 40s of NREM and 10s of REM (Fig. 1D), corresponding to a reduction of 22% in the network power consumption and an improvement of 1% in classification accuracy, in agreement with experimental data[8,11]; moreover sleep rhythms were found comparable with biological recordings[9] (Fig. 1B).
Cell-Assemblies trained over different examples of the same digit are grouped together in each area during NREM whereas those belonging to different cortical areas are associated through REM-stage (Fig. 1C).
Figure 1D demonstrates after-sleep cortico-cortical synapses homeostasis.

Acknowledgements

This work has been supported by the European Union Horizon 2020 Research and Innovation program under the FET Flagship Human Brain Project (grant agreement SGA3 n. 945539) and by the INFN APE Parallel/Distributed Computing laboratory. D. Cipollini contributed to this work during his master thesis research project at La Sapienza University of Rome.

Preferred form of presentation Poster & advertising flash talk
Topic area models and applications
Keywords Awake-NREM-REM cycles, Spiking Models, Classification task, Thalamo-cortical network, Apical Amplification-Drive-Isolation
Speaker time zone UTC+2
I agree to the copyright and license terms Yes
I agree to the declaration of honor Yes

Co-authors

Bruno Golosio (Department of Physics, University of Cagliari, Monserrato (CA), Italy and Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Cagliari, Monserrato (CA), Italy) Mr Cristiano Capone (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy) Mr Leonardo Tonielli (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy) Mr Cosimo Lupo (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy) Mr Davide Cipollini (5Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen, The Netherlands) Elena Pastorelli (Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Roma, Rome, Italy) Francesco Simula (Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Roma, Rome, Italy) GIANMARCO TIDDIA Mrs Giulia De Bonis (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy) Mrs Irene Bernava (2Istituto Nazionale di Fisica Nucleare (INFN) Sezione di Roma, Rome, Italy) Pier Stanislao Paolucci (Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Roma, Rome, Italy) chiara de luca (PhD Behavioural Neuroscience, La Sapienza University; INFN Sezione di Roma)

Presentation materials