Adaptive Network Modeling for Criterial Causation

Abstract

Propagation of activation of neurons depends on settings of a number of intrinsic characteristics of the network of neurons, such as synaptic connection strengths and excitability thresholds for neurons. These settings serve as criteria on the incoming signals for a neuron to get activated. As part of the plasticity of the neural processing these network characteristics also change over time. Such changes can be slow compared to propagation of activation, like in learning from a number of experiences, but they can also be fast, like in memory formation. From the informational perspective on the criteria, this can be considered a form of information formation, and the firing of neurons as driven by this information. This is called criterial causation. In this paper, an adaptive network model is presented modeling such criterial causation. Moreover, it is shown how criterial causation in the brain relates to the more general temporal factorisation principle for the world’s dynamics.