Attractor Dynamics and Thermodynamic Analogies in the Cerebral Cortex: Synchronous Oscillation, the Background EEG, and the Regulation of Attention

作者：Wright J J^*

来源：Bulletin of Mathematical Biology, 2011, 73(2): 436-457.

DOI：10.1007/s11538-010-9562-z

• Attention
• Cognition
• Cortical activation
• Cerebral cortex
• Gamma synchrony
• Synchronous oscillation
• EEG
• LFP
• Continuum neural field
• Attractor neural networks
• Thermodynamic analogies
• Null-spikes
• Phase-cones

Ongoing changes in attention and cognition depend upon cortical/subcortical interactions, which select sequences of different spatial patterns of activation in the cortex.
It is proposed that each pattern of cortical activation permits evolution of electrocortical wave activity toward statistically stationary states, analogous to thermodynamic equilibrium. In each steady-state, neurons fire with an intrinsic Poisson spike probability and also with a bursting pattern related to network oscillations. Excitatory cell dendrites act as a regenerative reservoir in which pulse generation is balanced against dissipations.
Equilibria exhibit contrasting limits. One limit, at high cortical activation, generates widespread zero-lag synchrony among excitatory cells, with partial suppression of noise. Excitatory and inhibitory cells approach zero-lag local correlation, with 1/4 cycle lag-correlation at greater distances of separation. The high-activation limit defines a correlated system of attractor basins, capable of co-ordinating synaptic modifications and intracortical signal generation. Suppression of noise would enhance convergence about attractor basins in the manner of simulated annealing, while, conversely, the persistence of some noise prevents network paralysis by phase locking. At the opposite limit-that of low activation-spikes and waves have low cross- and auto-correlation, but have wide-spectrum sensitivity to inputs. It is hypothesised that cortical regions, transiently at equilibrium near these extremes, engage in interaction with each other and with subcortical systems, to generate ongoing sequences of attention and cognition.
This account is compatible with classical and recently observed experimental phenomena. The principle features inferred from a simplified linear mathematical account are reproduced in a more physiologically realistic and non-linear numerical simulation.

• 出版日期2011-2