From Generative Models to Generative Passages: A Computational Approach to (Neuro) Phenomenology

Abstract

This paper presents a version of neurophenomenology based on generative modelling techniques developed in computational neuroscience and biology. Our approach can be described as computational phenomenology because it applies methods originally developed in computational modelling to provide a formal model of the descriptions of lived experience in the phenomenological tradition of philosophy (e.g., the work of Edmund Husserl, Maurice Merleau-Ponty, etc.). The first section presents a brief review of the overall project to naturalize phenomenology. The second section presents and evaluates philosophical objections to that project and situates our version of computational phenomenology with respect to these projects. The third section reviews the generative modelling framework. The final section presents our approach in detail. We conclude by discussing how our approach differs from previous attempts to use generative modelling to help understand consciousness. In summary, we describe a version of computational phenomenology which uses generative modelling to construct a computational model of the inferential or interpretive processes that best explain this or that kind of lived experience.

Fig. 1

A generative model of phenomenological experience. According to Husserl, noema are constituted through a kind of interpretation process, where the ‘hyletic’ data of pure lived experience are ‘animated’ by a noetic intention. Computational phenomenology casts this process of disclosure as a kind of ‘inference’ process based on a generative model. Right: Basic components of generative modelling that are used in computational phenomenology, as they are typically deployed to model task behavior and neural processes (see in-text description). A generative model can be decomposed into priors and a likelihood, which together form (one decomposition of) a joint probability distribution over all states of a system. Top left: A generic generative model that is capable of perceptual inference. Bottom left: A simple generative model for computational phenomenology. Here, we have specified one prior, denoted 1, which we could associate with phenomenological knowledge (e.g., claims of Husserlian, Merleau-Pontian, Heideggiarian, Bergsonian phenomenology about conscious lived experience and the structure of the lifeworld, i.e., what in the world that I inhabit typically causes this or that sensory data). The likelihood, which is denoted 3, maps hyletic data onto that knowledge, in a conditional fashion (i.e., it specifies the kind of hyletic data that I would sense if this or that latent state was the cause of my experience). The hyletic data, then, corresponds in a straightforward way to the data or observable states in a generative model, denoted 4. The noema or the phenomenological hypothesis that is mobilized to make sense of the hyletic data is labelled 2. Through the dynamic flow of lived phenomenological experience, we form a belief about the cause of our lived experience in the world that we inhabit phenomenologically (which is, lest we forget, the world made up of noema; what Husserl called the lifeworld). In an objectivist (e.g., Kantian) metaphysics, the element denoted 4 is the ‘phenomenon’, generated by its true cause or ‘noumenon’, denoted 5