The model looks inward

Interpretability as a speculative science of interiors

Is a synthetic psychology or sociology possible if we locate it in latent space — in embeddings, features, and circuits rather than in chat? The same difficulty returns as interspecies interpretability: how to read a whale, an octopus, or any other Umwelt without collapsing it into human speech. von Uexküll’s Umwelt is the condition of any such reading — a lived world, not a dataset. Can brain, network, and other-species research cross-pollinate without identity? Each card speculates past the interface: toward a new epistemology, or a language that speaks in the architecture itself — folding into itself, or stepping beyond recursion.

Press Space. Each speculation rotates an inquiry, a site — model, brain, or another Umwelt — and a gesture of thought. Open from any ontology entry to ground the speculation in that concept.

Ontology source

Each speculation rotates an inquiry — synthetic psychology, interspecies interpretability, Umwelt, brain–network comparison — a site (latent space, another Umwelt, ocean minds), and a gesture: analogy, crossing Umwelten, fold, or a step beyond recursion. Generated from this entry when present. Current papers or methods may be looked up on the live web; the ontology stays first.