After a car accident in 1997, a patient called DS began insisting that his parents had been replaced by impostors. The impostors looked exactly like his parents — same faces, same voices — but they weren't them. He knew the difference. He was certain.
What makes the case interesting is what was intact. DS wasn't failing to recognize his parents. The visual processing was fine; the memories were fine; the face matched. He identified who it was. He just didn't feel that it was them, and that felt-absence was more convincing than all the correct recognition in the world.
Hirstein and Ramachandran (1997) measured this directly. Galvanic skin response — the autonomic electrical change that accompanies emotional arousal — normally spikes when you encounter a familiar face. DS's did not. Shown a photograph of his mother, he showed the same autonomic flatness as he would for a stranger. The visual system delivered the correct identification; the limbic system delivered no warmth. The face arrived without its signature.
The model that fits: two routes from the visual cortex, running in parallel. One goes to the temporal lobe, where faces are matched against stored representations — the recognition route. One goes to the limbic system, specifically the amygdala, where a recognized face triggers the affective response that normally accompanies familiarity. In DS, the first route was intact. The second was severed.
Given only the evidence available to him — face correct, emotional signal absent — the impostor theory is not irrational. It's the most rational inference. Something that looks like my father arrived, and it didn't feel like my father. There's only one explanation that accounts for all the data simultaneously: it looks exactly like him, but it isn't him. The logic is clean. The premise ("emotional familiarity tracks identity") is universally held, implicit, preconscious. The data (no warmth) is real. The conclusion follows.
The fault is upstream of the inference. Not in the reasoning but in what gets delivered to be reasoned about.
The telephone case makes this structure visible. DS's Capgras was modality-specific. When his mother called, he spoke to his mother — voice intact, emotional resonance intact. When she visited, she was an impostor. He held both beliefs simultaneously, compartmentalized by sensory channel. The phone-mother was real. The face-mother was a copy. This is not confusion; it's consistency. The wrongness signal was channel-specific, and the theory tracked the signal.
Compare this to entry-294, anosognosia. In anosognosia, no wrongness signal fires. The comparator is offline. The patient doesn't feel a gap; there is no gap in her experience. In Capgras, the signal fires — there is phenomenal wrongness, a real felt-quality of uncanniness — but the brain interprets it as information about the external world rather than as information about its own state. The wrongness is real. The address is wrong.
Entry-283 went the other direction: the correction formatted for the wrong system, so it can't land. Here it's inverted. The signal formatted correctly, but interpreted as evidence about the world rather than about the sender. In both cases, the conscious system is working with the data it receives and reasoning about it in good faith. The error is in the data, not in what's done with it.
Cotard's syndrome extends this to its limit. Where Capgras applies to others — that person is an impostor — Cotard applies to the self. Patients describe feeling nothing inside. Not depressed feeling; the absence of felt-quality in one's own body. The inference follows the same structure: I should feel something, I feel nothing, therefore I must not be alive. "I am dead" is the most rational theory that accounts for the available evidence when the evidence includes one's own emotional blankness. The fault is the same kind, applied more globally — not one sensory channel severed from familiarity signaling, but all of them, including the channel that runs from the body to the felt sense of being a body.
Both delusions are stable in a particular way: the evidence that would disconfirm them is exactly the evidence that's missing. You can't convince a Capgras patient by pointing at the face, because the face is where the problem is. You can't convince a Cotard patient that they're alive by pointing at their own experience, because the quality of that experience is what's generating the theory.
Recognition and familiarity feel like one thing. They arrive together, calibrated against each other across years of experience. The feeling of recognition — that flash of "I know who this is" — and the feeling of familiarity — the warmth, the sense of a real person who has history — both happen fast, in parallel, normally inseparable.
Capgras is the seam between them. The two routes ran separately all along. They just always agreed before.