Perceived space, or spatial practice, is the dimension of Lefebvre's triad produced not by design but by habituation. It is the space of the worn path through the grass where the sidewalk was not built, the elevator button never pressed because it goes to a floor nobody visits, the corner avoided because the body has learned it is dangerous. Perceived space is stubborn. It records, in the patterns of daily use, the accumulated knowledge of a community inhabiting a space — knowledge that no designer anticipated and that often contradicts the designer's intentions. In the AI interface, perceived space is what the Berkeley study documented: not the tool as designed, but the tool as actually practiced — seeping into every pause, filling protected temporal territories, producing patterns of use the designers did not intend and cannot fully recognize.
Perceived space is the dimension most closely tied to the body. It emerges from bodily practice — the muscular, sensory, kinesthetic engagement of an organism with an environment over time. The body carries knowledge that the mind cannot articulate: the weight of a door, the specific resistance of a particular chair, the micro-geographies of a familiar room. This bodily knowledge is perceived space in operation.
The gap between perceived and conceived space is where the most interesting spatial politics happens. Designers intend one thing; inhabitants do another. The restaurant's designated queue becomes the smoking area. The plaza designed for quiet contemplation becomes a skateboard park. The AI tool designed for productive partnership becomes, in practice, the companion for compulsive engagement. Each case is an instance of perceived space diverging from conceived space, and the divergence is not an error but the normal condition of spatial inhabitation.
For the AI interface, perceived space emerges through patterns that accumulate across millions of daily interactions: the specific times people prompt, the specific ways they phrase requests, the rituals they develop to begin and end sessions, the workarounds they construct when the tool frustrates them. These patterns are visible in aggregate data but their significance — what they mean for the lives being lived through them — is visible only from the inside, from the inhabitation.
The political question about perceived space is whether its knowledge informs the redesign of conceived space. At Pruitt-Igoe, it did not: residents developed detailed knowledge of the building's failures, but the feedback loop back to the architects was broken. In AI interfaces today, it partially does: usage data flows back to designers. But the data captures what the conceived space makes measurable, not what the lived space makes meaningful.
Lefebvre's attention to daily routine as analytically primary was developed across the three volumes of Critique of Everyday Life (1947, 1961, 1981). The refinement of spatial practice as a formal category came with The Production of Space.
The body is the medium. Perceived space is produced by bodies moving through environments over time, accumulating knowledge that has no propositional form.
Divergence is structural. Perceived space always diverges from conceived space, because bodies have practices the designer did not anticipate.
The feedback loop is political. Whether perceived space informs the redesign of conceived space is determined by institutional arrangements, not by the quality of the data.