We develop neuronal criticality as a framework for the computational neurophenomenology of advanced meditation. Epistemologically, neuronal criticality constrains the possible mapping functions between phenomenological and mathematical structures through empirical (and theoretical) results on variations of specific informational and dynamical properties with the distance to criticality. After introducing theoretical and neurophysiological connections between neuronal criticality and active inference, previous theoretical models and empirical findings on advanced meditation are contextualized in terms of the distance to criticality. Attentional control dynamics are proposed to endogenously regulate the distance to criticality in a task-relative manner. Several phenomenological characteristics of minimal phenomenal experience are shown to naturally emerge in the vicinity of criticality from the integrated framework. This includes, amongst others, increased epistemic depth; the phenomenology of always already; and the attenuation of the first-person perspective in terms of a particular computational feedback-loop (i.e., epistemic ignition). Indeed, mathematically, more models of the modelling process are situated within the computational regime near criticality. Furthermore, leveraging tools developed in the field of information geometry, we show how, in the meditative endpoint of deconstruction called cessation of consciousness, the intrinsic Riemannian metric of the statistical manifold ‘collapses’ (i.e., becomes degenerate). The (in the limit) implied impossibility of belief-updating on that level is consistent with recent empirical results suggestive of critical dynamics during episodes of cessation. Neuronal criticality offers a promising framework for computational neurophenomenology and the computational neurophenomenology of advanced meditation, and enables new families of empirical and theoretical constraints for further theorizing.