Pillar 1 · The Crow
Why this species, before any model.
Of the animals whose voices we have any business asking models to map, the American crow (Corvus brachyrhynchos) is one of the strongest candidates. A brain-to-body ratio comparable to great apes. Documented tool use. Individual recognition of human faces across years. Cooperative family groups that hold territories, share food, and pass behavior down generations.
And a repertoire dense enough to warrant a map. Decades of work before ALP gave us six or eight named call types and a hunch that there was more. The new methods are recovering the "more" — graded variation, individual signature, dialect, hints of composition — visible at once on the same shared geometry.
Sub-page
↓ in M5
Vocal anatomy
How a crow makes sound — the syrinx, two-source vocal production, the 200 Hz – 8 kHz frequency window, and how it differs from a human larynx.
Sub-page · interactive
Repertoire atlas
The interactive vocal map. ~800 points, 9 clusters, click any to read context.
Sub-page
↓ in M5
Cognition & society
Mirror self-recognition, tool use, social learning, long-term human recognition (Marzluff mask experiments).