Universal principles justify the existence of concept cells

Abstract

The widespread consensus argues that the emergence of abstract concepts in the human brain, such as a “table”, requires complex, perfectly orchestrated interaction of myriads of neurons. However, this is not what converging experimental evidence suggests. Single neurons, the so-called concept cells (CCs), may be responsible for complex tasks performed by humans. This finding, with deep implications for neuroscience and theory of neural networks, has no solid theoretical grounds so far. Our recent advances in stochastic separability of highdimensional data have provided the basis to validate the existence of CCs. Here, starting from a few first principles, we layout biophysical foundations showing that CCs are not only possible but highly likely in brain structures such as the hippocampus. Three fundamental conditions, fulfilled by the human brain, ensure high cognitive functionality of single cells: a hierarchical feedforward organization of large laminar neuronal strata, a suprathreshold number of synaptic entries to principal neurons in the strata, and a magnitude of synaptic plasticity adequate for each neuronal stratum. We illustrate the approach on a simple example of acquiring “musical memory” and show how the concept of musical notes can emerge.

Dimensionality and ramping: Signatures of sentence integration in the dynamics of brains and deep language models

“…The present intrinsic dimensionality hypothesis stems from much research in the past decades, which has emphasized how biological neural networks use high-dimensional vector spaces to encode complex structures (Quiroga et al, 2005;Tyukin et al, 2019;Gorban et al, 2019;Calvo Tapia et al, 2020). For instance, human MEG signals can be decomposed into several dimensions that reflect the various contents of a visual sequence (Liu et al, 2019;Quentin et al, 2019;Al Roumi et al, (which was not certified by peer review) is the author/funder.…”

Toward Reflective Spiking Neural Networks Exploiting Memristive Devices

“… Calvo Tapia et al (2020b) extended results into the problem of building abstract concepts by binding individual items of the same kind. Figure 4B illustrates the model mimicking primary signaling pathways in the hippocampus.…”

“…At high dimensions, the neuron exhibits absolute selectivity to the input stimuli. Such neurons can be used to implement a variety of cognitive behaviors ( Tyukin et al, 2019 ; Calvo Tapia et al, 2020b ; see also Section “Novel Mathematical Principles for Spiking Neural Networks: Concept Cells and High-Dimensional Brain”). The switching dimension depends on the circuit components and starts at n = 50.…”

“…Learning the proposed HD neuron is automatic and goes the following way ( Calvo Tapia et al, 2020b ). At the beginning, the memristors have arbitrary initial resistances, which means that the neuron has some combination of the synaptic weights.…”

“…We note that the concept of a high-dimensional brain and analog machine vision complement each other and may bring this area to a qualitatively new level. Although we have described only the simplest selective effect emerging in HD neurons, more complex architecture (see, e.g., Calvo Tapia et al, 2020b ) are ready to be implemented in memristive architectures and SNNs.…”