You can also find my papers on my Google Scholar profile
(Disclaimer: The papers available on this website are provided for personal and educational use only. Any reproduction, distribution, or use of these materials for other purposes is strictly prohibited.)
(*) denotes corresponding/senior authorship
Natural Reading
Lexical Word Processing is Unaffected by Rapid Invisible Frequency Tagging in Reading: Evidence from Eye Movements
Prigorkina, V., Sheridan, H., Frisson, S., & Pan, Y. (*) (2025, September 9, under review)
📄 Paper | 💻 Code | 💾 datasets
Supervised a master’s student project. Here, we tried to address a fundamental question about the validity of RIFT: does high-frequency flicker alter eye movement patterns during reading? By analysing both the amplitude and latency of the classic word frequency effect, we found the answer to be no. Through careful experimental design and thorough data analysis, this study provides strong evidence for the validity of RIFT. RIFT opens up new possibilities for measuring attention in active vision, remaining imperceptible to participants and causing no interference with ongoing processing.
Parallel and dynamic attention allocation during natural reading
Pan, Y. (*), Frisson, S., Snell, J., Federmeier, K. D., & Jensen, O. (2025, bioRxiv, under review)
📄 Paper | 💻 Code | 💾 datasets
Using RIFT, we tagged two adjacent words with two different frequencies and observed neural evidence for both the lexical parafoveal effect (replicating our Nat Comm 2021 findings) and the foveal load effect. Importantly, we observed a dynamic shift of attention between foveal and parafoveal processing, depending on the processing demands of the currently fixated word. This flexible allocation of attentional resources appears to be key to fluent reading. Furthermore, this paper has important methodological significance as we extend RIFT from tagging a single item to tagging multiple items simultaneously, demonstrating its potential as a powerful method for tracking attention across multiple items.
Emotional content interacts with word frequency during parafoveal reading
Naveed, T., Evans, A., Ferrante, O., & Pan, Y. (*) (2024, under review)
📄 Paper | 💻 Code | 💾 datasets
Supervised an undergraduate student project. This time, we investigated parafoveal processing from the perspective of emotion. We found an intricate interaction between emotion and word frequency in parafoveal processing, a region where attentional resources are highly limited. This striking result underscores the processing priority of emotional information even when attention is constrained.
Fast hierarchical processing of orthographic and semantic parafoveal information during natural reading
Wang, L., Frisson, S., Pan, Y. (*), & Jensen, O. (2025, Nature Communications, 16(1), 8893)
📄 Paper | 💻 Code | 💾 datasets
Co-supervised PhD project. We applied Representational Similarity Analysis (RSA) to decode word representations at both orthographic and semantic levels, at different temporal latencies and in different brain regions. Crucially, this decoding was performed for parafoveal words, demonstrating RSA as an additional tool, alongside RIFT, for investigating parafoveal processing during reading.
Early parafoveal semantic integration in natural reading
Pan, Y. (*), Frisson, S., Federmeier, D. K., Jensen, O. (2024, eLife, 12, RP91327)
📄 Paper | 💻 Code | 💾 datasets
How much information can we extract from a word before we even fixate on it? Using Rapid Invisible Frequency Tagging (RIFT) in a semantic violation paradigm, we found that violation effects emerged even before readers directly fixated on the violating word. This suggests that readers not only extract semantic meaning from parafoveal words but also integrate this meaning into the preceding context, leading to a violation effect. These findings highlight the remarkable depth of parafoveal pre-processing and explain how words can be processed so efficiently during natural reading.
Saccades are locked to the phase of alpha oscillations during natural reading
Pan, Y. (*), Popov, T., Frisson, S., & Jensen, O. (2023, PLoS Biology, 21(1), e3001968)
📄 Paper | 💻 Code | 💾 datasets
Showed that oculomotor activity in reading is not noise but plays a critical role in word processing by phase-locking to endogenous alpha oscillations. This opens up new avenues for understanding the functional roles of neuronal oscillations in naturalistic scenarios involving eye movements.
An oscillatory pipelining mechanism supporting previewing during visual exploration and reading
Jensen, O., Pan, Y., Frisson, S., & Wang, L. (2021, Trends in Cognitive Sciences, 25(12), 1033-1044)
📄 Paper
Developed a neuroscience framework to explain how the human brain can process the currently fixated item (e.g., images, words) alongside the to-be-fixated item, while avoiding interference between foveal and parafoveal representations. We propose that this is achieved by encoding the two representations in different phases of the alpha cycle — an alpha-oscillation-based pipelining mechanism.
Neural evidence for lexical parafoveal processing
Pan, Y. (*), Frisson, S., & Jensen, O. (2021, Nature Communications, 12(1), 1-9)
📄 Paper | 💻 Code | 💾 datasets
Proposed a new technique, Rapid Invisible Frequency Tagging (RIFT), to measure attention during reading. Using this method, we showed that readers engage in substantial pre-processing of a word before fixating on it, including extracting lexical information (i.e., word frequency information), which facilitates subsequent processing. This work establishes a novel paradigm for studying attention in active vision, such as natural reading and visual search.
Attention & Memory & Perception
Higher baseline alpha power is associated with faster responses in visual search
Duecker, K., Shapiro, K. L., Hanslmayr, S., Griffiths, B. J., Quinn, A. J., Wolfe, J. M., Pan, Y., Pastuszak, A., & Jensen, O. (2025, bioRxiv)
📄 Paper
Combined a General Linear Model (GLM) approach with Rapid Invisible Frequency Tagging (RIFT) in a visual search task and found a correlation between alpha activity and task performance (i.e., higher alpha power, faster responses). Further analyses showed no correlation between alpha power and distractor responses, suggesting that increased alpha reflects enhanced top-down control in visual search.
Guided visual search is associated with target boosting and distractor suppression in early visual cortex
Duecker, K., Shapiro, K. L., Hanslmayr, S., Griffiths, B. J., Pan, Y., Wolfe, J. M., & Jensen, O. (2025, Communications Biology, 8 (1), 912)
📄 Paper | 💻 Code | 💾 datasets
Applied rapid invisible frequency tagging (RIFT) in a visual search paradigm and found evidence for both target enhancement and distractor suppression during guided vs. unguided search. The results show that feature-guidance modulates neuronal excitability as early as the primary visual cortex, supporting a priority-map-based mechanism of attention.
Anterior thalamic stimulation improves working memory precision judgments
Liu, J., Yu, T., Wu, J., Pan, Y., Tan, Z., Liu, R., Wang, X., Ren, L., & Wang, L. (2021, Brain stimulation 14 (5), 1073-1080)
📄 Paper
Investigated the role of the hippocampal–anterior thalamic pathway in working memory precision by delivering intracranial electrical stimulation to the anterior nucleus of the thalamus (ANT). We found that ANT stimulation enhanced working memory precision (i.e., not the number of items remembered, but the clarity and accuracy of those memories). Working with epilepsy patients provided a unique opportunity to conduct invasive (yet safe) brain stimulation and test the causal relationship between a specific brain region and a specific cognitive function.
Neural activity is dynamically modulated by memory load during the maintenance of spatial objects
Pan, Y., Tan, Z., Gao, Z., Li, Y., & Wang, L. (2018, Frontiers in psychology, 9, 1071)
📄 Paper
My first neuroscience project used scalp EEG to investigate the effect of working memory load (i.e., the number of items maintained in mind) on brain activity. We found that neural responses did not change linearly with memory load but instead showed dynamic modulation. In addition, occipital alpha power decreased with increasing load, supporting the functional role of alpha oscillations in “gating by inhibition”.
Lateral prefrontal cortex triggers bistable perception switching using human intracranial EEG
Xu, H., Qin, C., Bai, Y., Tan, Z., Pan, Y., Li, Y., Zhou, W., Zhang, T., & Wang, L. (2016, Advances in Psychological Science, 24 (Suppl.), 24)
📄 Paper
Investigated neural circuits underlying bistable perception in epilepsy patients with intracranial electrodes while performing a structure-from-motion task. We found that increased high-gamma activity in the lateral prefrontal and posterior parietal cortices preceded perceptual switches, supporting a top-down processing model in which the prefrontal cortex can trigger perceptual alternations independently of hMT+ activity.
Role of neural oscillations in maintenance of human visual working memory representations
Pan, Y., & Wang, L. (2016, Chinese Science Bulletin 61 (15), 1650-1660)
📄 Paper
Conducted a literature review on the functional role of neural oscillations in working memory, focusing on both single-frequency bands (delta, theta, alpha, beta, gamma) and cross-frequency coupling. This review laid the foundation for my PhD project, in which I applied a delayed-match-to-sample paradigm to investigate the maintenance of visual working memory. I collected intracranial datasets from over 30 epilepsy patients for this project (the datasets will be shared publicly in the near future)
Software/package development
A tutorial for measuring attention in active vision using rapid invisible frequency tagging (RIFT)
Pan, Y. (*)
📄 Paper | 💻 Code | 💾 datasets
I am currently developing a Python and MATLAB toolbox for conducting Rapid Invisible Frequency Tagging (RIFT) in experiments with and without eye movements. The toolbox will provide functionalities for:
* Stimulus presentation with an ultra-refresh-rate ProPixx projector (1440 Hz) or high-refresh-rate gaming monitor (360 Hz):
– tagging images in a visual search task
– tagging text in a reading task
– tagging the visual and auditory streams in a video-watching task
* Data analysis pipeline:
– co-registration of eye movements and brain signals (EEG & MEG)
– extensive removal of oculomotor-related artefacts
– computation of RIFT responses at single-participant, group, and even single-trial levels
– parametric and non-parametric statistical comparisons between conditions
– source localisation analyses
* Results visualisation
As RIFT is gaining growing attention across diverse fields, this toolbox and the accompanying methods paper aim to provide a comprehensive resource that explains the underlying rationale, standardises parameter selection, and unifies analysis pipelines.
Automatic and precise localization and cortical labeling of subdural and depth intracranial electrodes
Qin, C., Tan, Z., Pan, Y., Li, Y., Wang, L., Ren, L., Zhou, W., & Wang, L. (2017, Frontiers in neuroinformatics, 11, 10)
📄 Paper | 💻 Code
Developed a toolbox to automatically and precisely extract, localise, and visualise intracranial electrodes in patients with intractable epilepsy using pre-operative MRI and post-operative CT. The toolbox further allows electrode labelling according to multiple brain atlases. It is among the first tools to standardise the localisation and labelling of intracranial EEG electrodes, thereby facilitating automation and improving consistency in intracranial EEG research.