Neurobehavioral, Physiological, and Computational Processes of Auditory Object Learning in Mammals
Abstract
Previous efforts to build systems matching dolphins# expert echolocation abilities emphasized the acoustics of the task. Sonar systems improved # but still failed to match dolphin performance. We believe a key limitation of past approaches is that they ignored the cognitive processes allowing a dolphin to learn natural structures in sound, categorize acoustically variable sounds as #the same#when they come from the same source, and focus on whatever cues robustly identify an object in a given context by recruiting specialized brain networks. Our objective is to understand the cognitive and neural mechanisms allowing marine mammals and humans to learnhow to identify auditory objects in cluttered scenes. Our proposed technical approaches use behavior and neural measures to see ifcognitive processes control echolocation and hearing for marine mammals # as in humans. We will focus on dolphins but will also test sea lions, a trainable marine mammal whose smaller size and comfort on land support neuroimaging approaches less practical in dolphins. In humans, we will use behavior, neural measures, and neural stimulation to explore how cognitive brain networks coordinate auditory perception in complex settings. We will develop instrumentation to allow multi-modal neural and physiological sensing and build models operating at different levels of abstraction to gain insight into how different biological listeners learn to form auditory objects from limited data. Thrust 1 explores behavior in free-swimming, echolocating dolphins as they learn to represent and discriminate objects with varying acoustic backscatter. We will develop targets and distractors with known echo properties to identify the acoustic features that guide dolphins# decisions. These results will be compared to features used by convolutional neural nets to classify the same targets. We will quantify how dolphins move as they ensonify target and distractor objects and model the behavior to gain insight intothe difficulty and latency of their decisions. Computational models of decision making will be developed to characterize and predict dolphin behaviors. Using electroencephalography (EEG) and behavior, Thrust 2 compares the ability of dolphins to learn probabilistic temporal structure in sound sequences from two classes of behaviorally relevant sounds: target echoes (with stimulus choices guided by results of Thrust 1) and #syllables# from signature whistles of different individual dolphins. Quantitativemachine learning methods will be used to analyze EEG responses to reveal how learning changes object representation in the brain. Thrust 3 tests sea lions with behavior, EEG, and functional near-infrared spectroscopy (fNIRS) to determine whether these animals have specialized brain networks for spatial and nonspatial cognitive control. In humans, Thrust 3 will use behavior, EEG, fNIRS, and functional magnetic resonance imaging to characterize the roles of spatial and nonspatial cognitive brain networks. We will also explore whether we can manipulate attention and enhance auditory learning via transcranial electrical stimulation.Illuminating how marinemammals and humans learn and represent auditory objects will provide new perspectives that may lead to novel sonar signal processing approaches to detect and identify objects in complex, cluttered scenes. Similarities # but even more, differences # between the brain networks and processes that echolocating dolphins and other mammals use to learn and recognize sound patterns in complex settings will provide new insights into what acoustic information provides robust and reliable information in different environments. Studies into the efficacy of noninvasive neural stimulation could lead to safe methods for improving human operator performance in cognitively demanding situations.Approved for Public Release.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 12, 2023
- Source ID
- N000142312065
Entities
People
- Barbara Shinn-Cunningham
Organizations
- Carnegie Mellon University
- Office of Naval Research
- United States Navy