Embodied Computation

Embodied computation is the thesis that computation is not an abstract process independent of its physical substrate, but is fundamentally shaped by the body and environment in which it occurs. It rejects the view — dominant in classical computer science and much of artificial intelligence — that cognition and computation can be understood as symbol manipulation in a disembodied, Turing-equivalent system.

The Disembodiment Assumption and Its Critics

The classical view treats computation as a formal relationship between input symbols and output symbols, realizable in any sufficiently organized physical medium. The brain is seen as hardware; the mind is software. Embodied computation challenges this by arguing that the particular physical and geometric properties of the computing system — its sensors, effectors, material composition, and environmental embedding — are not merely implementation details but constitutive features of the computation itself.

This position has roots in cybernetics and has been developed in embodied cognition, enactivism, and new AI. The core argument is that many computations are only possible, or are only efficient, because they are performed by systems with specific bodily structures interacting with specific environments.

Relationship to Other Frameworks

Embodied computation overlaps with morphological computation, which focuses on how body shape and material properties compute; with emergent computation, which emphasizes distributed, self-organizing computational processes; and with neural computation, which studies how biological and artificial neural substrates perform computation. The unifying thread is that computation is a property of organized physical systems, not merely of formal symbol systems.