Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Adaptation in an uncertain environment: Pupil dilation as a proxy for locus coeruleus-norepinephrine activity and error-based learning
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
This poster is part of the Sandbox Series.
Hannah Mechtenberg1, Emily Myers1; 1University of Connecticut
INTRODUCTION Listeners regularly encounter unfamiliar talkers, noisy places, and idiosyncratic pronunciations of phonemes. Amidst this ocean of novelty and variability, there must be a system in place that allows each listener to rapidly adapt when necessary while simultaneously balancing the stability of existing representations (Friston et al., 2020; Kleinschmidt & Jaeger, 2015; Sohoglu et al., 2012). This problem becomes ever more complex in scenarios where listeners must contend with multiple concurrent challenges to perception and comprehension, such as when encountering a new talker with rapidly fluctuating noise occurring in the background. Pupillometry, or the measure of pupil area over time, has grown in popularity as a method by which to measure online effort allocation during speech perception and other demanding tasks (Brown et al., 2020; Zekveld & Kramer, 2014). Pupil dilation, modulated via the norepinephrine system (Poe et al., 2020), is also thought to index “arousal states” which are linked to the receptivity and reactivity of a given population of neurons wherein a highly aroused state is proposed to lead to greater plasticity and ability to adapt to a changing environment (Aston-Jones & Cohen, 2005). The proposed project tests predictions from a novel hypothesis of adaptation, wherein subcortical error signals—including those originating in the locus coeruleus—to first alert the listener of a challenging listening scenario and then iteratively, and specifically, tune the relevant cortical regions to achieve improved perception and/or comprehension. Using pupillometry as a proxy for the LC-NE system, this project seeks to test how listeners enter into an adaptive state, thus allowing for the recruitment of mental resources to improve their ability to pull a talker’s voice out from background noise. METHODS. In the proposed study, participants adapt not only to short, spoken sentences in different noise levels (Low, Medium, Hard) but also three different talkers (Female 1, Female 2, Female 3). By having only 42 trials for each novel talker, the task maximizes the amount of time spent in an adaptive state. Each block features a different talker, and the order of talkers was counterbalanced across participants. Within each block, there are 14 sentences embedded in each Low noise (+5dB SNR), Medium noise (+0dB SNR), and High noise (-5dB SNR) condition, which were presented in 12 different fixed, pseudorandom orders. Participants are tasked with listening to the sentences embedded in noise and verbally repeating what they heard. We hypothesize that we will replicate findings that there will be larger overall pupil dilations in high noise compared to low noise (e.g., Van Engen & McLaughlin, 2018), but, critically, we expect to find larger pupil dilations early in a block, when uncertainty is highest which then reduce as listeners improve their ability to understand the talker in noise. We have also included an exploratory cue manipulation, which may further capture whether the LC-NE system is proactive and flexible during adaptation to speech. CONCLUSION. The results of this project will create new avenues for future research into the role of error-driven adaptation, and an integrated LC-NE system.
Topic Areas: Speech Perception,