Back to search

East Tennessee State University

Testing P300-Based Biometric Authentication Paradigms

Abstract

dc:description.abstract

Bench research suggests that P300 amplitude is greater in response to self-relevant stimuli compared to familiar stimuli. This effect could enable a novel method of biometric authentication based on the self-relevant P300. This dissertation aims to: (1) determine the optimal presentation paradigm for leveraging self-relevance in P300-based biometric authentication, and (2) investigate P300 responses from self-relevant stimuli versus stimuli learned over the course of a week. Two presentation paradigms were compared – rapid serial visual presentation (RSVP) and matrix presentation (MP). Participant-specific RSVP and MP classifiers were trained by applying stepwise linear discriminant analysis to data collected during presentations of the participant’s own photos (Self-Relevant targets) in each respective paradigm during session one. Classifiers were tested using within-subject data to model genuine users and cross-subject data to model intruders. Within-subject testing utilized session two responses to self-relevant stimuli collected one week later. Rigorous simulation of intruder responses was achieved by instructing participants to study a set of modeled stranger photos (Learned targets) over the duration of the week. Cross-subject tests were conducted using participant responses to the learned stimuli collected in the second session. The MP classifiers had a mean within-subject classification accuracy of 99.92% compared to 96.75% in the RSVP classifiers. Conversely, the mean cross-subject classification accuracy of 4 the MP classifiers was 91.85% compared to 46.48% in the RSVP classifiers. However, generalized linear mixed effects modeling failed to find a statistically significant TestData (within-subject, cross-subject) × Paradigm (RSVP, MP) interaction in the classification accuracy data, likely due to limited statistical power caused by a ceiling effect in within-subject MP classification data. Linear mixed effects modeling suggested that presentation of Self-Relevant targets elicited greater P300 amplitude than Learned targets. P300 amplitudes from these two stimulus types did not change between sessions. An exploratory analysis revealed that the P300 amplitude difference between Self-Relevant and Learned targets was greatest in the RSVP paradigm. The results of this study support the feasibility of creating biometric authentication based on the self-relevant P300. Future research should replicate these results with larger samples and test a hybrid MP-RSVP paradigm that optimizes comfort and security.

Degree

thesis:*
Name thesis:degree_name
PhD (Doctor of Philosophy)
Level thesis:degree_level
Dissertation - unrestricted
Discipline thesis:degree_discipline
Psychology
Grantor dc:publisher
East Tennessee State University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Razzak, Jordan

Subjects

dc:subject × 9

Rights

dc:rights
Statement dc:rights
  • Copyright 2026 by Jordan Sahir Razzak. All rights reserved.
Language dc:language
eng

Identifiers

dc:identifier.*
Repository record dc:identifier
https://dc.etsu.edu/etd/4727

Chain of custody

source
Harvested from
East Tennessee State University
Base URL
dc.etsu.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Razzak, Jordan. Testing P300-Based Biometric Authentication Paradigms. Dissertation - unrestricted thesis, East Tennessee State University, 2026. https://dc.etsu.edu/etd/4727