[Comp-neuro] Can local inhibition lead to long-range competition?
Dylan Richard Muir
dylan.muir at unibas.ch
Mon Nov 10 10:58:55 CET 2014
We would like to draw your attention to a recent publication, where we examine the spatial range of competition expected for columnar visual cortex, given known features of cortical anatomy. Reprints are freely available from the author’s homepage (see link below).
Models of cortex that assume long-range competition are widespread, although it has long been known through anatomical studies that the spatial extent of inhibition is in fact more restricted than that of excitation.
We analyzed straightforward models of interacting cortical columns, and performed simulations of larger models, to show how the spatial scales of excitation and inhibition interact to produce competition through disynaptic inhibition.
We found that although disynaptic inhibition is widespread, the presence of competition across the cortical surface is predicted well by the anatomy of direct excitatory and inhibitory coupling. In comparison, the effect of multisynaptic network interactions was very small.
This implies that for realistic models of cortex, which should include short-range inhibition and longer-range excitation, the spatial extent of competition is likely to be even narrower than the range of inhibitory connections.
Muir and Cook, "Anatomical Constraints on Lateral Competition in Columnar Cortical Architectures", Neural Computation 26 8:1624–1666.
Dylan Muir | Post-doctoral researcher
Biozentrum, University of Basel | Klingelbergstrasse 50/70 | CH-4056 Basel
Phone: +41 61 267 17 75 | Email: dylan.muir at unibas.ch
www.biozentrum.unibas.ch | www.dylan-muir.com
More information about the Comp-neuro