Bayesian Modeling of the Dynamics of Phase Modulations and their Application to Auditory Evoked Responses at Different Loudness Scales

eMortezapouraghdam, Zeinab; Wilson, Robert C.; eSchwabe, Lars; Strauss, Daniel J. · 2016 · DOAJ

DOI: 10.3389/fncom.2016.00002

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the neural signatures of long-term habituation in auditory selective attention by analyzing the instantaneous phase dynamics of auditory event-related potentials (ERPs). The research is motivated by the need to objectively characterize how attention filters irrelevant sensory information, specifically focusing on the N100 wave component of ERPs. While previous studies often relied on amplitude changes, this work posits that phase jitter—representing a decrease in phase synchronization—is a more robust indicator of habituation. The authors aim to distinguish between stimuli of varying loudness levels (60, 70, 80, and 90 dB SPL) by modeling the gradual changes in phase concentration over trials, addressing the limitations of sliding-window methods which lack temporal precision and the ability to incorporate prior knowledge. To achieve this, the authors recorded EEG data from 19 healthy subjects exposed to pure-tone beeps at four distinct loudness levels. The data underwent rigorous preprocessing, including bandpass filtering, artifact removal, and denoising using a non-local means algorithm to preserve morphological regularities. Instantaneous phase information was extracted from the N100 wave (80–120 ms post-stimulus) using continuous wavelet transforms. The core methodological contribution is a Bayesian forward-backward model that estimates the concentration parameter ($\kappa$) of a von Mises distribution for the phase data. This model assumes that habituation is a gradual process, enforcing smoothness in the evolution of the concentration parameter over trials. By using a forward-backward sweep, the method mitigates initialization biases and provides high-temporal-resolution estimates of phase dispersion, allowing for the incorporation of prior knowledge about the habituation process. The results demonstrate that the Bayesian model effectively differentiates between stimuli based on the dynamics of phase concentration. For softer stimuli (60 and 70 dB), which are easier to habituate, the phase data showed increased dispersion (lower concentration) over the course of the experiment, reflecting a drift in attentional binding. In contrast, louder, aversive stimuli (80 and 90 dB) maintained high phase synchronization (high concentration) throughout the trials, indicating sustained attention and an absence of habituation. The Bayesian approach proved superior to moving-window maximum-likelihood estimates in detecting these gradual changes with higher precision and robustness against noise. The significance of this work lies in providing a sophisticated, objective tool for analyzing circular data in neurophysiology. By modeling phase dynamics rather than just amplitude, the study offers a more sensitive measure of attentional states and habituation. This method has potential clinical applications, such as calibrating cochlear implants for non-cooperative patients or understanding pathological attentional binding in conditions like schizophrenia and tinnitus. The ability to distinguish between closely spaced loudness levels using phase information highlights the utility of Bayesian modeling in capturing subtle neural signatures of selective attention.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• signal detection theory