Critical thresholds and dynamical network states in a neural architecture for cognitive tasks

Cognitive tasks are represented in a network, in which: - Nodes correspond to cell-assemblies with a critical threshold, which enables explanation of various psychological phenomena, like priming. - Dynamical network states (excitation loops) can propagate in specific, task-dependent ways ranging from singular chains to avalanches of excitation. - Temporary connections (binding) ensure the theoretically necessary, rapid, serial coupling of separate neural excitation patterns, corresponding at the cognitive level to binding of location, identity, rank-order, and task knowledge. Computer simulations of the network and empirical evidence for hypotheses on the critical threshold and serial binding show the viability of the model in the study of letter- and word recognition, change-blindness, and meta-cognition involving the tip-of-the-tongue state. The representation of several tasks within a single neural architecture produces constraints for the study and simulation of the architecture’s autonomous development. This is a necessary step in the search for biologically plausible mechanisms and conditions of cognitive self-organisation