Representile

Representile is a proposed term in cognitive science and philosophy of mind for an elemental unit within a representational architecture whose primary function is to form, store, and manipulate internal representations of external states. A representile is not a passive observation but an active bearer of content, akin to a neural population, latent unit, or symbolic node that contributes to the system’s model of the world. The term emphasizes the representational status of the unit rather than its specific medium or mechanism.

In computational models, representiles can correspond to latent variables in probabilistic models or to hidden units in neural networks. They participate in representational inference, updating their content as new data arrive and the system minimizes prediction error or maximizes likelihood. In robotics or AI, distributing representation across multiple representiles allows scalable, modular accounting of complex perceptual categories.

Distinctions: Representiles differ from raw sensory inputs (percepts) in that they carry content that can be reinterpreted, transformed, and recombined; they are also distinguished from units that trigger actions without internal content. They can be dynamic, with content changing over time as the system revises its beliefs.

Usage and criticism: The representile concept is useful for clarifying modular structure in theories of mind and AI, but critics warn that it risks reifying representation as a thing rather than a process, and that the boundaries of what counts as a representile are often ill-defined.

See also: mental representation, representation, latent variable, predictive coding, cognitive architecture.