Reality Construction in Cognitive Agents Through Processes of Info-computation

Gordana Dodig-Crnkovic,Rickard von Haugwitz

DOI: https://doi.org/10.1007/978-3-319-43784-2_10

2017-01-01

Abstract:What is reality for an agent? What is minimal cognition? How does the morphology of a cognitive agent affect cognition? These are still open questions among scientists and philosophers. In this chapter we propose the idea of info-computational nature as a framework for answering those questions. Within the info-computational framework, information is defined as a structure (for an agent), and computation as the dynamics of information (information processing). To an agent, nature therefore appears as an informational structure with computational dynamics. Both information and computation in this context have broader meaning than in everyday use, and both are necessarily grounded in physical implementation. Evolution of increasingly complex living agents is understood as a process of morphological (physical, embodied) computation driven by agents’ interactions with the environment. It is a process much more complex than random variation; instead the mechanisms of change are morphological computational processes of self-organisation (and re-organisation). Reality for an agent emerges as a result of interactions with the environment together with internal information processing. Following Maturana and Varela, we take cognition to be the process of living of an organism, and thus it appears on different levels of complexity, from cellular via organismic to social. The simpler the agent, the simpler its “reality” defined by the network of networks of info-computational processes, which constitute its cognition. The debated topic of consciousness takes its natural place in this framework, as a process of information integration that we suggest naturally evolved in organisms with a nervous system. Computing nature/pancomputationalism is sometimes confused with panpsychism or claimed to necessarily imply panpsychism, which we show is not the case. Even though we focus on natural systems in this chapter, the info-computational approach is general and can be used to model both biological and artifactual cognitive agents.