Skip to content

Conscious access through a model of the world and body

Shogo Makioka

preprint DOI: 10.31234/osf.io/f2bh8 (opens in new tab)

Study at a glance

AI-extracted from the abstract
Characteristics Theoretical or philosophical paper
Key points Proposes that a "world/body model," generated by projecting modality-specific latent spaces learned through adversarial learning onto a single internal model with an innate spatiotemporal frame, mediates conscious access. Argues this accounts for rich sensory access, cross-modal exchange, and the unity of consciousness, and hypothesizes that qualia arise from information without explicit cross-modal correspondence.

Abstract

A novel framework in which a world/body model mediates conscious access is proposed. The world/body model is assumed to be generated by projecting the latent spaces acquired through adversarial learning in each modality onto a single internal model with an innately constrained temporal and spatial frame of reference. This framework has the following features. (1) Since the latent space holds detailed information about the external world and one's body, it is possible to explain the process of conscious access to rich sensory information. (2) Specifically examining the process by which information is exchanged between modalities is possible by postulating the world/body model as an intermediate representation. Based on the findings on synesthesia, the spatial frame of reference in the world/body model is considered to be based on a Cartesian coordinate system. (3) The phenomenal nature of consciousness can be discussed from a functional perspective by examining the process by which detailed sensory information is represented and accessed via the world/body model. The unity of consciousness is naturally explained by assuming that conscious content corresponds to a single-world/body model. A hypothesis is presented that qualia originates in information that does not have an explicit correspondence between modalities.

Citations in the library