Information processing and intracellular 'neural' (protein) networks: considerations regarding the diffusion-based...

The possibility that the disposition of subsets of proteins within the cell can retain memory traces and may act therefore in a computationial role has been advanced and more recently refined by Bray (Nature (1995) 376, 307-312). The proposition is not without its merits but inevitably has a number of associated difficulties, some of which are discussed in this article. These relate to the nature of the computational units envisaged (analog vs digital), their limitations in the number of stable patterns they can accommodate, the reliance on diffusion at the molecular level as the 'governing principle', and the complication of the turnover of proteins through degradative mechanisms. These issues suggest that certain modifications of the original model are required.