Perceptual maps and learning

A perceiver doesn’t receive stimuli directly; it builds an internal arrangement of them — which sounds count as similar, which are far apart — and everything downstream reads that arrangement rather than the world. Psychophysics measures the arrangement by measuring judgments.

Two findings from the founding source

The Mainz study of pulsed-sound sequences human-mouse-perceptual-maps reports both:

The map is largely stimulus-driven. Humans and mice, with substantially different auditory hardware, build remarkably similar maps of the same sequences. If two such different systems agree, much of the structure is being supplied by the regularities in the sound, not invented by the listener.

Learning refines rather than rebuilds. The maps stayed stable before, during and after training. So training sharpens discriminations inside an existing structure instead of reorganizing it — a claim about the shape of learning, not merely its occurrence.

Why it belongs beside Kahneman and Jung

The founding corpus’s shared premise is that the mind runs on structure its owner didn’t choose and can’t inspect (synthesis). Here that structure is measurable, cross-species, and present before the learning starts, which makes this the corpus’s most tractable instance — and the standard the other two are read against, since Kahneman’s frames and Jung’s archetypes are the same claim at scales where the evidence gets much harder.

A term to hold lightly

The coverage reaches for world model, a phrase currently doing heavy work in AI. A perceptual similarity map and a learned generative model of an environment are not obviously the same thing, and the shared phrase is an analogy until someone demonstrates otherwise. Worth flagging because this hub’s other spokes use the AI sense constantly.

Related

human-mouse-perceptual-maps · focusing-illusion · analytical-psychology · synthesis