Specialization of neural processing during active acoustic sensing in marine mammals and humans
Abstract
"This project combines behavioral and neural measures to study the mechanisms andstrategies both odontocetes and humans use to hear (i.e., process acoustic energyemitted by sound sources) and how this relates to how odontocetes echolocate. Workis divided into three interrelated thrusts. Thrust I will explore how dolphins andporpoises process sounds in complex settings through a combination of behavioral andnon-invasive neural measures. We will study how odontocetes analyze different classesof sounds to test the hypothesis that different brain networks are engaged whenprocessing echolocation sounds versus communication signals. In particular, we willtest for lateralization of auditory processing of communication vs echolocation signals indolphins and porpoises. Active sensing behavior will be characterized in freelyswimming animals and in stationed animals using non-invasive EEG. We will developEEG for freely swimming animals and continue exploring the utility of fNIRS to captureneural signals in odontocetes. Thrust II will build on our ongoing work studying thebrain networks controlling hearing in humans. As in our current work, we will combinebehavioral performance and neuroimaging methods (EEG, MEG, fMRI, and fNIRS)using new data fusion techniques to understand the neural mechanisms of activesensing of acoustic inputs. We will also develop an EEG-compatible virtual-realityplatform capable of generating realistic auditory and visual scenes that respondrealistically to user movements. Using this new tool, we will measure behavioral andneural signals to understand how information is processed in specialized brain networksas listeners engage in games that require active exploration of a simulated environment.Thrust III will develop echolocation-specific models for auditory scene analysis. Allreceived signals in echolocation scenes contain delayed, attenuated, and filtered copiesof the same signal. This suggests that the time-frequency footprint used as animportant cue for stream formation in passive listening is far less informative inecholocation scene analysis. In contrast, echolocating animals control which spatialregion they interrogate by choosing where they direct their sonar beams. The animalsalso influence when echoes are received by choosing when to transmit sonar signals.We will investigate the role of both bottom-up echo feature propagation and top-downattention processes which modulate how features are combined to form auditoryobjects. The models will also investigate how object representations are refinedthrough successive sonar transmissions, and how the signaling strategies in bothspatial directivity of sonar beams and timing of echoes are influenced by the uncertaintyabout objects within the scene. The model predictions will be compared with recordedanimal behavior in echolocation experiments from Thrust I for validation and refinement."
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 31, 2020
- Source ID
- N000142012709
Entities
People
- Barbara Shinn-Cunningham
Organizations
- Massachusetts Institute of Technology
- Office of Naval Research
- United States Navy