Adaptive Networks at the Crossroad of AI and Formal, Biological, Medical and Social Sciences

In this paper, it is illustrated how a network-oriented modeling approach based on temporal-causal networks can be used to model adaptive processes in different domains and thus serve as a unifying factor for multiple sciences. The approach makes use of mathematical relations and functions as declarative building blocks to be used in a standard temporal-causal network format. A dedicated software environment is available which includes a combination function library for aggregation of multiple impacts within a temporal-causal network, with more than 35 already predefined basic combination functions. This software environment makes design and simulation for such network models relatively easy. The approach is illustrated by three examples of adaptive network models: a first-order adaptive network model for bonding by homophily, a second-order adaptive network model for plasticity and metaplasticity for emotion regulation dysfunction in disorders, and a fourth-order adaptive network model for evolutionary processes related to pathogens and pregnancy