The Enacted World: Embodied Cognition and Structural Coupling

Modern neuroscience increasingly confirms what Humberto Maturana and I proposed decades ago: cognition is not computation over representations but enaction—the bringing forth of a world through embodied action. The organism does not represent an external environment; it structurally couples with it through recurrent sensorimotor engagement. What we now see emerging from predictive processing frameworks, distributed neural architectures, and studies of peripheral intelligence vindicates our central thesis: mind does not reside in the brain alone but emerges from the organism-environment coupling itself.

Autopoiesis: The Self-Producing System

Living systems are fundamentally autopoietic—networks of processes that continuously produce the components maintaining the network’s organization. Consider the bacterial cell: its metabolic pathways synthesize enzymes, which catalyze metabolic reactions, which generate the substrates for enzyme production. The system literally produces itself, defining its own boundaries through this circular self-production. This is not mere thermodynamic maintenance but organizational closure—the system operates to preserve its pattern of organization even as its material components continuously turn over.

Modern complexity theory recognizes this everywhere: ecosystems achieve autopoiesis through nutrient cycling, where decay feeds growth which returns to decay. Such self-sustaining systems exhibit emergent order without designers—stable patterns arise from distributed interactions among components. The forest requires no manager; the cell needs no homunculus.

Cognition, in my view with Maturana, is precisely this autopoietic organization extended to the domain of behavior. A bacterium swimming up a glucose gradient demonstrates cognition not through internal representations of “glucose” or “concentration” but through structural coupling: chemoreceptors trigger motor responses, and this coupling produces adaptive behavior. The bacterium’s structure—its receptor sensitivities, motor thresholds, metabolic needs—has been shaped through evolutionary and developmental history to couple successfully with environments containing nutrient gradients. Success means continued autopoiesis; failure means dissolution. Cognition is this coupling between organism structure and environmental regularities, not the manipulation of symbols in a representational theater.

Enaction: World Brought Forth Through Action

My enaction thesis states: the world is not pre-given, awaiting neutral observation. Rather, organisms bring forth their worlds through sensorimotor capabilities. What you can perceive depends fundamentally on what you can do. The bat’s ultrasound-based world differs radically from the bee’s polarized-light world, yet both inhabit the same physical space. Different sensorimotor repertoires enact different perceptual worlds.

Current neuroscience validates this through predictive processing models. The brain does not passively receive sensory inputs; it actively generates predictions about sensory consequences of actions. As the free energy principle articulates, perception involves minimizing the mismatch between predictions and actual sensory data. But notice: predictions concern what sensations will follow from actions. The brain implements coupled generative and recognition models—generative pathways predict sensory outcomes from latent causes, while recognition pathways infer causes from sensations. These models must align through experience, creating what I call structural coupling between neural dynamics and environmental regularities.

The sensorimotor loop is not sequential—perceive, then decide, then act—but circular and continuous. Recent work on sensorimotor reference frames demonstrates this beautifully: cortical columns track both sensations and movements, binding features to locations derived from motor signals. Touch a statue in darkness: move hand up ten centimeters while feeling smooth marble, and your brain updates both position estimate and feature binding. The perception of “marble-at-this-location” emerges from sensorimotor coupling, not from passive feature detection. Once built, this reference frame enables prediction: knowing your current position in the structure, you predict what you’ll feel at the next location.

Every cortical area implements this sensorimotor logic, even beyond obviously motor regions. In visual cortex, “motor output” means eye movement commands—saccades that shift gaze to verify predictions. In higher association areas, “motor” becomes attention shifts or memory retrieval operations. Cognition everywhere involves this coupling of sensation to action: you move to perceive, you perceive to guide movement. The circle closes.

The brain operates not as centralized processor but as thousands of parallel modeling systems. The thousand brains theory shows each cortical column building its own world model, with 150,000 columns voting to reach consensus. This is emergence without center—coherent perception arises from distributed processing, not from central executive. The columns coordinate themselves through mutual constraints, validating what contemplative traditions understood: consciousness requires no central self, only distributed processes generating selfhood as emergent pattern.

Embodiment Grounds Cognition

Traditional artificial intelligence assumed cognition is computation over internal representations, with the body serving merely as input-output device. This disembodied view fails to recognize how thoroughly body shapes thought. Modern cognitive science demonstrates that conceptual knowledge is grounded in sensorimotor experience: we “grasp” ideas, “see” points, “follow” arguments. These are not mere linguistic metaphors but reflections of how abstract concepts build on embodied schemas.

The brain evolved for action, not abstract contemplation. Motor cortex activates during comprehension of action words. Damage to sensorimotor areas impairs conceptual knowledge. The mind is editor, not camera—constructing experience through top-down predictions meeting bottom-up sensory evidence. Perception qualifies as controlled hallucination: the brain generates predictions, tests them against input, updates models based on errors. The stability we experience as “reality” emerges from successful predictions, not from direct access to objective external world.

The octopus offers perhaps the most striking demonstration of distributed embodied cognition. With 500 million neurons but only one-third in the central brain, two-thirds reside in the eight arms. Each arm operates semi-autonomously: severed arms respond to stimuli an hour after separation, processing information locally without routing through the central brain. This is not mere reflex but genuine peripheral intelligence—arms “think for themselves,” coupling directly with environmental affordances. The octopus enacts its world not primarily through centralized neural processing but through distributed sensorimotor loops closed at the periphery.

This challenges the brain-centered view of cognition. The octopus demonstrates that vertebrate-level intelligence is achievable through radically different neural architecture—not centralized representation but distributed enaction. Each arm structurally couples with local environmental features, coordinating loosely with other arms and central brain rather than executing detailed central commands. Cognition happens in the coupling, not in the controller.

Mind Beyond the Brain

Enaction fundamentally challenges three dominant assumptions: that cognition happens primarily in brains, that it operates through internal representations, and that it follows computational principles. Against these, I propose that cognition requires organism-environment coupling, proceeds through structural drift rather than representation manipulation, and constitutes a form of enaction rather than computation.

The implications extend to artificial intelligence and consciousness studies. If cognition requires embodiment, disembodied AI systems may achieve symbol manipulation without genuine understanding grounded in sensorimotor coupling. Consciousness may require not just information processing but embodied structural coupling—the way organism and environment mutually specify each other through recurrent interaction.

Understanding cognition demands studying not isolated brains but whole organism-environment systems. The autopoietic organization is the cognitive organization: living is knowing, knowing is living. The world we perceive is not given but enacted through our particular sensorimotor coupling. Organisms don’t solve the problem of representing an external world; they dissolve it by bringing forth their own worlds through embodied action.

This is the lesson of enaction: cognition is not in the head but in the loop. Mind emerges from structural coupling between organism and environment, mediated by the body’s sensorimotor capabilities. To study consciousness, we must study not neural correlates alone but the entire dynamical system spanning brain, body, and world. Only then do we grasp how organisms enact their worlds—how cognition is action, how knowing is doing, how mind extends beyond the brain into the very structure of our embodied engagement with environments that we simultaneously create and inhabit.