pith. sign in

arxiv: 2606.20598 · v1 · pith:2GL3KY2Nnew · submitted 2026-05-19 · 💻 cs.HC · cs.AI

Using Biometrics to Understand AI-Assisted Coding Performance and its Perception

Paolo Burelli , Fabio Calefato , Daniela Grassi , Mihaela Yurieva Hristova , Nicole Novielli , Alberto Antonio , Romano , Paolo Tell This is my paper

Pith reviewed 2026-06-30 18:37 UTC · model grok-4.3

classification 💻 cs.HC cs.AI
keywords AI-assisted programmingbiometricscognitive engagementEEGeye-trackingelectrodermal activityNASA-TLXsoftware development
0
0 comments X

The pith

Biometric signals show AI-assisted coding reduces cognitive engagement compared to solo coding.

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

The study measures brain and body signals while developers write code with and without AI assistance in a crossover design across two sites. Developers showed a lower EEG theta-to-alpha ratio in the first task and higher gaze blink rate in the second when using AI, patterns the authors link to offloading generative work to the model. These differences appeared the same for undergraduates and graduates, while performance correlations with skin conductance and certain workload ratings held only in the non-AI condition. A reader would care because the results challenge the view that AI tools merely speed up existing coding habits and instead treat the activity as cognitively distinct.

Core claim

In the within-subjects comparison, the EEG θ/α ratio was lower during the first task and gaze blink rate higher during the second under AI assistance, both taken as signs of reduced cognitive engagement from offloading effort to the model; electrodermal activity correlated with performance only without AI, and the pattern did not vary by student experience level.

What carries the argument

The within-subjects crossover comparison of neurophysiological measures (EEG θ/α ratio, gaze blink rate, electrodermal activity) and NASA-TLX workload scores between AI-assisted and non-assisted coding conditions.

If this is right

  • AI code assistants should be designed around a shift in cognitive processes rather than an assumption of simple acceleration.
  • Biometric monitoring could be integrated into AI-augmented development environments.
  • Developer experience level does not change the observed physiological differences between conditions.
  • Performance links to objective workload appear only in the absence of AI assistance.

Where Pith is reading between the lines

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

  • Tool builders might explore interfaces that detect offloading moments and adjust suggestions accordingly.
  • Training programs could address how to maintain engagement when generative work is handed to models.
  • Future studies could test whether these biometric shifts predict long-term skill retention or error rates.

Load-bearing premise

The selected physiological signals accurately reflect changes in cognitive engagement and load while writing code.

What would settle it

A replication that finds identical EEG θ/α ratios and blink rates across AI and non-AI conditions, or finds that skin conductance still predicts performance under AI assistance.

Figures

Figures reproduced from arXiv: 2606.20598 by Alberto Antonio, Daniela Grassi, Fabio Calefato, Mihaela Yurieva Hristova, Nicole Novielli, Paolo Burelli, Paolo Tell, Romano.

Figure 1
Figure 1. Figure 1: Experimental protocol phases. 4.3 Experimental Protocol At each site, the experiment was conducted in a controlled laboratory envi￾ronment to minimize external distractions and ensure consistent measurement conditions. The overall timeline required about 90 minutes to complete (see [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Development environment screenshots for AI-assisted vs. non-assisted [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Experimental setup for participants in both Uniba and ITU sites. [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Project execution timeline showing the parallel activities at the two [PITH_FULL_IMAGE:figures/full_fig_p022_4.png] view at source ↗
read the original abstract

AI-based code assistants are transforming software development, yet we lack empirical evidence on how they affect developers' cognitive processes. We present a multisite study investigating the neurophysiological correlates of AI-assisted programming through a within-subjects crossover design. We recruited participants at two universities (Bari, Italy, and Copenhagen, Denmark) and collected electroencephalography, eye-tracking, electrodermal activity, and heart rate variability data alongside a rubric-based performance score and self-reported workload across six dimensions using the NASA Task Load Index (NASA-TLX). We tested four hypotheses addressing physiological differences between AI-assisted and non-assisted conditions, the moderating role of developer experience, the association between physiology and performance, and the alignment between subjective perceptions and objective measures. Under AI assistance, the EEG $\theta/\alpha$ ratio was lower during the first task and the gaze blink rate was higher during the second, both consistent with reduced cognitive engagement when developers offload generative effort to the model. This pattern did not differ between undergraduate and graduate students. Electrodermal activity correlated with performance under the non-AI condition but not under AI. Among the six NASA-TLX dimensions of self-reported workload, only Physical demand was associated with performance under the non-AI condition but not under AI. These findings suggest that AI-assisted programming is not a faster version of solo coding but a cognitively distinct activity, with implications for the design of AI assistants and for biometric monitoring in AI-augmented development.

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

2 major / 2 minor

Summary. The manuscript reports a multisite within-subjects crossover study recruiting participants at two universities and collecting EEG, eye-tracking, EDA, HRV, rubric-based performance scores, and NASA-TLX workload ratings during AI-assisted versus non-assisted coding tasks. It tests four hypotheses and reports lower EEG θ/α ratio in the first task and higher gaze blink rate in the second under AI assistance (interpreted as reduced cognitive engagement when offloading to the model), no moderation by undergraduate/graduate experience level, EDA-performance correlation only in the non-AI condition, and Physical demand-performance association only in the non-AI condition. The authors conclude that AI-assisted programming constitutes a cognitively distinct activity rather than a faster version of solo coding.

Significance. If the chosen biometric signals validly index cognitive engagement and load during code writing, the work supplies objective evidence that AI assistance alters developers' cognitive processes in ways not reducible to speed gains, with direct implications for AI assistant design and biometric monitoring in software engineering. The within-subjects design, multi-site recruitment, and joint collection of physiological, performance, and subjective measures are methodological strengths that would support generalizability if the signal-to-construct mappings are substantiated.

major comments (2)
  1. [Abstract and Results] Abstract and Results (physiological findings): The central claim that lower θ/α during the first task and higher blink rate during the second indicate 'reduced cognitive engagement when developers offload generative effort to the model' treats these signals as direct, domain-appropriate indices without reported task-specific calibration, validation against coding performance, or controls for confounds such as altered visual scanning of AI suggestions versus code, motor interface demands, or non-engagement arousal.
  2. [Methods] Methods (hypotheses and design): The within-subjects crossover and experience-level moderation test do not address whether EEG θ/α ratio, blink rate, or EDA track the intended cognitive constructs in code-writing tasks; no calibration data, baseline comparisons, or validation against established coding-load measures are described, leaving the dissociation between AI and non-AI conditions dependent on untested mappings.
minor comments (2)
  1. [Abstract] Abstract omits sample sizes, exclusion criteria, statistical tests, effect sizes, and exact task ordering, which are required to evaluate the reported differences and the claim of no experience-level moderation.
  2. Notation for the four hypotheses and the six NASA-TLX dimensions should be explicitly numbered or labeled when first introduced to improve traceability to the results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major point below with clarifications drawn from the study design and literature, and indicate revisions where appropriate to improve transparency.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results (physiological findings): The central claim that lower θ/α during the first task and higher blink rate during the second indicate 'reduced cognitive engagement when developers offload generative effort to the model' treats these signals as direct, domain-appropriate indices without reported task-specific calibration, validation against coding performance, or controls for confounds such as altered visual scanning of AI suggestions versus code, motor interface demands, or non-engagement arousal.

    Authors: We agree that the manuscript does not report new task-specific calibration or direct validation of θ/α and blink rate against coding performance within this dataset. These signals were selected based on prior literature establishing their sensitivity to cognitive engagement and workload in visual-motor tasks. The within-subjects crossover design controls for individual differences, and the multi-site recruitment supports generalizability of the observed condition differences. To strengthen the paper, we will add an explicit limitations subsection in the Discussion that acknowledges the absence of domain-specific calibration, discusses potential confounds such as visual scanning of AI output, and notes that interpretations rest on established mappings rather than new validation data. revision: partial

  2. Referee: [Methods] Methods (hypotheses and design): The within-subjects crossover and experience-level moderation test do not address whether EEG θ/α ratio, blink rate, or EDA track the intended cognitive constructs in code-writing tasks; no calibration data, baseline comparisons, or validation against established coding-load measures are described, leaving the dissociation between AI and non-AI conditions dependent on untested mappings.

    Authors: The four hypotheses were grounded in established neurophysiological literature linking θ/α ratio to cognitive engagement, blink rate to workload, and EDA to arousal, rather than being derived from new calibration within this study. No baseline comparisons or coding-specific validation data were collected. We will revise the Methods and Discussion sections to include additional citations to validation studies in related domains and to state the interpretive assumptions more explicitly, thereby clarifying the evidential basis for the reported dissociations without claiming new construct validation. revision: partial

Circularity Check

0 steps flagged

Empirical biometric study with no definitional or fitted reductions

full rationale

The paper reports a within-subjects crossover experiment collecting EEG, eye-tracking, EDA, HRV, performance scores, and NASA-TLX ratings to compare AI-assisted vs. solo coding. No equations, parameter fits, or predictions are defined in terms of the target outcomes; differences are measured directly against external signals and rubric scores. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. The central claim rests on observed physiological and performance differences rather than any internal redefinition or renaming of inputs. This is a standard empirical design whose validity can be assessed against external benchmarks without circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the selected biometric channels index cognitive engagement during coding; no free parameters or invented entities are introduced in the abstract.

axioms (1)
  • domain assumption EEG θ/α ratio, gaze blink rate, and electrodermal activity are valid proxies for cognitive engagement and workload in programming tasks.
    Invoked to interpret lower θ/α and higher blink rate as reduced engagement under AI assistance.

pith-pipeline@v0.9.1-grok · 5815 in / 1139 out tokens · 23574 ms · 2026-06-30T18:37:01.398084+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

154 extracted references · 54 canonical work pages · 2 internal anchors

  1. [1]

    2011 , publisher=

    Thinking, Fast and Slow , author=. 2011 , publisher=

    2011

  2. [2]

    1995 , publisher=

    Cognition in the Wild , author=. 1995 , publisher=

    1995

  3. [5]

    2019 , issn =

    Heart Rate and Heart Rate Variability Correlate with Clinical Reasoning Performance and Self-Reported Measures of Cognitive Load , journal =. 2019 , issn =. doi:10.1038/s41598-019-50280-3 , url =

    work page doi:10.1038/s41598-019-50280-3 2019

  4. [6]

    Biomedical Signal Processing and Control , volume=

    Acute mental stress assessment via short term HRV analysis in healthy adults: A systematic review with meta-analysis , author=. Biomedical Signal Processing and Control , volume=. 2015 , publisher=

    2015

  5. [7]

    Proceedings of the 46th IEEE/ACM international conference on software engineering , pages=

    A large-scale survey on the usability of ai programming assistants: Successes and challenges , author=. Proceedings of the 46th IEEE/ACM international conference on software engineering , pages=

  6. [8]

    Proceedings of the 16th international conference on Intelligent user interfaces , pages=

    Eye activity as a measure of human mental effort in HCI , author=. Proceedings of the 16th international conference on Intelligent user interfaces , pages=

  7. [9]

    2012 , publisher=

    Electrodermal activity , author=. 2012 , publisher=

    2012

  8. [10]

    Handbook of psychophysiology , volume=

    The electrodermal system , author=. Handbook of psychophysiology , volume=

  9. [11]

    Applied ergonomics , volume=

    Measuring mental workload using physiological measures: A systematic review , author=. Applied ergonomics , volume=. 2019 , publisher=

    2019

  10. [12]

    Psychiatry investigation , volume=

    Stress and heart rate variability: a meta-analysis and review of the literature , author=. Psychiatry investigation , volume=

  11. [15]

    Culemann, Wolf and Hooge, Ignace T. C. and Niehorster, Diederick C. and Heine, Angela and Nystr. Blink parameters are confounded by vertical eye orientation in video-based eye tracking: Comparing pupil- and eyelid-based methods , journal=. 2026 , month=. doi:10.3758/s13428-026-02984-4 , url=

    work page doi:10.3758/s13428-026-02984-4 2026

  12. [17]

    Information , VOLUME =

    Al-gawwam, Sarmad and Benaissa, Mohammed , TITLE =. Information , VOLUME =. 2018 , NUMBER =

    2018

  13. [18]

    Pupil dilation as an indicator of cognitive workload in human-computer interaction , journal =

    Pomplun, Marc and Sunkara, Sindhura , year =. Pupil dilation as an indicator of cognitive workload in human-computer interaction , journal =

  14. [19]

    Proceedings of the 16th International Conference on Intelligent User Interfaces , pages =

    Chen, Siyuan and Epps, Julien and Ruiz, Natalie and Chen, Fang , title =. Proceedings of the 16th International Conference on Intelligent User Interfaces , pages =. 2011 , isbn =. doi:10.1145/1943403.1943454 , abstract =

    work page doi:10.1145/1943403.1943454 2011

  15. [20]

    2022 , issn =

    Parietal Alpha Oscillations: Cognitive Load and Mental Toughness , journal =. 2022 , issn =. doi:https://doi.org/10.3390/brainsci12091135 , url =

    work page doi:10.3390/brainsci12091135 2022

  16. [22]

    Frontiers in Computer Science , volume=

    Deriving a Cortisol-Related Stress Indicator From Wearable Skin Conductance Measurements: Quantitative Model & Experimental Validation , author=. Frontiers in Computer Science , volume=. 2020 , publisher=

    2020

  17. [24]

    cvxEDA: A Convex Optimization Approach to Electrodermal Activity Processing , year=

    Greco, Alberto and Valenza, Gaetano and Lanata, Antonio and Scilingo, Enzo Pasquale and Citi, Luca , journal=. cvxEDA: A Convex Optimization Approach to Electrodermal Activity Processing , year=

  18. [27]

    2023 IEEE Conference on Games (CoG) , pages=

    Investigating Perceived and Mechanical Challenge in Games Through Cognitive Activity , author=. 2023 IEEE Conference on Games (CoG) , pages=. 2023 , organization=

    2023

  19. [30]

    2023 , eprint=

    The Impact of AI on Developer Productivity: Evidence from GitHub Copilot , author=. 2023 , eprint=

    2023

  20. [31]

    Hofmeister and Christian K

    Janet Siegmund and Norman Peitek and Chris Parnin and Sven Apel and Johannes C. Hofmeister and Christian K. Measuring neural efficiency of program comprehension , booktitle =. 2017 , url =

    2017

  21. [32]

    Program Comprehension and Code Complexity Metrics: An fMRI Study , booktitle =

    Norman Peitek and Sven Apel and Chris Parnin and Andr. Program Comprehension and Code Complexity Metrics: An fMRI Study , booktitle =. 2021 , url =

    2021

  22. [33]

    Understanding understanding source code with functional magnetic resonance imaging , booktitle =

    Janet Siegmund and Christian K. Understanding understanding source code with functional magnetic resonance imaging , booktitle =. 2014 , url =

    2014

  23. [35]

    2020 , booktitle =

    Girardi, Daniela and Novielli, Nicole and Fucci, Davide and Lanubile, Filippo , title =. 2020 , booktitle =

    2020

  24. [36]

    Predicting Bugs by Monitoring Developers During Task Execution , year=

    Laudato, Gennaro and Scalabrino, Simone and Novielli, Nicole and Lanubile, Filippo and Oliveto, Rocco , booktitle=. Predicting Bugs by Monitoring Developers During Task Execution , year=

  25. [37]

    1994 , publisher=

    Cognitive load theory , author=. 1994 , publisher=

    1994

  26. [39]

    Performance and Metacognition Disconnect when Reasoning in Human-AI Interaction , author=. Proc. of the 2024 CHI Conf. on Human Factors in Computing Systems , year=

    2024

  27. [40]

    Decoding the representation of code in the brain: an fMRI study of code review and expertise , booktitle =

    Benjamin Floyd and Tyler Santander and Westley Weimer , editor =. Decoding the representation of code in the brain: an fMRI study of code review and expertise , booktitle =. 2017 , url =

    2017

  28. [41]

    Adesope , editor =

    Sarah Fakhoury and Yuzhan Ma and Venera Arnaoudova and Olusola O. Adesope , editor =. The effect of poor source code lexicon and readability on developers' cognitive load , booktitle =. 2018 , url =

    2018

  29. [42]

    Quantifying programmers' mental workload during program comprehension based on cerebral blood flow measurement: a controlled experiment , booktitle =

    Takao Nakagawa and Yasutaka Kamei and Hidetake Uwano and Akito Monden and Ken. Quantifying programmers' mental workload during program comprehension based on cerebral blood flow measurement: a controlled experiment , booktitle =. 2014 , url =

    2014

  30. [43]

    Neurological divide: an fMRI study of prose and code writing , booktitle =

    Ryan Krueger and Yu Huang and Xinyu Liu and Tyler Santander and Westley Weimer and Kevin Leach , editor =. Neurological divide: an fMRI study of prose and code writing , booktitle =. 2020 , url =

    2020

  31. [44]

    Yoshiharu Ikutani and Hidetake Uwano , title =. Int. J. Networked Distributed Comput. , volume =. 2014 , url =

    2014

  32. [46]

    Distilling neural representations of data structure manipulation using fMRI and fNIRS , booktitle =

    Yu Huang and Xinyu Liu and Ryan Krueger and Tyler Santander and Xiaosu Hu and Kevin Leach and Westley Weimer , editor =. Distilling neural representations of data structure manipulation using fMRI and fNIRS , booktitle =. 2019 , url =

    2019

  33. [47]

    eLife , issn =

    Comprehension of computer code relies primarily on domain-general executive brain regions , author =. eLife , issn =. doi:10.7554/eLife.58906 , keywords =

    work page doi:10.7554/elife.58906

  34. [48]

    Connecting the dots: rethinking the relationship between code and prose writing with functional connectivity , booktitle =

    Zachary Karas and Andrew Jahn and Westley Weimer and Yu Huang , editor =. Connecting the dots: rethinking the relationship between code and prose writing with functional connectivity , booktitle =. 2021 , url =

    2021

  35. [49]

    Ma, Qianou Christina and Wu, Sherry Tongshuang and Koedinger, Ken , year =. Is. Proc. of the Workshop on Empowering Education with

  36. [50]

    Andrei, Oana and Sojtory, Zoltan , editor =. Proc. of the 2024 Conf. on United Kingdom & Ireland Computing Education Research,. 2024 , file =. doi:10.1145/3689535.3689546 , pages =

    work page doi:10.1145/3689535.3689546 2024

  37. [52]

    2023 , eprinttype =

    Chen, Eason and Huang, Ray and Liang, Justa and Chen, Damien and Hung, Pierce , urldate =. 2023 , eprinttype =. doi:10.48550/arXiv.2310.13896 , shorttitle =. 2310.13896 [cs] , keywords =

    work page doi:10.48550/arxiv.2310.13896 2023

  38. [53]

    Pair programming with. Comput. Stat. , author =. 2024 , file =. doi:10.1007/S00180-023-01437-2 , pages =

    work page doi:10.1007/s00180-023-01437-2 2024

  39. [54]

    Classification of human-human and human-

    Takai, Yuma and Washizaki, Hironori and Yoshioka, Nobukazu and Fukazawa, Yoshiaki , year =. Classification of human-human and human-. 30th Conf. on Pattern Languages of Programs (

  40. [55]

    Pair Programming with

    Friedmann, David Vadim , year =. Pair Programming with

  41. [56]

    Neuroscience & Biobehavioral Reviews , volume=

    Measuring neurophysiological signals in aircraft pilots and car drivers for the assessment of mental workload, fatigue and drowsiness , author=. Neuroscience & Biobehavioral Reviews , volume=. 2014 , publisher=

    2014

  42. [57]

    2021 , month = apr, url =

    Research Proves Your Brain Needs Breaks:. 2021 , month = apr, url =

    2021

  43. [58]

    Trends in cognitive sciences , volume=

    Frontal theta as a mechanism for cognitive control , author=. Trends in cognitive sciences , volume=. 2014 , publisher=

    2014

  44. [59]

    Generative. Proc. of the 58th Hawaii Int'l Conf. on System Sciences , author =. 2025 , langid =

    2025

  45. [60]

    Impact of generative

    Ganeshan, Nivethika and Kobbenes, Simon , year =. Impact of generative

  46. [61]

    Thomas Fritz and Andrew Begel and Sebastian C. M. Using psycho-physiological measures to assess task difficulty in software development , booktitle =. 2014 , url =

    2014

  47. [62]

    Sebastian C. M. Stuck and Frustrated or in Flow and Happy: Sensing Developers' Emotions and Progress , booktitle =. 2015 , url =

    2015

  48. [63]

    A replication study on code comprehension and expertise using lightweight biometric sensors , booktitle =

    Davide Fucci and Daniela Girardi and Nicole Novielli and Luigi Quaranta and Filippo Lanubile , editor =. A replication study on code comprehension and expertise using lightweight biometric sensors , booktitle =. 2019 , url =

    2019

  49. [64]

    Software design cognitive aspects , series =

    Fran. Software design cognitive aspects , series =. 2001 , url =

    2001

  50. [65]

    Cognitive psychology , volume=

    Stimulus structures and mental representations in expert comprehension of computer programs , author=. Cognitive psychology , volume=. 1987 , publisher=

    1987

  51. [66]

    Are Students Representatives of Professionals in Software Engineering Experiments? , booktitle =

    Iflaah Salman and Ayse Tosun Misirli and Natalia Juristo Juzgado , editor =. Are Students Representatives of Professionals in Software Engineering Experiments? , booktitle =. 2015 , url =

    2015

  52. [67]

    2016 , url =

    Sira Vegas and Cecilia Apa and Natalia Juristo Juzgado , title =. 2016 , url =

    2016

  53. [68]

    Journal of Behavioral and Experimental Finance , volume=

    oTree—An open-source platform for laboratory, online, and field experiments , author=. Journal of Behavioral and Experimental Finance , volume=. 2016 , publisher=

    2016

  54. [70]

    James and Nadia Polikarpova , title =

    Shraddha Barke and Michael B. James and Nadia Polikarpova , title =. Proc. of the ACM on Programming Languages , volume =. 2023 , doi =

    2023

  55. [71]

    Hussein Mozannar and Gagan Bansal and Adam Fourney and Eric Horvitz , title =. Proc. of the 2024 CHI Conf. on Human Factors in Computing Systems , pages =. 2024 , doi =

    2024

  56. [72]

    Glassman , title =

    Priyan Vaithilingam and Tianyi Zhang and Elena L. Glassman , title =. Proc. of the 2022 CHI Conf. on Human Factors in Computing Systems Extended Abstracts , pages =. 2022 , doi =

    2022

  57. [73]

    2024 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC) , pages =

    Ningzhi Tang and Meng Chen and Zheng Ning and Aakash Bansal and Yu Huang and Collin McMillan and Toby Jia-Jun Li , title =. 2024 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC) , pages =. 2024 , doi =

    2024

  58. [74]

    Becker, N

    Joel Becker and Nate Rush and Beth Barnes and David Rein , title =. arXiv preprint arXiv:2507.09089 , year =

    work page arXiv

  59. [76]

    Lev Tankelevitch and Viktor Kewenig and Auste Simkute and Ava Elizabeth Scott and Advait Sarkar and Abigail Sellen and Sean Rintel , title =. Proc. of the 2024 CHI Conf. on Human Factors in Computing Systems , pages =. 2024 , doi =

    2024

  60. [77]

    Hao-Ping Lee and Advait Sarkar and Lev Tankelevitch and Ian Drosos and Sean Rintel and Richard Banks and Nicholas Wilson , title =. Proc. of the 2025 CHI Conf. on Human Factors in Computing Systems , pages =. 2025 , doi =

    2025

  61. [78]

    Meyerhoff , title =

    Sandra Grinschgl and Frank Papenmeier and Hauke S. Meyerhoff , title =. Quarterly Journal of Experimental Psychology , volume =. 2021 , doi =

    2021

  62. [79]

    2026 , month =

    Margaret-Anne Storey , title =. 2026 , month =

    2026

  63. [80]

    Keeping Humans in the Driver's Seat: A Conceptual Framework for. Proc. of the 34th ACM Symposium on the Foundations of Software Engineering (FSE '26), Companion , year =

  64. [81]

    How AI impacts skill formation,

    Judy Hanwen Shen and Alex Tamkin , title =. arXiv preprint arXiv:2601.20245 , year =

    work page arXiv

  65. [82]

    Sreecharan Sankaranarayanan , title =. Proc. of the 13th ACM Conf. on Learning at Scale (L@S '26) , year =

  66. [83]

    arXiv preprint arXiv:2602.10540 , year =

    Arty Starr and Margaret-Anne Storey , title =. arXiv preprint arXiv:2602.10540 , year =

    work page arXiv

  67. [84]

    Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task

    Nataliya Kosmyna and Eugene Hauptmann and Ye Tong Yuan and Jessica Situ and Xian-Hao Liao and Ashly Vivian Beresnitzky and Iris Braunstein and Pattie Maes , title =. arXiv preprint arXiv:2506.08872 , year =

    work page internal anchor Pith review Pith/arXiv arXiv

  68. [85]

    LaToza and Brittany Johnson , title =

    Ebtesam Al Haque and Chris Brown and Thomas D. LaToza and Brittany Johnson , title =. 2025 , eprint =

    2025

  69. [86]

    arXiv preprint arXiv:2511.08177 , year =

    Yasmine Elfares and G\". arXiv preprint arXiv:2511.08177 , year =

    work page arXiv

  70. [87]

    Archives of general psychiatry , volume=

    Move over anova: progress in analyzing repeated-measures data andits reflection in papers published in the archives of general psychiatry , author=. Archives of general psychiatry , volume=. 2004 , publisher=

    2004

  71. [88]

    Advances in Methods and Practices in Psychological Science , volume=

    An introduction to linear mixed-effects modeling in R , author=. Advances in Methods and Practices in Psychological Science , volume=. 2021 , publisher=

    2021

  72. [89]

    2026 , note =

    Mahmoud, May and Asare, Eric and Shahid, Nisa and Sakr, Nourhan and Nadi, Sarah , title =. 2026 , note =

    2026

  73. [90]

    Empirical Software Engineering , volume=

    A practical guide on conducting eye tracking studies in software engineering , author=. Empirical Software Engineering , volume=. 2020 , publisher=

    2020

  74. [91]

    2016 IEEE Conference on Norbert Wiener in the 21st Century (21CW) , pages=

    Blink rate variability during resting and reading sessions , author=. 2016 IEEE Conference on Norbert Wiener in the 21st Century (21CW) , pages=. 2016 , organization=

    2016

  75. [92]

    Computers in Human Behavior , volume=

    Cognitive demand, digital screens and blink rate , author=. Computers in Human Behavior , volume=. 2015 , publisher=

    2015

  76. [93]

    International Journal of Human-Computer Studies , volume=

    Expertise-dependent visual attention strategies develop over time during debugging with multiple code representations , author=. International Journal of Human-Computer Studies , volume=. 2012 , publisher=

    2012

  77. [94]

    Information 9(4), doi:10.3390/info9040093, ://www.mdpi.com/2078-2489/9/4/93

    Al-gawwam S, Benaissa M (2018) Robust eye blink detection based on eye landmarks and savitzky–golay filtering. Information 9(4), doi:10.3390/info9040093, ://www.mdpi.com/2078-2489/9/4/93

    work page doi:10.3390/info9040093 2018

  78. [95]

    Sensors 19(5):1121, doi:10.3390/s19051121, ://doi.org/10.3390/s19051121, pMID: 30841622

    Alsaeedi N, Wloka D (2019) Real-time eyeblink detector and eye state classifier for virtual reality ( VR ) headsets ( Head-Mounted Displays, HMDs ). Sensors 19(5):1121, doi:10.3390/s19051121, ://doi.org/10.3390/s19051121, pMID: 30841622

    work page doi:10.3390/s19051121 2019

  79. [96]

    Scientific Data 11(1):1000, doi:10.1038/s41597-024-03738-7, ://doi.org/10.1038/s41597-024-03738-7

    Anders C, Moontaha S, Real S, Arnrich B (2024) Unobtrusive measurement of cognitive load and physiological signals in uncontrolled environments. Scientific Data 11(1):1000, doi:10.1038/s41597-024-03738-7, ://doi.org/10.1038/s41597-024-03738-7

    work page doi:10.1038/s41597-024-03738-7 2024

  80. [97]

    In: Proc

    Barke S, James MB, Polikarpova N (2023) Grounded copilot: How programmers interact with code-generating models. In: Proc. of the ACM on Programming Languages, vol 7, pp 85:1--85:27, doi:10.1145/3586030, distinguished Paper Award

    work page doi:10.1145/3586030 2023

Showing first 80 references.