Wild Architecture: Explaining Cognition via Self-Sustaining Systems

Introduction

For some time now, one of the leading assumptions in the development of cognitive architectures has been Marr’s (1982) tri-level theory of explanation; the idea that the proper approach in developing a cognitive architecture (and thus explaining cognition) is to

• 1.

  Determine the computations necessary to completing a cognitive task (e.g., sorting a list of numbers from lowest to highest),

• 2.

  Generate a representation of the inputs, outputs, and algorithms an information-processing system would need to complete the task, and

• 3.

  Actually build (i.e., implement) a system capable of executing the algorithms.

The purpose of the present paper is to examine issues that have proven challenging to Marr’s implementation approach to explaining cognition. Three particular challenges are the issues of consciousness, embodiment, and representation. After examining these challenges, we will present an approach to describing cognitive architectures (Wild Systems Theory—WST, Jordan, 20213) that addresses each challenge, while simultaneously shifting the focus of modeling from looking to biology for inspiration, to looking at a more fundamental property that biological systems share with many other types of systems, including chemical, psychological, and cultural—specifically, the ability of certain far from equilibrium systems to generate catalysts that feedback into and sustain the processes that produced them; what Kauffman (1995) refers to as ‘autocatalytic’ systems, and what Jordan (2013) refers to as self-sustaining, or wild systems.