Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Sensory Encoding and Decision-making in Speech Perception in Noise

Poster D37 in Poster Session D with Social Hour, Friday, October 7, 5:30 - 7:15 pm EDT, Millennium Hall

Jacie R. McHaney1, Bharath Chandrasekaran1; 1University of Pittsburgh

Speech processing in noise (SPIN) has been predominantly analyzed through the lens of sensory processing. SPIN requires reconstructing sound properties and meaningfully mapping them onto established object representations. Object identification has sensory and decisional components that have been mapped to the auditory cortical network and an extended non-auditory network. As noise levels increase, the non-auditory network shows increased activation and more information in representation. Our goal of the current study is to re-examine SPIN using a neurobiologically inspired drift-diffusion model (DDM) of decisional processes that accounts for accuracies and response times. Per the DDM, listeners noisily accumulate sensory evidence until they have acquired sufficient evidence to make a decision. Model parameters tied to the efficiency of evidence accumulation and response caution relate to different neural operations of the decision-making process. Here, we examine the extent to which these computational model parameters relate to objective and subjective experiences of SPIN with an extensive battery of sensory and neural tests. Young adult, native English speakers with normal hearing thresholds participated in this study. To assess decision-making processes during SPIN, participants completed a phoneme in noise categorization task that consisted of 380 trials. Stimuli were synthetically-generated /ba/, /da/, and /ga/ phonemes in quiet or masked in speech-shaped noise at +8, -2, -6, and -9 dB signal-to-noise ratios (SNR). Participants were instructed to categorize phonemes as quickly and as accurately as possible. Accuracies and response times from this task were implemented into a DDM. Here, we examined the evidence accumulation and decision threshold parameters from the DDM to understand the processes underlying decision-making during SPIN. Evidence accumulation reflects extraction of information from the stimulus that is relevant for decision making, wherein lower evidence accumulation rates reflect more difficulty extracting relevant information. The decision threshold parameter is a measure of response caution that reflects the tradeoff between speed and accuracy, where larger decision thresholds indicate greater response caution. Participants also completed a self-report of hearing using the Speech, Spatial, and Qualities of Hearing Scale, measures of cognition, and QuickSIN, a standardized measure of SPIN abilities. In a separate session, participants listened to continuous speech in quiet and in noise while electroencephalography was recorded. Results from the DDM indicate robust relationships between evidence accumulation in noise and self-reported measures of speech abilities. Additionally, both evidence accumulation and decision thresholds in noise strongly correlated with pure tone averages. These results indicate that listeners with better hearing thresholds who report fewer listening difficulties, accumulate evidence more efficiently and are more cautious responders, favoring accuracy over speed of decision-making. Ongoing work in phase two examines the extent to which DDM parameters predict cortical tracking of the continuous speech envelope in quiet and in noise in the same individuals. Taken together, these results suggest that incorporating response times into models of SPIN provides valuable information that could lead to the development of fast, clinically-relevant tests that capture real-world SPIN performance.

Topic Areas: Speech Perception, Computational Approaches

SNL Account Login

Forgot Password?
Create an Account