Rodrigo Garro-Rivero

Core knowledge theory holds that human cognition originates from a small set of innate modular systems, each of which is informationally encapsulated from the others. We subject this theory to scrutiny by critically examining one influential division within it: the distinction between an encapsulated core knowledge system for place (representing where things are) and an encapsulated core knowledge system for forms (representing object structure). Reviewing behavioral, developmental, and comparative evidence long taken to establish a double dissociation between these systems, we argue that the case for either system’s encapsulation is considerably weaker, and the counterevidence more extensive, than typically acknowledged. In place of their strict encapsulation, we propose that place and form systems are better distinguished not by the information they have access to but by the proprietary algorithms that govern how they prioritize and process spatial information. In turn, these results suggest that core knowledge does not constitute a single, architecturally uniform, psychological kind, but a heterogeneous family of systems that vary in the nature and degree of their modularity.

Cognitive scientists and philosophers regularly appeal to representational formats to explain patterns of failure and success in psychological tasks. Often, this involves them analogizing mental representations to familiar forms of artefactual representations, such as sentences, pictures, graphs, maps, etc. In this paper, I suggest that this general strategy sometimes leads researchers to mistakenly attribute irrelevant properties of artefactual representations to our mental representations. I illustrate this point using the format of visuospatial representations as a case study. Recent evidence suggests that humans encode visuospatial relations in a polar coordinate format, comprising an independent pairing of angle and distance from an anchor point. I argue, however, that these mental coordinates differ markedly from familiar mathematical coordinates by comprising analog rather than digital constituents. The result is a distinctive hybrid representational format that problematizes familiar ways of categorizing mental representations and illustrates how mental representations often contain overlooked combinations of properties.