ConceptGraphSearchPerceptual

Perceptual search is what happens when you see several attributes in your visual field and you want to know what is this. The attributes might be facts like what has a collar and four legs, or they might be fragments of an image. These features form the seeds for a search for the closest matching nodes within the graph.

The perceptual search follows relationships in the reverse direction from the arrows in the graph, so it can search downwards in the inheritance hierarchy.

• Source: 2025-10-08 Killing AI’s Energy Hog

Picture yourself walking into a room, your eyes pick up edges, colors, shadows, textures, your ears register faint sounds of motion or conversation. All of this is raw input. It's useless until your brain begins the real work of perceptual search. Each of these features sparks activity in associated neurons. One detects straight lines, another responds to motion, another to a certain pitch of sound. But this activity doesn't stay isolated. It spreads like sparks racing outward across a web of information following relationships in the reverse direction. Candidate nodes begin to glow. Four legs, wagging tail, maybe a dog. Pointy ears and whiskers, maybe a cat. Dark wooden surface, maybe a table. Dozens of hypotheses compete in parallel. And recognition happens when one cluster of neurons reinforces itself strongly enough that the brain converges on, this is phyto.

• Source: 2025-10-02 How you search your brain!