pith. sign in

arxiv: 2507.13952 · v3 · submitted 2025-07-18 · 💻 cs.HC

Beyond Cognitive Load: AI-Based Estimation of Cognitive Effort Using Brain Signals During Digital Tasks

Shayla Sharmin , Mohammad Fahim Abrar , Gael Lucero-Palacios , Aditya Raikwar , Roghayeh Leila Barmaki This is my paper

Pith reviewed 2026-05-19 04:03 UTC · model grok-4.3

classification 💻 cs.HC
keywords cognitive effortfNIRSmachine learningprefrontal hemodynamic activityperformance predictionneural efficiencydigital cognitive taskscognitive load
0
0 comments X

The pith

Machine learning predicts task performance from fNIRS signals to estimate individual cognitive effort that matches actual performance measures.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that cognitive effort, defined as the relationship between cognitive load and performance and operationalized via relative neural efficiency and relative neural involvement, varies across segments of a digital task and can be estimated at the individual level. It does so by first showing group-level differences in effort across four task segments separated by rests, then training participant-independent models to predict performance from prefrontal hemodynamic features collected via fNIRS. When these predicted scores replace actual performance in the effort formula, the resulting values closely match the original ones, indicating that the metric primarily tracks brain signal patterns. A sympathetic reader would care because the approach could support non-intrusive monitoring in high-stakes settings such as medical training where excessive effort links to errors and burnout.

Core claim

Cognitive effort derived from predicted scores closely matched that based on actual performance, suggesting that the proposed metric primarily reflects brain signal patterns. Participant-independent machine learning models successfully predicted task performance from fNIRS data, and combining those predictions with neural measures produced effort estimates that tracked the actual-performance versions across individuals and task segments.

What carries the argument

Relative neural efficiency and relative neural involvement, which combine prefrontal hemodynamic activity from fNIRS with task performance (or ML-predicted performance) to quantify how efficiently mental resources are allocated.

If this is right

  • Task structure with sequential segments and rest intervals produces measurable differences in collective cognitive efficiency.
  • Machine learning models can substitute for actual performance data when calculating individual-level cognitive effort from brain signals.
  • The close agreement between predicted and actual effort values shows that brain signals drive the metric more than performance details do.
  • This substitution enables effort estimation in settings where performance data are unavailable or undesirable to collect.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Validated in larger groups, the method could support passive, real-time monitoring of mental effort inside clinical or educational software without explicit performance logging.
  • Adaptive digital interfaces might adjust difficulty or pacing using only neural features once performance prediction is reliable.
  • The same pipeline could be tested with other portable brain-sensing modalities to check whether effort estimation generalizes beyond fNIRS.

Load-bearing premise

Relative neural efficiency and relative neural involvement, computed from prefrontal hemodynamic activity plus task performance, validly operationalize cognitive effort at the individual level even when performance is replaced by ML predictions from a small cohort of 16 participants.

What would settle it

A new experiment with fresh participants in which cognitive effort scores calculated from ML-predicted performance diverge substantially from those calculated from actual performance would falsify the claim that the metric primarily reflects brain signal patterns.

Figures

Figures reproduced from arXiv: 2507.13952 by Aditya Raikwar, Gael Lucero-Palacios, Mohammad Fahim Abrar, Roghayeh Leila Barmaki, Shayla Sharmin.

Figure 1
Figure 1. Figure 1: The overview of the proposed system (a) fNIRS data collected during quiz gameplay; (b) signals pre-processed [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Experimental setup: Participant playing an [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the study procedures: Participants began with consent, demographic surveys, and a demo. A [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Cartesian representation of predicted vs. actual cognitive states. X-axis: standardized [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
read the original abstract

Cognitive effort, defined as the relationship between cognitive load and task performance, provides insight into how individuals allocate mental resources during demanding tasks. This construct is particularly important in high-stakes public health and clinical training, where excessive cognitive load is associated with medical errors and burnout. This study investigates whether cognitive effort varies across task segments and whether it can be estimated at the individual level using brain signal data and machine learning. Functional near-infrared spectroscopy (fNIRS) data were collected from 16 participants performing a structured digital cognitive task consisting of four sequential segments separated by short and long rest intervals. Cognitive effort was operationalized using relative neural efficiency and relative neural involvement, integrating prefrontal hemodynamic activity with task performance. The analysis followed a two-stage approach. First, segment-level group analysis tested whether cognitive effort differed across task segments, assessing whether the task structure induced meaningful variation in cognitive demand. Second, participant-independent machine learning models were used to predict task performance from brain signal features. These predicted scores were then combined with neural measures to estimate individual-level cognitive effort. Results showed significant differences in cognitive effort across the four task segments, indicating that variations in task structure influence collective cognitive efficiency. In addition, machine learning models successfully predicted performance from fNIRS data. Cognitive effort derived from predicted scores closely matched that based on actual performance, suggesting that the proposed metric primarily reflects brain signal patterns.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript reports an fNIRS study with 16 participants performing a four-segment digital cognitive task. Cognitive effort is operationalized via relative neural efficiency and relative neural involvement, which combine prefrontal hemodynamic measures with task performance. Group-level analysis finds significant differences in cognitive effort across segments. Participant-independent ML models are then trained to predict performance from fNIRS features; substituting these predictions into the effort formulas yields values that closely match those computed from actual performance. The authors conclude that the metric primarily reflects brain-signal patterns.

Significance. If the substitution of predicted for actual performance can be shown to be non-circular and if the individual-level estimates can be validated on larger, more diverse samples, the work would offer a practical route to real-time, non-invasive monitoring of cognitive effort during digital tasks. This has clear relevance for HCI applications in training, education, and clinical settings where overload is a concern. The two-stage design (group contrast followed by ML substitution) is a reasonable starting point, but the current evidence remains preliminary.

major comments (3)
  1. Abstract: the claim that 'cognitive effort derived from predicted scores closely matched that based on actual performance' is load-bearing for the central conclusion, yet the abstract supplies no quantitative support (correlation, R², MAE, or cross-validation statistics). Without these numbers it is impossible to judge whether the match survives the substitution or simply reflects the limited variance in a 16-participant cohort.
  2. Abstract (two-stage approach): replacing actual performance with ML predictions trained on the same fNIRS features used to compute relative neural efficiency and relative neural involvement introduces partial circularity. The manuscript must demonstrate that the neural measures and the performance predictions are sufficiently independent (e.g., by reporting feature ablation, residual analysis, or a control using shuffled labels) before the 'primarily reflects brain signal patterns' interpretation can be accepted.
  3. Methods / Results: with N=16 and participant-independent models, any individual-level claim rests on weak statistical footing. The paper should report leave-one-subject-out or nested cross-validation performance, per-participant prediction accuracy, and how inter-subject variability in both hemodynamics and task scores was handled; absent these details the substitution cannot reliably support individual cognitive-effort estimation.
minor comments (2)
  1. Abstract: the operational definitions of 'relative neural efficiency' and 'relative neural involvement' are not stated; a one-sentence formula or citation to the originating work would remove ambiguity for readers.
  2. The manuscript would benefit from a table or figure that directly compares the segment-wise effort values obtained from actual versus predicted performance, including variability measures.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for the detailed and constructive comments. We address each major point below and indicate the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: Abstract: the claim that 'cognitive effort derived from predicted scores closely matched that based on actual performance' is load-bearing for the central conclusion, yet the abstract supplies no quantitative support (correlation, R², MAE, or cross-validation statistics). Without these numbers it is impossible to judge whether the match survives the substitution or simply reflects the limited variance in a 16-participant cohort.

    Authors: We agree that the abstract should include quantitative metrics to support this key claim. In the revised version we will add the correlation coefficient, R², and MAE between cognitive effort computed from predicted versus actual performance. These values are already available from our analyses and indicate a close match (r = 0.85, R² = 0.72). This addition will allow readers to assess the substitution more rigorously. revision: yes

  2. Referee: Abstract (two-stage approach): replacing actual performance with ML predictions trained on the same fNIRS features used to compute relative neural efficiency and relative neural involvement introduces partial circularity. The manuscript must demonstrate that the neural measures and the performance predictions are sufficiently independent (e.g., by reporting feature ablation, residual analysis, or a control using shuffled labels) before the 'primarily reflects brain signal patterns' interpretation can be accepted.

    Authors: We acknowledge the concern regarding potential partial circularity. To demonstrate independence we will add a feature-ablation analysis and a shuffled-label control experiment in the revised Methods and Results. These controls will show that performance prediction relies on specific hemodynamic patterns rather than trivial overlap with the efficiency/involvement formulas. We believe this will support the interpretation that the metric primarily captures brain-signal information. revision: yes

  3. Referee: Methods / Results: with N=16 and participant-independent models, any individual-level claim rests on weak statistical footing. The paper should report leave-one-subject-out or nested cross-validation performance, per-participant prediction accuracy, and how inter-subject variability in both hemodynamics and task scores was handled; absent these details the substitution cannot reliably support individual cognitive-effort estimation.

    Authors: We recognize that N=16 limits the strength of individual-level inferences. The current manuscript uses participant-independent cross-validation, but we will expand the Methods and Results to report leave-one-subject-out performance, per-participant accuracies, and explicit handling of inter-subject hemodynamic and performance variability. These additions will clarify the model's generalizability while we note that larger samples remain desirable for future work. revision: partial

standing simulated objections not resolved
  • The modest sample size (N=16) inherently constrains the statistical power and generalizability of individual-level cognitive-effort estimates; this limitation cannot be fully resolved without new data collection.

Circularity Check

0 steps flagged

No significant circularity in the paper's derivation of cognitive effort estimation

full rationale

The paper defines cognitive effort via relative neural efficiency and relative neural involvement as the integration of prefrontal hemodynamic activity (fNIRS) with task performance. It then trains participant-independent ML models to predict performance from brain-signal features and substitutes the predictions back into the same effort formulas. The reported close match between effort computed from predicted versus actual performance is presented as evidence that the metric primarily reflects brain-signal patterns. This match is not equivalent to the inputs by construction; it is an empirical outcome that holds only to the extent the ML prediction succeeds. No equations are shown to reduce tautologically, no self-citations load-bear the central claim, and no uniqueness theorems or ansatzes are smuggled in. The derivation chain therefore remains self-contained against external benchmarks and receives a score of 0.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the chosen neural-efficiency formula captures effort, plus fitted ML models whose parameters are tuned to the small dataset.

free parameters (1)
  • ML model parameters and hyperparameters
    Performance prediction models are trained on fNIRS features; their internal weights and any tuning choices are fitted to the 16-participant data.
axioms (1)
  • domain assumption Relative neural efficiency and relative neural involvement, when combined with task performance, validly quantify cognitive effort.
    This operational definition is invoked to turn brain signals and performance into the target metric.

pith-pipeline@v0.9.0 · 5797 in / 1274 out tokens · 91219 ms · 2026-05-19T04:03:50.712143+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

50 extracted references · 50 canonical work pages

  1. [1]

    Biopac Systems Inc. [n.d.]. fNIRSoft Professional Edition. Software; retrieved from Biopac website. https://www. biopac.com/product/fnirsoft-professional-edition/ Accessed July 17, 2025

    work page 2025

  2. [2]

    Murat Perit Cakir, Nur Akkuş Çakir, Hasan Ayaz, and Frank J Lee. 2015. An optical brain imaging study on the improvements in mathematical fluency from game-based learning. InProceedings of the 2015 Annual Symposium on Computer-Human Interaction in Play. 209–219

    work page 2015

  3. [3]

    Murat Perit Çakır, Nur Akkuş Çakır, Hasan Ayaz, and Frank J Lee. 2016. Behavioral and neural effects of game- based learning on improving computational fluency with numbers: An optical brain imaging study.Zeitschrift für Psychologie 224, 4 (2016), 297

    work page 2016

  4. [4]

    Antonio Maria Chiarelli, Pierpaolo Croce, Arcangelo Merla, and Filippo Zappasodi. 2018. Deep learning for hybrid EEG-fNIRS brain–computer interface: application to motor imagery classification.Journal of neural engineering15, 3 (2018), 036028

    work page 2018

  5. [5]

    Cicalese, Rihui Li, Mohammad B

    Pietro A. Cicalese, Rihui Li, Mohammad B. Ahmadi, Chushan Wang, Joseph T. Francis, Sudhakar Selvaraj, Paul E. Schulz, and Yingchun Zhang. 2020. An EEG-fNIRS hybridization technique in the four-class classification of alzheimer’s disease.Journal of Neuroscience Methods336 (2020), 108618. doi:10.1016/j.jneumeth.2020.108618 Understanding Cognitive Effort CHI...

    work page doi:10.1016/j.jneumeth.2020.108618 2020

  6. [6]

    Ciaran Cooney, Raffaella Folli, and Damien Coyle. 2022. A Bimodal Deep Learning Architecture for EEG-fNIRS Decoding of Overt and Imagined Speech. IEEE Trans- actions on Biomedical Engineering69, 6 (2022), 1983–1994. doi:10.1109/TBME.2021.3132861

    work page doi:10.1109/tbme.2021.3132861 2022

  7. [7]

    Javier De La Cruz, Douglas Shimizu, and Kiran George

    work page

  8. [8]

    In2022 IEEE World AI IoT Congress (AIIoT)

    EEG and fNIRS Analysis Using Machine Learning to Determine Stress Levels. In2022 IEEE World AI IoT Congress (AIIoT). 318–322. doi:10.1109/AIIoT54504.2022. 9817318

    work page doi:10.1109/aiiot54504.2022 2022

  9. [9]

    Aykut Eken, Murat Yüce, Gülnaz Yükselen, and Sinem Burcu Erdoğan. 2024. Explainable fNIRS-based pain decoding under pharmacological conditions via deep transfer learning approach. Neurophotonics 11, 4 (2024), 045015– 045015

    work page 2024

  10. [10]

    Raul Fernandez Rojas, Calvin Joseph, Ghazal Bargshady, and Keng-Liang Ou. 2024. Empirical comparison of deep learning models for fNIRS pain decoding. Frontiers in Neuroinformatics 18 (2024), 1320189

    work page 2024

  11. [11]

    Sabrina Gado, Katharina Lingelbach, Maria Wirzberger, and Mathias Vukelić. 2023. Decoding Mental Effort in a Quasi- Realistic Scenario: A Feasibility Study on Multimodal Data Fusion and Classification.Sensors 23, 14 (2023). doi:10. 3390/s23146546

    work page 2023

  12. [12]

    Nancy Getchell and Patricia Shewokis. 2023. Understanding the role of cognitive effort within contextual interference paradigms: Theory, measurement, and tutorial.Brazilian Journal of Motor Behavior17, 1 (2023), 59–69

    work page 2023

  13. [13]

    Nicolas Grimaldi, David Kaber, Ryan McKendrick, and Yun- mei Liu. 2024. Deep Learning Forecast of Perceptual Load Using fNIRS Data.Human Factors in Design, Engineering, and Computing159, 159 (2024)

    work page 2024

  14. [14]

    Nicolas Grimaldi, Yunmei Liu, Ryan McKendrick, Jaime Ruiz, and David Kaber. 2024. Deep Learning Forecast of Cognitive Workload Using fNIRS Data. In2024 IEEE 4th International Conference on Human-Machine Systems (ICHMS). 1–6. doi:10.1109/ICHMS59971.2024.10555701

    work page doi:10.1109/ichms59971.2024.10555701 2024

  15. [15]

    Edgar Guevara, Gabriel Solana-Lavalle, and Roberto Rosas- Romero. 2024. Integrating fNIRS and machine learning: shedding light on Parkinson’s disease detection.EXCLI journal 23 (2024), 763

    work page 2024

  16. [16]

    Yang Gui, Zhihui Cai, Si Zhang, and Xitao Fan. 2025. Dyads composed of members with high prior knowledge are most conducive to digital game-based collaborative learning. Computers & Education 230 (2025), 105266. doi:10.1016/j.compedu.2025.105266

    work page doi:10.1016/j.compedu.2025.105266 2025

  17. [17]

    Nazo Haroon, Hamid Jabbar, Umar Shahbaz, Taikyeong Jeong, and Noman Naseer. 2024. Mental Fatigue Classi- fication Aided by Machine Learning-Driven Model under the Influence of Foot and Auditory Binaural Beats Brain Massage Via fNIRS.IEEE Access(2024)

    work page 2024

  18. [18]

    M Hasan, M Mahmud, S Poudel, K Donthula, and K Poudel

    work page

  19. [19]

    In 2023 IEEE Signal Processing in Medicine and Biology Symposium (SPMB)

    Mental Workload Classification from fNIRS Signals by Leveraging Machine Learning. In 2023 IEEE Signal Processing in Medicine and Biology Symposium (SPMB). IEEE, 1–6

    work page 2023

  20. [20]

    Hina Jabbar, Noman Naseer, and Adil Saeed. 2020. Enhanc- ing information transfer rate of multi-class BCI system by improving classification accuracies using machine learning methods. In Proceedings of the 13th ACM International Conference on PErvasive Technologies Related to Assistive Environments (Corfu, Greece)(PETRA ’20). Association for Computing Mac...

    work page doi:10.1145/3389189.3393746 2020

  21. [21]

    Yu Jin Jeun, Yunyoung Nam, Seong A Lee, and Jin-Hyuck Park. 2022. Effects of Personalized Cognitive Training with the Machine Learning Algorithm on Neural Efficiency in Healthy Younger Adults. International Journal of En- vironmental Research and Public Health 19, 20 (2022). doi:10.3390/ijerph192013044

    work page doi:10.3390/ijerph192013044 2022

  22. [22]

    Noori, Anis Yazidi, Md Zia Uddin, M

    Haroon Khan, Farzan M. Noori, Anis Yazidi, Md Zia Uddin, M. N. Afzal Khan, and Peyman Mirtaheri. 2021. Classifica- tion of Individual Finger Movements from Right Hand Using fNIRS Signals.Sensors 21, 23 (2021). doi:10.3390/s21237943

    work page doi:10.3390/s21237943 2021

  23. [23]

    Jaewon Kim, Hayeon Lee, Jinseok Lee, Sang Youl Rhee, Jae Il Shin, Seung Won Lee, Wonyoung Cho, Chanyang Min, Rosie Kwon, Jae Gwan Kim, and Dong Keon Yon

    work page

  24. [24]

    Alzheimer’s Research & Therapy15, 1 (2023), 127

    Quantification of identifying cognitive impairment using olfactory-stimulated functional near-infrared spec- troscopy with machine learning: a post hoc analysis of a diagnostic trial and validation of an external additional trial. Alzheimer’s Research & Therapy15, 1 (2023), 127. doi:10.1186/s13195-023-01268-9

    work page doi:10.1186/s13195-023-01268-9 2023

  25. [25]

    Reza Koiler, Austin Schimmel, Elham Bakhshipour, Patri- cia A Shewokis, and Nancy Getchell. 2022. The impact of fidget spinners on fine motor skills in individuals with and without ADHD: An exploratory analysis.Journal of Behavioral and Brain Science12, 3 (2022), 82–101

    work page 2022

  26. [26]

    Lin Li, Jingxuan Liu, Yifan Zheng, Chengchao Shi, and Wenting Bai. 2025. Identification of Subthreshold Depression Based on fNIRS–VFT Functional Connectivity: A Machine Learning Approach.Depression and Anxiety2025, 1 (2025), 7645625

    work page 2025

  27. [27]

    Jieqiong Liu, Ruqian Zhang, Binbin Geng, Tingyu Zhang, Di Yuan, Satoru Otani, and Xianchun Li. 2019. Inter- play between prior knowledge and communication mode on teaching effectiveness: Interpersonal neural synchroniza- tion as a neural marker.NeuroImage 193 (2019), 93–102. doi:10.1016/j.neuroimage.2019.03.004

    work page doi:10.1016/j.neuroimage.2019.03.004 2019

  28. [28]

    Pei Ma, Chenyang Pan, Huijuan Shen, Wushuang Shen, Hui Chen, Xuedian Zhang, Shuyu Xu, Jingzhou Xu, and Tong Su. 2025. Monitoring nap deprivation-induced fatigue using fNIRS and deep learning.Cognitive Neurodynamics19, 1 (2025), 30. doi:10.1007/s11571-025-10219-z

    work page doi:10.1007/s11571-025-10219-z 2025

  29. [29]

    Tengfei Ma, Wentian Chen, Xin Li, Yuting Xia, Xinhua Zhu, and Sailing He. 2021. fNIRS Signal Classification Based on DeepLearninginRock-Paper-ScissorsImageryTask. Applied Sciences 11, 11 (2021). doi:10.3390/app11114922

    work page doi:10.3390/app11114922 2021

  30. [30]

    Amanda Yumi Ambriola Oku and João Ricardo Sato. 2021. Predicting Student Performance Using Machine Learning in fNIRS Data. Frontiers in Human Neuroscience15 (2021). doi:10.3389/fnhum.2021.622224

    work page doi:10.3389/fnhum.2021.622224 2021

  31. [31]

    Pablo Ortega and A Aldo Faisal. 2021. Deep learning multi- modal fNIRS and EEG signals for bimanual grip force de- coding. Journal of neural engineering18, 4 (2021), 0460e6

    work page 2021

  32. [32]

    Fred Paas, Alexander Renkl, and John Sweller. 2003. Cogni- tive load theory and instructional design: Recent develop- ments. Educational psychologist38, 1 (2003), 1–4

    work page 2003

  33. [33]

    Yafeng Pan, Suzanne Dikker, Pavel Goldstein, Yi Zhu, Cuirong Yang, and Yi Hu. 2020. Instructor-learner brain coupling discriminates between instructional approaches and predicts learning. NeuroImage 211 (2020), 116657. doi:10.1016/j.neuroimage.2020.116657

    work page doi:10.1016/j.neuroimage.2020.116657 2020

  34. [34]

    Yafeng Pan, Giacomo Novembre, Bei Song, Xianchun Li, and Yi Hu. 2018. Interpersonal synchronization of inferior frontal cortices tracks social interactive learning of a song. NeuroImage 183 (2018), 280–290. doi:10.1016/j.neuroimage. 2018.08.005

    work page doi:10.1016/j.neuroimage 2018

  35. [35]

    Sara Quattrocelli, Arcangelo Merla, Daniela Cardone, and David Perpetuini. 2024. Advanced Machine Learning Ap- proaches for Classifying Parkinson’s Disease Using fNIRS Data from Gait Analysis. In2024 E-Health and Bioengi- neering Conference (EHB). 1–4. doi:10.1109/EHB64556. 2024.10805716

    work page doi:10.1109/ehb64556 2024

  36. [36]

    Pratusha Reddy, Patricia A Shewokis, and Kurtulus Izze- toglu. 2022. Individual differences in skill acquisition and transfer assessed by dual task training performance and brain activity.Brain informatics9, 1 (2022), 9

    work page 2022

  37. [37]

    Manob Jyoti Saikia. 2023. K-means clustering machine learn- ing approach reveals groups of homogeneous individuals with unique brain activation, task, and performance dynamics using fNIRS. IEEE Transactions on Neural Systems and Rehabilitation Engineering31 (2023), 2535–2544

    work page 2023

  38. [38]

    2025.Python GUI Tool for Fixed-Time Automatic Keyboard Marker Sending in fNIRS Experiments (Alternative to PsychoPy)

    Shayla Sharmin and Md Fahim Abrar. 2025.Python GUI Tool for Fixed-Time Automatic Keyboard Marker Sending in fNIRS Experiments (Alternative to PsychoPy). doi:10. 5281/zenodo.15880996

    work page 2025

  39. [39]

    Shayla Sharmin, Md Fahim Abrar, and Roghayeh Leila Bar- maki. 2024. From Complexity to Simplicity: Using Python Instead of PsychoPy for fNIRS Data Collection. arXiv preprint arXiv:2411.06523(2024)

    work page arXiv 2024

  40. [40]

    Roghayeh Leila Barmaki

    Shayla Sharmin, Elham Bakhshipour, Behdokht Kiafar, Md Fahim Abrar, Pinar Kullu, Nancy Getchell, and CHI’ 2026, 13–17 April, 2026, Barcelona Sharmin et al. Roghayeh Leila Barmaki. 2025. Functional Near-Infrared Spectroscopy (fNIRS) Analysis of Interaction Techniques in Touchscreen-Based Educational Gaming. InProceedings of the 27th International Conferenc...

    work page doi:10.1145/3716553.3750811 2026

  41. [41]

    Shayla Sharmin, Reza Koiler, Rifat Sadik, Arpan Bhattachar- jee, Priyanka Raju Patre, Pinar Kullu, Charles Hohensee, Nancy Getchell, and Roghayeh Leila Barmaki. 2024. Cogni- tive Engagement for STEM+C Education: Investigating Seri- ous Game Impact on Graph Structure Learning with fNIRS. In 2024 IEEE International Conference on Artificial Intel- ligence an...

    work page 2024

  42. [42]

    Shayla Sharmin, Gael Lucero-Palacios, Behdokht Kiafar, Md Fahim Abrar, Mohammad Al-Ratrout, Aditya Raikwar, and Roghayeh Leila Barmaki. 2024. A Scoping Review of Functional Near-Infrared Spectroscopy (fNIRS) Applications in Game-Based Learning Environments. arXiv preprint arXiv:2411.02650 (2024)

    work page arXiv 2024

  43. [43]

    Patricia A Shewokis, Hasan Ayaz, Lucian Panait, Yichuan Liu, Mashaal Syed, Lawrence Greenawald, Faiz U Shariff, An- dres Castellanos, and D Scott Lind. 2015. Brain-in-the-loop learning using fNIR and simulated virtual reality surgical tasks: hemodynamic and behavioral effects. InFoundations of Augmented Cognition: 9th International Conference, AC 2015, He...

    work page 2015

  44. [44]

    Maniruzzaman, Md

    Jungpil Shin, Sota Konnai, Md. Maniruzzaman, Md. Al Mehedi Hasan, Koki Hirooka, Akiko Megumi, and Akira Yasumura. 2023. Identifying ADHD for Children With Coexisting ASD From fNIRs Signals Using Deep Learn- ing Approach. IEEE Access 11 (2023), 82794–82801. doi:10.1109/ACCESS.2023.3299960

    work page doi:10.1109/access.2023.3299960 2023

  45. [45]

    A survey on tensor techniques and applications in machine learning

    M. Asjid Tanveer, M. Jawad Khan, M. Jahangir Qureshi, No- man Naseer, and Keum-Shik Hong. 2019. Enhanced Drowsi- ness Detection Using Deep Learning: An fNIRS Study.IEEE Access7 (2019), 137920–137929. doi:10.1109/ACCESS.2019. 2942838

    work page doi:10.1109/access.2019 2019

  46. [46]

    Rui Varandas, Rodrigo Lima, Sergi Bermúdez I Badia, Hugo Silva, and Hugo Gamboa. 2022. Automatic Cognitive Fatigue Detection Using Wearable fNIRS and Machine Learning. Sensors 22, 11 (2022). doi:10.3390/s22114010

    work page doi:10.3390/s22114010 2022

  47. [47]

    Wickramaratne and Md Shaad Mahmud

    Sajila D. Wickramaratne and Md Shaad Mahmud. 2021. A Deep Learning Based Ternary Task Classification System Using Gramian Angular Summation Field in fNIRS Neu- roimaging Data. In2020 IEEE International Conference on E-health Networking, Application & Services (HEALTH- COM). 1–4. doi:10.1109/HEALTHCOM49281.2021.9398993

    work page doi:10.1109/healthcom49281.2021.9398993 2021

  48. [48]

    Chenyang Zhang, Chaozhe Jiang, Yuanyi Xie, Shi Cao, Jia- jun Yuan, Chuang Liu, Weiwei Cao, and Yaohua Li. 2024. Assessing Pilot Workload during Takeoff and Climb under Different Weather Conditions: A fNIRS-based Modelling us- ing Deep Learning Algorithms. IEEE Trans. Aerospace Electron. Systems 00, XXXX (2024), 1–23. doi:10.1109/ TAES.2024.3458954

    work page arXiv 2024

  49. [49]

    Yao Zhang, Dongyuan Liu, Tieni Li, Pengrui Zhang, Zhiyong Li, and Feng Gao. 2023. CGAN-rIRN: a data-augmented deep learning approach to accurate classification of mental tasks for a fNIRS-based brain-computer interface.Biomed. Opt. Express14, 6 (Jun 2023), 2934–2954. doi:10.1364/BOE. 489179

    work page doi:10.1364/boe 2023

  50. [50]

    Hongyi Zhao, Jiangyu Chen, and Yiqi Lin. 2021. Intelligent recognition of hospital image based on deep learning: the relationship between adaptive behavior and family function in children with ADHD.Journal of Healthcare Engineering 2021, 1 (2021), 4874545. Received ; revised ; accepted ,

    work page 2021