P292 Single-trial detection of lambda responses in free-viewing EEG measurements
Iffah Syafiqah Suhaili*1, Zoltan Nagy1,2, Zoltan Juhasz1 1Department of Electrical Engineering and Information Systems, University of Pannonia, 8200 Veszprem, Hungary 2Heart and Vascular Centre, Semmelweis University, 1085 Budapest, Hungary
*Email: ssiffah@phd.uni-pannon.hu Introduction
Visual lambda responses are occipital activations evoked by saccadic eye movements. Their study is important for understanding visual processing during natural viewing conditions. Traditionally, lambda waves are detected by averaging many short epochs in which lambda responses are phase-locked to stimulus. In natural viewing conditions, especially in experiments where trials span many seconds, their detection is difficult, and averaged ERP/based methods are not applicable as saccades occur in an unpredictable, non-time-locked manner. This study presents a novel method that can detect individual lambda responses in single trials without averaging, allowing for more naturalistic experimental designs.
Methods 80 art paintings were presented to 29 healthy volunteers. Each painting was displayed for 8 seconds in a random order, each followed by a 4-second blank screen. 128-channel EEG data were registered using a Biosemi ActiveTwo EEG device. Participants were instructed to explore the painting and then respond by pressing a LIKE or DISLIKE button. After high-pass (1 Hz) and low-pass (40 Hz) filtering, the signals were decomposed into independent components using the Infomax Independent Component Analysis (ICA) method[1]. Simultaneously, eye movements were recorded with a Tobii Pro Fusion eye-tracker at 250 Hz sampling rate. As the final step of the pre-processing, the EEG and eye-tracking data were synchronized.
Results Besides the usual eye-related artefact components (horizontal and vertical eye movements), ICA decomposition produced a characteristic component displaying a distinct, rhythmic pulse-train pattern during the 8-second viewing period that diminished in the 4-second blank interval. The location of this brain source component was on the parieto-occipital electrodes (Pz –Oz). Overlaying the eye-tracking events (saccade onset and offset) on the ICA activation plot clearly shows that the pulses are time-locked to the saccade offsets, with an average latency of 82 ms. Fig. 1 illustrates these findings in detail.
Discussion ICA can reliably detect saccade-related lambda waves in free-viewing experiments lasting at least 15 minutes. This method helps determine the number and temporal distribution of saccades characterizing perceptual behaviour (e.g. engagement, attention) in natural viewing experiments. Lambda wave properties (peak amplitude, peak latency, inter-peak distance) allow further quantitative analysis and can act as synchronization markers in segmenting sessions into saccade-evoked epochs locked to lambda peaks. Identifying the lambda component improves eye-movement artefact removal by including parieto-occipital activations. We hope this method will lead to new experimental approaches that advance our understanding of the human visual system.
Figure 1. ICA results highlighting saccade-related lambda waves. a) ICA activation plot of two stimulus-locked paintings (epoch 22 and 23) highlighting the lambda response component (IC 3) occur only during the stimulus presentation. b) Scalp topography map of IC 3 over parieto-occipital region. c) A zoomed-in single trial segment circled in (a), displaying three lambda peaks aligned with saccade events. Acknowledgements This research was funded by the University Research Fellowship Programme (EKOP) (Code: 2024-2.1.1-EKOP-2024-00025/58) of Ministry of Culture and Innovation from the National Fund for Research, Development and Innovation. References 1.Lee, T.-W., Girolami, M., & Sejnowski, T. J. (1999). Independent Component Analysis Using an Extended Infomax Algorithm for Mixed Subgaussian and Supergaussian Sources.Neural Computation,11(2), 417–441. Neural Computation. https://doi.org/10.1162/089976699300016719
