Context: The enactivist tradition, out of which neurophenomenology arose, rejects various internalisms – including the representationalist and information-processing metaphors – but remains wedded to one further internalism: the claim that the structure of perceptual experience is directly, constitutively linked only to internal, brain-based dynamics. Problem: I aim to reject this internalism and defend an alternative analysis. Method: The paper presents a direct-realist, externalist, sensorimotor account of perceptual experience. It uses the concept of counterfactual meaningful action to defend this view against various objections. Results: This account of experience matches certain first-person features of experience better than an internalist account could. It is fully tractable as “normal science.” Implications: The neuroscientific conception of brain function should change from that of internal representation or modelling to that of enabling meaningful, embodied action in ways that constitutively involve the world. Neurophenomenology should aim to match the structure of first-person experience with the structure of meaningful agent-world interactions, not with that of brain dynamics. Constructivist content: The sensorimotor approach shows us what external objects are, such that we may enact them, and what experience is, such that it may present us with those enacted objects.
Open peer commentary on the article “Exploration of the Functional Properties of Interaction: Computer Models and Pointers for Theory” by Etienne B. Roesch, Matthew Spencer, Slawomir J. Nasuto, Thomas Tanay & J. Mark Bishop. Upshot: Artificial life computer simulations hold the potential for demonstrating the kinds of bottom-up, cooperative, self-organizing processes that underlie the self-construction of observer-actors. This is a worthwhile, if limited, attempt to use such simulations to address this set of core constructivist concerns. Although we concur with much of the philosophical perspective in the target article, we take issue with some of the implied positions related to dynamical systems, sensorimotor contingency theory, and neural information processing. Ideally, we would like to see computational approaches more directly address adaptive, constructive processes and mechanisms operant in minds and brains. This would entail using tasks that are more relevant to the psychology of human and animal learning than performing digit sums or sorts. It also could involve relating the dynamics of agents more explicitly to ensembles of communicating neural assemblies.
The binary divide between traditional cognitivist and enactivist paradigms is tied to their respective commitments to understanding cognition as based on knowing that as opposed to knowing how. Using O’Regan’s and Noë‘s landmark sensorimotor contingency theory of perceptual experience as a foil, I demonstrate how easy it is to fall into conservative thinking. Although their account is advertised as decidedly ‘skill-based’, on close inspection it shows itself to be riddled with suppositions threatening to reduce it to a rules-and-representations approach. To remain properly enactivist it must be purged of such commitments and indeed all commitment to mediating knowledge: it must embrace a more radical enactivism