The flow of information between micro and macro scales plays an important 

role in biological computation and cognition. Processing can be pictured 

in terms of cross-scale flow of of information (Conrad, 1984, 1995).

We have implemented a virtual system, called the hypernetwork model, 

that makes it possible to formally represent vertical information flow and to 

investigate the constraints to which it is subject (Segovia-Juarez and Conrad, 1999).

The hypernetwork model is a multi-scale architecture that stresses the 

flow of information among the different hierarchical levels. 

Influences from the environment filter down into lower levels (organs, cells,

 molecules), are dynamically integrated by processes at each of these levels, 

and then percolate up to higher levels through selective amplification. 

The levels are defined by scale features. The model, however, does not 

simply introduce levels ad hoc. Ultimately all interactions are between 

elementary particles. In the hypernetwork model the interactions are mediated 

by complementary pairing between string representations of macromolecules. 

Higher level units of organization are defined by scale features, at present 

by the density of molecular interactions. The hypernetwork architecture 

currently comprises three levels: organism, cellular, and molecular. 

The cells are networks of pairwise interactions among the molecular components. 

The organism is a network of interactions between effector and receptor molecules 

of cells. The cellular networks can be viewed as a highly abstracted representation 

of conformational or reaction cascades in the cell. The global network is molded 

for functional performance (simple classification tasks at the present stage) 

through error feedback acting on the structure and location of the molecular

components. The hypernetwork model has been developed to answer questions 

such as: what factors control how environmental signals are distributed through 

the network; how do influences filter down from macro to micro scales; 

what factors determine which microscale interactions are selected for amplification 

to the macro level; and how are interactions orchestrated within and between levels

to yield coherent function? Elucidating such issues will hopefully provide 

insights into the architectural constraints operative in models of information 

processing and cognition that draw on microscale features.

References

Conrad, M. (1984). Microscopic-macroscopic interface in biological 

information processing. Biosystems, 16:345-363.

Conrad, M. (1995). Cross-scale interactions in biomolecular information 

processing. Optical Memory and Neural Networks, 4(2):89-98.

Segovia-Juarez, J., and Conrad, M. (1999). Hypernetwork Model of Biological 

Information Processing, to appear in the Proceedings of the 1999 Congress 

of Evolutionary Computation (July 06-09, 1999, Washington D.C.).