Tononi1994mbc Sporns Neural Complexity
From enfascination
Contents |
A measure for brain complexity: relating functional segregation and integration in the nervous system
<bibtex> @article{tononi1994mbc,
title=Template:A measure for brain complexity: relating functional segregation and integration in the nervous system, author={Tononi, G. and Sporns, O. and Edelman, GM}, journal={Proceedings of the National Academy of Sciences}, volume={91}, number={11}, pages={5033--5037}, year={1994}, publisher={National Acad Sciences}
} </bibtex>
Tononi and Sporns, in trying to account for the opposing forces of 'integration' and 'segregation' develop an information theoretic metric for the latter and for the combination of the two, called Integration and neural complexity, resp. Neural complexity gives a high scores to systems exhibiting high levels of both integration and segregation. The metric is used on simulated neural data.
Integration is a generalization of mutual information to many parts (instead of between two). It is the sum of the differences between the entropies of the parts of a system and the entropy of the whole considered all at once. Neural complexity calculates this value for each scale of the system, taking the difference between the integration at some scale and the system-wide integration, and gets the sum across scales. The result is a single number, characterizing its complexity. I don't have my head around all the information theory yet.
Quotes
"It remains to be seen how neural complexity is altered by a succession of diverse signals from the world to yield adaptive behavior"
- I like how Olaf stays very mindful of the whole environment/embodiment thing, it keeps him very responsible wrt the grandness of claims he is making.
Questions
- Has this metric been used yet on real data?
- Has it been used (in neuroscience or elsewhere) by others?
- What are bipartitions?
Citations to read
- particularly stuff by Edelman who, i think, was Olaf's advisor.He has something from the late 80s titled "neuronal group selection", ah here: <bibtex>@book{edelman1987ndt,
title=Template:Neural Darwinism: The theory of neuronal group selection, author={Edelman, G.M.}, year={1987}, publisher={Basic Books New York}
}</bibtex>