MIT lab's PNAS paper: humans use internalized object acoustics to parse impact sounds
JoshHMcDermott · x · 2026-11-10
McDermott's lab (Agarwal, Traer, Schwartz & McDermott) published a PNAS 2026 paper showing that perceiving physical variables from sound is an ill-posed scene analysis problem, solved by humans via internalized knowledge of how objects make sound. The paper includes a generative model, human experiments, and impulse responses; data and code are open-sourced on GitHub (mcdermottLab/impact-sound-perception), aiming to open doors for more work on this aspect of human hearing.
More from Research
- Wiener's 1960 Alignment Warning Revisited: Patterning as a Control Problem — geoffreyirving · 2026-11-11
- NEO-VLA: encoder-free native VLA turns pixels into actions, beats encoder-based baselines — liuziwei7 · 2026-11-11
- Why AI Slop Is So Frustrating to Read, per Zonder Labs — floguo · 2026-11-11
- Terminal-Bench thanks community auditors, publishes task flaw reporting process — ajratner · 2026-11-11
- Berkeley Paper: Harness Tweaks Cut Agent Costs Up to 2.6x Without Changing Models — dair_ai · 2026-11-11
- Reframe: Post-Training Framework Makes Visual Agents Actually Use What They See — _akhaliq · 2026-11-11