World-enactment is the concept, central to enactive philosophy, that the world an organism perceives is not the objective world described by physics but a meaningful world constituted by the meeting of the organism's bodily capacities with environmental features. Jakob von Uexküll's concept of the Umwelt, Heidegger's analysis of being-in-the-world, and J.J. Gibson's theory of affordances all contributed to the view that worlds are species-relative and enacted, not simply given. A tick's world consists of butyric acid, warmth, and hair. A human's world consists of affordances — chairs for sitting, doorways for passing, coffee cups for grasping — constituted by the relationship between bodily capacities and environmental features.
The concept of the Umwelt was developed by Estonian-German biologist Jakob von Uexküll in the early 20th century, most famously in his 1934 essay A Foray into the Worlds of Animals and Humans. Uexküll argued that each species inhabits a perceptual world uniquely shaped by its sensory and motor capacities, and that no single objective world is shared by all organisms. The tick's Umwelt consists of three perceptual dimensions; the bat's includes echolocation; the human's is structured by bipedal locomotion, manipulative hands, and linguistic culture.
Heidegger extended this with his analysis of being-in-the-world in Being and Time (1927), arguing that humans do not encounter neutral objects that are subsequently invested with meaning but always already inhabit a meaningful world of tools, projects, and concerns. J.J. Gibson's ecological psychology developed the theory of affordances — the idea that environmental features are perceived directly in terms of the actions they afford to a particular organism. A chair affords sitting for a human; it does not afford sitting for a fish.
For Noë, these threads converge on a single claim with profound implications for AI. A world is not a database. It is an enacted meaningful structure that emerges in the coupling of a living organism with its environment. The 'world' available to a large language model is the archive of linguistic traces left by organisms that enacted worlds. The model processes these traces with extraordinary sophistication. It does not enact a world. And the difference is not a deficiency that more data could fix; it is an ontological gap between the trace and the living.
David Z. Morris captured this asymmetry with a now-famous line drawn from Noë's framework: 'A robot can't make coffee because a robot can't taste coffee.' The observation is not about the engineering of coffee-making machines, which exist. It is about what the activity 'making coffee' consists in when performed by an organism for whom coffee matters — an organism that adjusts the grind because the last cup was bitter, that tastes as it goes, that has decades of mornings structured by the ritual. The robot executes. The person enacts.
Jakob von Uexküll, Umwelt und Innenwelt der Tiere (1909), and A Foray into the Worlds of Animals and Humans (1934). Further developed by Heidegger (Being and Time, 1927), J.J. Gibson (The Ecological Approach to Visual Perception, 1979), and formalized within enactivism by Varela, Thompson, Rosch, and Noë.
Worlds are enacted, not given. Perception is the bringing-forth of a meaningful structure through the organism's species-specific capacities.
The Umwelt is species-relative. Each kind of organism inhabits a perceptual world uniquely shaped by its body.
Affordances are relational. Environmental features are perceived as possibilities for action, constituted by the meeting of body and world.
Data is not world. An archive of linguistic traces left by world-enacting organisms is not itself a world.
The robot cannot taste. AI processes patterns derived from embodied engagement without undergoing the engagement.
Representationalists argue that what enactivists call 'world-enactment' is just a dramatic rebranding of perceptual processing that could be fully described in representational-computational terms. Enactivists respond that the difference is not rhetorical but ontological — that a representation of a world is not a world, any more than a map is a territory.