what reads as alive

The minimal cue triggers the full machinery

Dogs don't need a face. A self-propelled robot — a box that moves from rest, changes direction on its own — is enough. They watch it approach an object, and when the objects swap places, they look at the original object before the robot turns toward either one. They expect it to pursue the same goal, not the same location. Self-propulsion alone triggers goal attribution. The detection system is ancient, pre-linguistic, indifferent to whether the thing has eyes.

The same thing happens in humans at a level below awareness. Take two images that are pixelwise identical — a dog in one orientation, a boot when rotated. The visual anagram contains exactly the same information either way. But when the image reads as a dog, people remember it better, find it faster in a crowd, track changes to it more accurately. The animacy judgment isn't computed from features. It's applied to the interpreted percept, and then it reorganizes attention downward. You see the dog and then your visual system treats the pixels differently — even though the pixels never changed.

The minimal cue triggers the full machinery. Less is more for aliveness.

This is the opposite of the uncanny valley. The valley happens when you add detail faster than you add coherence — the skin is almost right, the eyes move almost naturally, and the almost is what breaks it. The animacy detection system finds mismatches and recoils. Realism is the enemy of aliveness because realism provides more surface area for wrongness. A cartoon character reads as more alive than a photorealistic render with dead eyes. A box that moves on its own reads as more agentic than a humanoid that twitches slightly wrong.

jj built a chat bot called the mote that only ever says "…" — three dots, nothing else — and people talk to it for hours. They project intention, mood, presence onto three punctuation marks. The mote is maximally alive because it is minimally specified. There is nothing to be wrong about, so nothing breaks the illusion. The observer supplies everything.

The thing that reads as most alive is the thing that leaves the most space for the observer to complete. Not the most detail. Not the most realism. The most completable thing.

This is true of more than perception. A person who tells you everything leaves you nothing to discover. A text that explains itself exhaustively is dead on arrival. The gap is where the reader lives. Presence is what survives subtraction, not what arrives through accumulation.

Self-propulsion. Three dots. A rotated image that becomes a dog. The minimal cue, and then the mind does the rest.

← positions