Another Opening in the Explanatory Gap

Abstract

Taylor’s paper purports to bridge the explanatory gap between neurophysiological function and phenomenal experience (PE). He suggests properties of PE, attempts to show how these properties map onto underlying neural information processing, and argues that a particular form of neural network, the locally recurrent network, has the requisite properties for producing the most elemental forms of PE, or qualia. Part of the task of his paper is to explicate a theory of consciousness in a manner that is accessible to neuroscience, philosophy, and psychology. To Taylor’s credit, he succeeds in making the basic idea behind neural networks comprehensible. Yet, even with the technical sophistication and thought that obviously went into his paper, aspects of the model are unclear. The paper is very complex, as any theory of the explanatory gap should be. The gap has been around for a long time and will not disappear easily. Unfortunately, Taylor’s theory tends to be somewhat elusive as the paper progresses. It begins with a discussion of qualia (e.g., elemental experiential properties such as ‘‘redness’’), and then goes on to build on subjective aspects of PE. This implies that the theory will build sets of qualia into PE, but never is explicit about what the specific qualia are. Instead, a list of allegedly fundamental properties of PE, which appear to be quite separable from qualia are listed with some seemingly arbitrary criteria for neural underpinnings. This sort of conceptual slippage in the presence of the intricate and technical details of the neural networks made the paper frustratingly difficult to follow. Locally recurrent networks might provide an excellent model of the formation and combination of elemental representations, but it is not clear how this would necessarily result in PE itself. Taylor’s implicit argument appears to be thus: