Why Neurons Are Not the Right Level of Abstraction for Implementing Cognition

Claude Touzet (1)

1 LNIA - Laboratoire de Neurosciences intégratives et adaptatives

The cortex accounts for 70% of the brain volume. The human cortex is made of micro-columns, arrangements of 110 cortical neurons (Mountcastle), grouped in by the thousand in so-called macro-colums (or columns) which belong to the same functional unit as exemplified by Nobel laureates Hubel and Wiesel with the orientation columns of the primary visual cortex. The cortical column activity does not exhibit the limitations of single neurons: activation can be sustained for very long periods (sec.) instead of been transient and subject to fatigue. Therefore, the cortical column has been proposed as the building block of cognition by several researchers, but to not effect – since explanations about how the cognition works at the column level were missing. Thanks to the Theory of neuronal Cognition, it is no more the case. The cortex functionality is cut into small areas: the cortical maps. Today, about 80 cortical maps are known in the primary and secondary cortex [1]. These maps form a hierarchical organization. A cortical map is a functional structure encompassing several thousands of cortical columns. The function of such maps (also known as Kohonen maps) is to build topographic (i.e., organized and localized) representations of the input stimulii (events). This organization is such that similar inputs activate either the same cortical column or neighboring columns. Also, the more frequent the stimulus, the greater the number of cortical columns involved. Each map acts as a novelty detector and a filter. Events are reported as patterns of activations on various maps, each map specialized in a specific " dimension ". Spatial and temporal coordinates of events are linked to activations within the hippo-campus and define de facto the episodic memory. Learning is achieved at neuronal level using the famous Hebb's law: " Neurons active in the same time frame window reinforce their connections ". This rule does not respect " causality ". This, plus the fact that there is at least as much feedback connections as there are feed-forward ones, explain why a high level cortical activation generates a low level cortical pattern of activations – the same one that would trigger this high level activity. Therefore, our opinion is that the true building block of the cognition is a set of feed-forward and feedback connections between at least two maps, each map a novelty detector.

Domaines: Neurosciences

Claude Touzet. Why Neurons Are Not the Right Level of Abstraction for Implementing Cognition. Biologically Inspired Cognitive Architectures 2012: Proceedings of the Third Annual meeting of the BICA Society, 196, Springer, pp.317-318, 2013, Advances in Intelligent Systems and Computing, ⟨10.1007/978-3-642-34274-5_54⟩. ⟨hal-01338036⟩