Neural Selection in Corticothalamocortical Loops: Heterogeneity as a Computational Resource.

Neuroscience has long relied on (typically binary) categorisations to manage the complexity of brain organisation. Yet growing anatomical and physiological evidence reveals that these dichotomies artificially partition systems that are inherently continuous and heterogeneous. In the thalamus, neurons display mixed afferent-efferent motifs and graded physiological properties that defy rigid categorisation. In the cerebral cortex, laminar somatic labels obscure the contribution of multilayer active dendritic nonlinearities, such as burst spiking, which radically reshapes modes of neural communication. Here, we propose that heterogeneity is not taxonomic noise, rather it is a key functional resource. Specifically, we argue that corticothalamocortical loops implement a dynamical process algorithmically akin to natural selection. Cortical and thalamic heterogeneity provides variation across neural coalitions; inhibitory competition, particularly within the thalamic reticular nucleus, enacts selection among competing whole-brain signals; and neural amplification, paradigmatically via subcortical projecting bursting promotes inheritance by recruiting selected neural coalitions across time. This perspective emphasises the corticothalamocortical system as a dynamic substrate for neural selection, rather than a static circuit motif. By integrating anatomy, physiology, and dynamical systems theory, we outline a framework that leverages neural heterogeneity for flexible, adaptive behaviour.