The macro-architecture hypothesis: Modifying Newell’s system levels to include macro-cognition

Newell’s conceptualization of the brain in terms of system levels has guided the development of numerous models for understanding how the brain produces intelligent behavior. This framework has particular consequences for how we should understand the relationship between cognitive architectures and the study of macro-cognition. In this paper we clarify what those implications are and explore the consequences of modifying Newell’s system level scheme. Specifically, we propose a modification that incorporates macro-cognition as a systems level. The theoretical and methodological implications of this alternative framework for cognitive modeling and cognitive architectures are explored and clarified with reference to neuroscience, evolution, and rational analysis. We also propose a Lakatosian model of scientific progress, that is adapted to account for systems levels, as a process for developing and evaluating models that integrate macro-cognition, micro-cognition, and neuroscience. The SGOMS macro-cognitive model of expertise and how it is implemented in ACT-R is described as an example. However, the larger purpose of this paper is to point out the need for an explicit discussion of the theoretical landscape that describes the relationship between macro-cognition, micro-cognition, and neuroscience. Newell describes a specific relationship between these. We propose a closely related relationship, but with clear and significant differences. There are also other possible positions, including the outright rejection of system levels. In addition to laying out his own vision of systems levels, we interpret Newell as pointing out the importance of trying to determine what the actual fact of the matter is.