Eye fixation related potentials in a target search task

Healy, Graham; Smeaton, Alan F. · 2011 · OpenAlex-citations

DOI: 10.1109/iembs.2011.6091043

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 feasibility of using Eye Fixation Related Potentials (EFRPs) to detect targets in visual search tasks that require active eye movements, addressing a limitation in previous Brain-Computer Interface (BCI) research. Prior work relied on Rapid Serial Visual Presentation (RSVP) paradigms where targets were highly salient and detectable without eye movements. The authors sought to determine if EEG signals could distinguish between fixations on target versus non-target objects in scenarios where targets are not pre-attentively salient, such as medical imaging or security screening, thereby enabling the identification of both the target and its specific location within an image. The experimental design involved seven subjects performing a visual search task on a screen displaying four corner stimuli, one of which was a target (a broken circle with two lines) and the others non-targets (broken circles with three lines). Subjects followed a prescribed search pattern, fixating on each corner object while maintaining central fixation otherwise, ensuring they were unaware of the target’s identity until the moment of fixation. EEG data was recorded using a 16-channel system, and eye movements were tracked via Electrooculogram (EOG) to timestamp fixations. The researchers extracted EEG windows from 0 to 1000 milliseconds post-fixation. To assess discriminative capability, they employed Support Vector Machine (SVM) classifiers with radial basis functions, using 20-fold cross-validation to calculate Area Under the Curve (AUC) metrics for target versus non-target fixation classification. The results demonstrated distinct neural signatures associated with target detection. ERP analysis revealed a posterior negative component (visual N2) peaking between 250 and 350 milliseconds, which was enhanced for target objects. More significantly, a widely distributed positive component consistent with P3b activity emerged around 500 milliseconds post-fixation for targets, primarily over occipital and parietal regions. Machine learning analysis confirmed the utility of these signals, achieving an average AUC of 0.76 across subjects when using all EEG channels. Classification performance was significantly lower when using only anterior channels (AUC 0.58) or pre-fixation data (AUC 0.50), confirming that the discriminative information arises post-fixation and is localized to posterior brain regions. The findings establish that EFRPs can effectively differentiate between target and non-target fixations even when targets lack pre-attentive salience. This extends the application of EEG-based image annotation beyond simple RSVP tasks to complex visual search scenarios requiring serial inspection. The study confirms that neural markers of target detection are offset from the time of fixation and are detectable via standard ERP components like P3b. This capability supports the development of BCIs for real-world applications, such as assisting radiologists or security screeners, by providing a method to identify targets and their locations within images through neural signals alone.

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 2026-06-26; verification: verified.

Topics

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

• visual search