Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Altered phonetic representation in tumor-afflicted auditory cortex during speech perception
Poster Session D, Saturday, September 13, 5:00 - 6:30 pm, Field House
This poster is part of the Sandbox Series.
Yike Li1, Sena Oten2, Shawn Hervey-Jumper2, David Brang1; 1University of Michigan, Ann Arbor, 2University of California, San Francisco
Glial brain tumors form electrical and chemical synapses with healthy neurons to accelerate tumor proliferation, potentially altering local neural computations. Prior work has shown that glioma-infiltrated cortex remains similarly active during speech production but with decreased entropy, suggesting reduced information processing capacity. The current study investigates how tumor presence affects auditory information processing in the human superior temporal gyrus (STG) during naturalistic speech perception. Subdural electrocorticography (ECoG) data were collected from tumor patients during neurosurgery. Task-related changes in high-gamma power (HGp, 70-150 Hz) were used to measure neural selectivity for phonetic features. Distributed changes in population responses were quantified by the classification accuracy of support vector machine (SVM) classifiers trained to decode phonetic categories from HGp. Preliminary results show that while tumor-afflicted areas achieve a comparable mean decoding accuracy to normal-appearing regions, they exhibit greater variability in decodability, suggesting a noisier phonetic representation. These findings align with previous research on phonetic feature encoding in human STG and provide new insights into speech information processing in the human auditory cortex. This work highlights the partial preservation of neural computation in tumor-infiltrated auditory regions, which is both scientifically and clinically significant as it offers a foundation for further understanding the mechanism of altered neural decodability as well as perceptual outcomes in patients following surgical resection.
Topic Areas: Speech Perception, Computational Approaches