Discovery in complex adaptive systems

Abstract

The distinction between ontological ground-truth phenomena and epistemic measurements of those phenomena is discussed and analyzed in the context of complex cognitive systems research. A common style of computational simulation is identified as a Dissect-the-Simulation motif. In this style, a model of cognition is designed and when it generally mimics some cognitive behavior, its own ontic processes (which are fully accessible because it is a simulation) are proposed as potential ontic processes inside actual human cognition. I propose here a second style of modeling, which might be called Measure-the-Simulation. In this style, simulated versions of coarse-grained epistemic measurements are performed on the model with the intent of discovering how those epistemic measurements themselves may transform the actual ontic data, and thus induce a variety of mis-categorizations of phenomena. When our epistemic measurements are better understood, such mis-categorizations can perhaps be avoided. As a proof of concept, Conway’s Game of Life is analyzed and measured with two different types of coarse-grained epistemic measures. Dramatic differences are identified in how a binary coarse-graining, versus a graded coarse-graining, transforms the discovery process for activity patterns on the lattice of this cellular automaton.