Towards an Active Inference Account of Deep Meditative Deconstruction

Abstract

This paper explores the intersection of deep meditative deconstruction, in particular the Buddhist defabrication process and its associated phenomenology, and computational mechanisms under the active inference framework (AIF). We contextualise states such as the jhānas within this process, drawing on both Buddhist theoretical frameworks and contemporary phenomenological understanding. We take a step towards ‘translating’ Buddhist meditation-based phenomenology into computational neurophenomenology. We demonstrate how Buddhist defabrication can be understood as a deconstructive process driving inference ever lower in an agent’s hierarchical generative model by the repeated release of mental tensing associated with clinging and aversion. We cast this release of mental tensing as corresponding to a hierarchical level-specific reduction in belief precision, permitting the interpretation of Buddhist notions such as equanimity and meditative stillness under AIF. We then illustrate the deconstruction process up to its conclusion in a cessation of phenomenal experience, touching on how states traversed during the process may inform areas related to core-knowledge structuring and the generation of experience.