Implementing engrams from a machine learning perspective: the relevance of a latent space

TL;DR

This paper explores how autoencoders in neural networks relate to biological engrams, emphasizing the importance of latent space dimensionality and its implications for cognitive capacity and machine learning.

Contribution

It analyzes the role of latent space in autoencoders as models of engrams, linking neural structure differences to cognitive abilities and machine learning potential.

Findings

Latent space dimensionality correlates with information complexity.

Species differences in connectome relate to cognitive capacities.

Machine learning systems can surpass biological limitations.

Abstract

In our previous work, we proposed that engrams in the brain could be biologically implemented as autoencoders over recurrent neural networks. These autoencoders would comprise basic excitatory/inhibitory motifs, with credit assignment deriving from a simple homeostatic criterion. This brief note examines the relevance of the latent space in these autoencoders. We consider the relationship between the dimensionality of these autoencoders and the complexity of the information being encoded. We discuss how observed differences between species in their connectome could be linked to their cognitive capacities. Finally, we link this analysis with a basic but often overlooked fact: human cognition is likely limited by our own brain structure. However, this limitation does not apply to machine learning systems, and we should be aware of the need to learn how to exploit this augmented vision of the nature.