Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

Confidence in Sound Localization Reflects Calibrated Uncertainty Estimation

Lakshmi Narasimhan Govindarajan¹, Sagarika Alavilli², Josh Mcdermott¹; ¹Massachusetts Institute of Technology, ²Harvard University, Harvard University

Presenter: Lakshmi Narasimhan Govindarajan

Humans localize sounds using a combination of binaural and monaural cues. However, the location of a sound remains ambiguous under many conditions. Because sound localization is often used to guide behavior, representing the uncertainty of a sound’s location is likely to be critical to decisions about where and when to act. However, little is known about whether humans represent the uncertainty associated with a sound’s location and whether any such representations are calibrated to the accuracy of localization. To study these issues, we developed a new class of stimulus-computable models to enable the representation of uncertainty. We optimized the model for sound localization in natural conditions and then compared its uncertainty estimates to those of humans.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF