Thinking as Analogy-Making: Toward a Neural Process Account of General Intelligence

Abstract

What is the secret of human intelligence? A key discovery in psychology is that performance correlations across diverse cognitive tasks are explained by a few broad abilities and one overarching general factor, which is also predictive of real-life achievements. Whether these factors correspond to biological processes is a century-old debate. While previous research focused on localizing their correlates in brain structure, connectivity, and activation levels, the mechanisms of neural information processing related to intelligence are still unexplored. I outline a new approach integrating psychometrics with neuroscientific advances in identifying the computations underlying single tasks from their representational geometry to provide a novel perspective on this topic. In particular, I propose a neural process account of the general factor that builds on the central role of structure mapping—the process of abstracting and reasoning based on relational knowledge—in human cognition. Neural coding properties in the hippocampal and prefrontal–parietal systems that enable inferential leaps through structural abstraction might contribute to the general factor. In general, integrating neuro-representational and psychometric research has the potential to uncover core principles of natural intelligence.