Representations in human sciences: time for reconsideration

Abstract

The concept of representation has played a fundamental role in cognitive science since its inception in the 1950s. Representation gained this emphasis insofar as cognitive psychology and linguistics since their beginnings were based on a “computer metaphor” and, accordingly, on the classical Turing concept of computations as operations with symbols construed as representations of some data or content. With the emergence of numerous non-classical models
of computing in recent decades (distributed, analog, genetic, quantum, etc.), representation has been attacked by various anti-representationalist or “post-cognitive” schools. This article shows that the complete rejection of representations in theoretical schemes of radical connectionism and radical enactivism, as well as the irrelevance of this concept in the cognitive paradigm of “dynamic systems,” means a step back to traditional natural science with its weak ability to explain cognitive phenomena. Meanwhile, the strong representationalism closely related to computational symbolism is fraught with paradoxes, such as the “homunculus paradox,” and its interpretation on real neurophysiological facts is a theoretical problem. On the contrary, moderate, or weak, representationalism proceeds from the fact that complex computing systems use representations of computational outcomes in their subsystems or at their particular levels as input for computations in other subsystems or at other levels. Therefore, the concept of representation is of operational value to cognitive science, but it cannot be a universal clue to cognitive or philosophical problems.

Keywords:

computation, representation, cognitive science, symbolism, connectionism, enactivism, dynamic systems, predictive mind