Eccentricity Confound in EEG-based Visual Attention Decoding from Gaze-Fixated Neural Tracking of Motion in Natural Videos
Yuanyuan Yao 1 , Celina Salamanca González 1 , Simon Geirnaert 1,2 , Céline R. Gillebert 3,5 , Tinne Tuytelaars 4 , Alexander Bertrand 1 · 2026
Read Eccentricity Confound in EEG-based Visual Attention Decoding from Gaze-Fixated Neural Tracking of Motion in Natural Videos by Yuanyuan Yao 1 , Celina Salamanca González 1 , Simon Geirnaert 1,2 , Céline R. Gillebert 3,5 , Tinne Tuytelaars 4 , Alexander Bertrand 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-15223.txt.
Introduction
Yao and colleagues show that object distance from fixation confounds EEG decoding of visual attention. Standard methods read stronger motion-neural coupling as higher attention, but eye-movement artifacts and stimulus traits can mimic it. With gaze fixed across three tasks varying eccentricity and attention, neural tracking of motion still holds and shifts with attention, yet weakens at larger eccentricities. The team concludes past free-viewing results reflect real neural processing, but eccentricity limits decoders that equate coupling with attention alone.
Catalog record
| CREATOR | Yuanyuan Yao 1 , Celina Salamanca González 1 , Simon Geirnaert 1,2 , Céline R. Gillebert 3,5 , Tinne Tuytelaars 4 , Alexander Bertrand 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41217 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.15223
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Eccentricity Confound in EEG-based Visual Attention Decoding from Gaze-Fixated Neural Tracking of Motion in Natural Videos Yuanyuan Yao 1 , Celina Salamanca González 1 , Simon Geirnaert 1,2 , Céline R. Gillebert 3,5 , Tinne Tuytelaars 4 , Alexander Bertrand 1 Abstract Objective. Decoding visual attention from brain signals during naturalistic video viewing has emerged as a new direction in brain-computer interface research. Current methods assume that stronger coupling between object motion and neural activity indicates higher attention, but this can be confounded by eye movement artifacts and stimulus properties. This study investigates how visual eccentricity—the distance between a visual object and the fixation point—affects neural responses when eye movement artifacts are controlled. Approach. EEG signals were recorded across three tasks that manipulated object eccentricity and attention conditions while participants maintained gaze fixation. Correlation analysis and match-mismatch decoding were performed to quantify the neural tracking of object motion. Main results. The analysis supports three conclusions: (1) neural tracking of object motion in natural videos works under gaze fixation; (2) the strength of neural tracking under gaze fixation is predictive of attention; and (3) there exists a significant eccentricity confound in the EEG responses, with poorer neural tracking of motion at larger eccentricities. Significance. These results provide critical evidence that findings from previous free-viewing studies reflect genuine neural processing rather than mere oculomotor artifacts. However, the identified eccentricity effect highlights a major limitation for current decoding approaches that assume coupling strength reflects attention levels alone. † † address: 1 KU Leuven, Department of Electrical Engineering, STADIUS Center for Dynamical Systems, Signal Processing and Data Analytics, Belgium 2 KU Leuven, Department of Neurosciences, Research Group ExpORL, Belgium 3 KU Leuven, Brain and Cognition, Belgium 4 KU Leuven, Department of Electrical Engineering, Processing Speech and Images (PSI), Belgium 5 Leuven Brain Institute (LBI), Belgium 1 Introduction Neural activity is modulated by attention: neurons processing task-relevant information exhibit increased firing rates and reduced variability, while those processing irrelevant information are suppressed [ 21 ] . Therefore, stronger neural responses are expected with attended stimuli, leading to stronger temporal coupling between stimulus and neural activity. This modulation has been observed not only with simple synthetic stimuli but also with naturalistic speech, images, and videos [ 16 , 1 , 6 , 24 ] . These findings form the foundation for numerous studies aiming to decode visual or auditory attention from brain signals [ 23 , 2 , 18 ] , where stronger stimulus-response coupling is typically interpreted as evidence of higher attention levels towards the stimulus. However, attention is not the only factor that influences stimulus-brain coupling strength. When using naturalistic stimuli, various confounding factors can modulate this coupling. For example, Li et al. [ 15 ] reconstructed speech envelopes from electroencephalography (EEG) signals and used the correlation between reconstructed and actual envelopes as a metric for auditory attention. While this score was higher when participants were attentive and lower during distraction, it also correlated with the time-varying standard deviation of the speech envelope, suggesting that the score was also sensitive to stimulus dynamics rather than attention alone. When this confound was not addressed, decoding performance deteriorated. This exemplifies that stimulus-brain coupling is not a pure measure of top-down goal-directed attention, but is also dependent on bottom-up sensory stimulus properties [ 3 , 11 ] . Moreover, such bottom-up influences may not cancel out in complex naturalistic conditions and could systematically bias decoding results in real-world applications. These concerns extend naturally to newly emerging EEG-based visual attention decoding paradigms with potential applications in education, neuromarketing, and clinical diagnostics [ 6 , 24 , 23 ] . We hypothesize that the spatial location of visual objects relative to gaze position (i.e., eccentricity, calculated from the center of the object’s bounding box) could be a potential confound: a higher stimulus-brain coupling may not necessarily indicate higher attention, but could also result from the object being located at a more central location in the visual field. This is motivated by the well-established nonuniformity of visual processing, with maximal visual acuity and pattern recognition capabilities at the fovea and a rapid decline toward the periphery [ 4 , 20 ] . A direct link between eccentricity and neural responses has been demonstrated previously. For instance, in a functional magnetic resonance imaging study by Wang et al. [ 22 ] , participants were presented with images of different categories (faces, houses, etc.) at different eccentricity positions. They reported that the neural activation was most intense when stimuli were in the central visual field, and the magnitude of responses declined as eccentricity increased. Beyond these stimulus-dependent effects, eye movements present another persistent confound in free-viewing paradigms. Under natural conditions, attention is typically accompanied by overt behavior such as directing gaze toward the attended object, which is known as overt attention. This is particularly relevant for paradigms relying on the motion information in naturalistic videos, as eye movements (saccades and smooth pursuit) are naturally evoked by the movements in the videos, and thus may strongly correlate with the video motion features. These eye movements can influence stimulus-EEG correlations through several mechanisms: (1) (residual) electrooculogram (EOG) artifacts in the EEG signals, (2) neural motor activity associated with the planning and execution of eye movements, and (3) neural responses to gaze-induced visual motion, akin to a moving-camera effect where the eyes serve as the camera. This ambiguity undermines findings from prior studies that reported significant neural tracking of video motion under free-viewing conditions [ 6 , 24 ] , since it remains unclear to what extent these effects reflect genuine neural processing of motion information rather than oculomotor contributions. In [ 23 ] , eye gaze data were explicitly shown to be correlated with video motion features, and using gaze data alone could achieve comparable attention decoding performance to EEG-based decoders, which again raises concerns about the neural origin of the reported effects. Similarly, Ki et al. [ 13 ] showed that the neural responses are spatially selective and are enhanced for task-related locations, but since their paradigm relied on overt attention, doubts remain as to whether the reported enhancement was simply an artifact of eye gaze bias toward task-relevant locations, rather than a genuine neural correlate of covert attentional selection. Since disentangling genuine neural correlates of attention from oculomotor confounds is particularly challenging under free-viewing conditions, we follow a common strategy in cognitive neuroscience: employing a covert attention paradigm. Participants maintain fixation while attending to stimuli in the visual periphery, thereby eliminating eye-movement confounds at the cost of reduced ecological validity. This controlled approach allows us to rigorously test two key questions: (1) whether neural tracking of video motion persists in the absence of eye movement contributions, and (2) how eccentricity of visual stimulation systematically influences neural tracking of naturalistic videos. The remainder of this paper is organized as follows. Section 2 describes the experimental protocol, including visual stimuli, tasks, and data acquisition. Section 3 details preprocessing and correlation analysis methods. Section 4 presents results on the effects of attention and eccentricity on single-subject decoding performance and inter-subject correlation (ISC). Section 5 discusses the implications of our findings. Finally, Section 6 concludes the paper. 2 Experimental Protocol 14 healthy young adults with normal