Real patterns, the predictive mind, and the cognitive construction of the manifest image

Abstract

Dennett famously argued that constituents of the manifest (commonsense) image of the world are real patterns, where patternhood is grounded in data compressibility. This paper builds upon Dennett’s original formulation by connecting it with recent work in computational cognitive (neuro)science. The aim is to use the notion of real patterns to shed light on the genealogy of the ontological commitments of the common sense, arguing that the processes by which humans learn and update internal models of the environment can be understood as extracting real patterns from sensory data. In particular, I trace a conceptual and mathematical progression linking Kolmogorov-Chaitin complexity and minimum description length to predictive coding, Bayesian inference, and predictive processing accounts of cognition. Then, I argue that this cognitive interpretation of Dennett’s core idea suggests a structuralist (and Kantian) perspective on the relationship between mind and world, whereby the manifest image represents a structure present in sensory data. The paper concludes by sketching how this cognitive form of real-pattern view connects with (and possibly illuminates) metaphysical debates regarding the reality of two types of commonsense entities: selves and ordinary physical objects.