Sequence learning in a single trial: a spiking neurons model based on hippocampal circuitry

Abstract

In contrast with our everyday experience using brain circuits, it can take a prohibitively long time to train a computational system to produce the correct sequence of outputs in the presence of a series of inputs. This suggests that something important is missing in the way in which models are trying to reproduce basic cognitive functions. In this work, we introduce a new neuronal network architecture that is able to learn, in a single trial, an arbitrary long sequence of any known objects. The key point of the model is the explicit use of mechanisms and circuitry observed in the hippocampus, which allow the model to reach a level of efficiency and accuracy that, to the best of our knowledge, is not possible with abstract network implementations. By directly following the natural system's layout and circuitry, this type of implementation has the additional advantage that the results can be more easily compared to experimental data, allowing a deeper and more direct understanding of the mechanisms underlying cognitive functions and dysfunctions, and opening the way to a new generation of learning architectures.

Linking Brain Structure, Activity, and Cognitive Function through Computation

“…Simulation is increasingly enabled by the computational capabilities and capacities becoming available in Fenix (see below) to handle the very large data representing a human brain, and is in fact driving the development of computer science through its requirements. In the past few years, models of the cerebral cortex (Markram et al, 2015)), hippocampus (Coppolino et al, 2021), cerebellum (Casali et al, 2020), basal ganglia (Grillner and Robertson, 2016), typically at the cellular/circuit level, large-scale brain-simulations based on point neurons (Potjans and Diesmann, 2014) or mean-field network modelling (Goldman et al, 2021), as well as models of cognitive functions, such as spatial navigation (Coppolino et al, 2021), object recognition, scene understanding, visuo-motor functions, attention, perception and learning have been developed, and are being constantly improved.…”

“…In contrast with our everyday experience using brain circuits, it can take a prohibitively long time to train a computational system to produce the correct sequence of outputs in the presence of a series of inputs. By directly following the natural system's layout and circuitry of the hippocampus, models allow a level of efficiency and accuracy to be reached that opens the way to a new generation of learning architectures, including one shot learning (Coppolino et al, 2021).…”