Pith. sign in

REVIEW 6 major objections 6 minor 77 references

Baseline behaviour in human vision

T0 review · 6 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Gaze patterns contain a universal baseline behaviour: a fixed, memoryless rule based only on the previous saccade's length and angle predicts eye movements across tasks, ages, and stimuli, implying a hardwired motor prior.

desk verdict A large-scale demonstration that relative saccade lengths carry cross-task predictive information, but the paper's claim to model spatial gaze patterns is not supported by its length-only evaluation. read the letter →

arxiv 2507.14573 v1 pith:C7DS2YTZ submitted 2025-07-19 q-bio.NC

Thomas Fabian This is my paper · ORCID
classification q-bio.NC
keywords visualbehaviourgazedynamicsrelativesaccadelengthscomputationalmodellingcognitiveneuroscienceMarkovchainscanpathpredictionmotorprior
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that human gaze behaviour contains a universal, context-invariant component: a memoryless rule that uses only the previous saccade's length and angle can describe real eye movements during reading, visual search, and scene perception, in both adults and children, with no information about the task, stimulus, or viewer. The rule is formalised as BURRITOS (Baseline Using Relative Responses Independent of Task, Observer, and Stimulus), a Markov-chain model that sets the next saccade's length by multiplying the current length by a ratio sampled from the distribution of relative saccade lengths, and sets its angle from an empirical mixture of directions weighted toward horizontal and vertical movement with a suppressed return. Across 36 public eye-tracking experiments (4,170 participants, 233,584 trials, more than 69 million fixations), the model beats a uniform random baseline in discriminating real saccade lengths, aligning probability with fixation locations, and reproducing the length distribution in most categories. The authors take this as the first evidence for an intrinsic oculomotor motor prior: a baseline behaviour that task- and stimulus-driven signals modulate rather than create.

What carries the argument

The load-bearing object is the distribution of relative saccade lengths and the fixed transition kernel derived from it. The paper pools saccades from its corpus, divides each length by the preceding length to obtain ratios, and fits the pooled ratio histogram in two parts: a fitted decay function below one and a power law $y = (x \cdot 6.73)^{-2.03}$ above one. BURRITOS is a Markov chain whose state stores only the current fixation and the last saccade's length and angle; the next length is the previous length multiplied by a ratio sampled from this distribution, and the next angle is drawn from a mixture of von Mises distributions centred on the horizontal and vertical directions, with probabilities in the 40° range pointing back to the last fixation reduced by 70% to implement inhibition of return. Because the kernel is fixed and applied identically to every trial without retraining, any predictive success it achieves is attributable to the regularity itself rather than to task-specific fitting, which is what lets the authors call it a baseline.

What would settle it

Estimate the relative saccade length distribution separately for each of the four categories, or for each of the 36 datasets, and test whether the fitted kernels agree within confidence intervals; if reading or visual search yields a materially different curve, the single universal kernel is refuted. A complementary test is to hold out whole datasets when estimating the kernel and evaluate BURRITOS only on datasets never seen during fitting, since the zero-shot claim requires the model to generalise to unseen corpora rather than merely to match the corpus that generated its transition probabilities.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that gaze patterns follow a baseline behaviour: the distribution of relative saccade lengths, meaning each saccade's length divided by the length of the one before it, is stable enough across tasks, observers, and stimuli that a single fixed transition kernel built from it predicts scanpaths without retraining. BURRITOS instantiates this kernel as a Markov chain over fixation positions, sampling each new length as the previous length times a random ratio and each new angle from a cardinal-biased distribution with a memoryless inhibition-of-return term. Evaluated on its 36-experiment corpus, the model significantly outperforms a uniform random model on saccade-length discrimination (AUC) and saliency alignment (NSS) in every category, reading, visual search, scene perception, and children, and on full-distribution fidelity (EMD) in reading, scene perception, and children, with visual search statistically tied. That combination of generality and parsimony supports the paper's central conclusion: a substantial fraction of scanpath structure emerges from the visual system's own dynamics, so eye-movement patterns are comparable across contexts, contrary to the prevailing task-centric view.

Load-bearing premise

The argument stands or falls on whether the distribution of relative saccade lengths estimated from the pooled corpus is a true universal property, because the model is evaluated on data that went into producing its kernel, so if that distribution is partly an artefact of the specific datasets pooled to estimate it, the cross-task baseline claim is overstated.

Editorial extensions

If this is right

  • Gaze models gain a mandatory null: any model that claims task- or stimulus-specific explanatory power must first beat a zero-information, fixed-kernel baseline, and its advantage can then be attributed to content-driven mechanisms.
  • Scanpath prediction becomes possible without saliency maps, centre bias, or memory, at least for long trials: the kernel's AUC and NSS improve steadily with trial length, reaching well above chance past roughly 100 fixations.
  • The same transition kernel applies to adults and children and across language families, so gaze patterns from different contexts can be compared against one normative baseline rather than only within-task.
  • The neurobiological corollary is that, since brain activity predicts eye-movement patterns, a context-invariant kernel implies a hardwired default visual attention network whose activity is modulated, not created, by task demands.
  • Clinically and practically, deviations from the normative kernel become candidate low-resource biomarkers for oculomotor changes in disease, and the kernel itself can serve as a zero-shot saliency prior in computer vision.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The kernel's multiplicative structure, where the next length equals the current length times a random ratio, makes gaze a scale-invariant random walk, so the same regularity might also govern saccade timing if applied to fixation durations; the paper does not compute that extension.
  • The near-chance performance on short trials suggests the kernel describes the stationary statistics of gaze rather than its initiation; a stronger test of the motor-prior hypothesis would check whether the kernel also constrains the first few saccades once the starting fixation is controlled.
  • The one category where full-distribution fidelity failed, visual search, is a natural probe of how much context actually modulates the baseline; a hybrid model that lets task difficulty gate the kernel would quantify the modulation the paper leaves as a consequence.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

6 major / 6 minor

Summary. The manuscript introduces BURRITOS, a Markov-chain model that generates saccade lengths by multiplying the previous saccade length by a sample from a 'relative saccade length' distribution, and saccade angles from a von Mises mixture with an inhibition-of-return term, using no stimulus, task, or observer information. The model is evaluated on a large corpus of 36 eye-tracking datasets (4,170 participants; 233,584 trials) spanning reading, visual search, scene perception, and children. The authors report that BURRITOS outperforms a uniform baseline on AUC, NSS, and EMD, with the exception of EMD in visual search, and interpret this as evidence for a context-invariant oculomotor baseline and a default visual attention network.

Significance. If the central claim were established, this would be a valuable contribution: a simple, information-agnostic baseline model for gaze prediction, with implications for benchmarking specialized models and for theories of oculomotor priors. The assembled corpus is exceptionally large and heterogeneous, and the formalization of the relative saccade length distribution is a useful step. The paper also has genuine strengths in transparency (public data, explicit model equations) and in proposing a parsimonious, falsifiable baseline. However, as detailed below, the reported evaluations do not currently test the central claim about spatial gaze patterns, and several methodological issues (EMD definition, circular evaluation, parameter freedom) mean the conclusions are not yet supported.

major comments (6)
  1. [Methods, Performance evaluation (EMD formula)] The formula EMD = Σ_j |p_j − p'_j| Δc is not the Earth Mover's Distance or first-order Wasserstein distance; that distance for one-dimensional histograms is Σ_k |Σ_{j≤k}(p_j − p'_j)| Δc. The formula as written is a scaled total variation distance. Because the text interprets EMD as the 'average number of pixels by which a predicted saccade differs from the empirical one' and uses the EMD results to claim distributional fidelity, the EMD-based conclusions are not supported as reported. Please recalculate with the correct Wasserstein metric or relabel the measure and adjust the interpretation.
  2. [Model evaluation / Performance evaluation] All three metrics (AUC, NSS, EMD) are computed on the one-dimensional bin index b_i of the empirical saccade length only; angle and fixation position are never scored. The NSS is computed over length bins, not image locations, and the EMD is over length histograms, not spatial scanpaths. Consequently, the reported results do not establish that BURRITOS describes 'gaze patterns' or that the angle kernel and inhibition-of-return term contribute anything; a model with the same length kernel and uniformly random angles would receive identical scores on all three metrics. The central claim of context-invariant spatial gaze behaviour therefore requires evaluation with spatial scanpath metrics (e.g., saccade-angle distributions, fixation maps, scanpath similarity measures).
  3. [Methods, Data and Distribution estimation / Discussion] The relative saccade length distribution and the angle distribution are estimated from the same 36 datasets on which the model is evaluated (the Methods state that the data and preprocessing procedure follow ref. 20, and the corpus description is the same set of experiments). Thus the comparisons do not test zero-shot generalization to new contexts; they test in-sample fit. Moreover, the model is not parameter-free: the power-law and Gaussian parameters are fitted, the von Mises mixture is fitted to the pooled angles, and the inhibition-of-return range (40°) and strength (70%) are hand-chosen. A leave-one-dataset-out or external validation is needed to support the universal-baseline claim.
  4. [Methods, Distribution estimation] The functional form labelled 'exponential' for relative saccade lengths smaller than one is given as y = 0.023 * exp(-1.13 * (x - 0.59)^2), which contains a squared term and is a Gaussian, not an exponential. This is not a mere typo: the claimed exponential form is part of the formalized universal regularity, and the paragraph describing the model selection ('best-fitting functions are the exponential function...') is internally inconsistent. Please correct the equation and the text, and provide the model-selection details.
  5. [Results, Baseline comparison across all experiments] The comparison of BURRITOS with the uniform baseline across all datasets uses Welch's t-test on 36 dataset means (t = 2.51, 3.58, -6.69). Because both models are evaluated on exactly the same trials, the observations are paired; the appropriate test is a paired t-test or Wilcoxon signed-rank test. The unpaired test ignores the pairing and can yield anti-conservative p-values, so the reported significance levels should be recomputed.
  6. [Results, Correlation of model performance with trial length / Abstract] The correlation analysis shows that AUC and NSS are near chance for short trials and only exceed 0.7 for trials containing more than about 100 fixations. Many natural viewing tasks, and especially the children's data, contain short trials. The abstract's unconditional claim that the model 'can describe gaze patterns during reading, visual search, and scene perception' is therefore too strong; please report the distribution of trial lengths per category and qualify the claim accordingly.
minor comments (6)
  1. [Fig. 3 caption] The caption states that for EMD crosses below the diagonal indicate better performance, then states that for EMD crosses above the diagonal imply better performance; please correct the contradiction.
  2. [Methods, Model structure] The von Mises mixture g(θ) is described verbally, but the weights w_f, w_o, w_r and concentrations κ_f, κ_o, κ_r are never reported; without these values the angle kernel is not reproducible.
  3. [Methods, Performance evaluation] The AUC computation samples a negative score by uniformly drawing a bin index; for the uniform baseline all scores are equal, so the stated chance level of 0.5 depends on a tie-handling convention that is not specified.
  4. [Data availability] The data availability statement says all datasets are publicly available in the repositories of the respective publications, but no list of URLs or accessions is provided; a supplementary table would aid reproducibility.
  5. [References] Reference 7 spells the publisher as 'Plenium'; it should be 'Plenum'.
  6. [Results, Baseline comparison across all experiments] The term 'dataset-by-dataset covariance' (Fig. 3) appears to mean co-variation or correlation; please use standard terminology.

Circularity Check

3 steps flagged · score 6.0 of 10

Transition kernel fitted to the same corpus that is then 'predicted'; all metrics score only length bins, so the central gaze-pattern claim is partly circular.

  1. fitted input called prediction [Methods, 'Distribution estimation' and 'Performance evaluation'; Results, 'Model evaluation']
    "For estimating the distribution of relative saccade lengths, we use the data and preprocessing procedure from the original publication on relative saccade lengths 20. ... To measure the performance of BURRITOS and the baseline model, we let both models predict the fixations for each empirical trial in our eye-tracking corpus."

    The kernel is estimated from the same corpus on which it is evaluated. The 36-dataset corpus is assembled 'follow[ing] the procedure that was used to investigate the regularity of relative saccade lengths 20', and the distribution is estimated from 'the data and preprocessing procedure from the original publication'. BURRITOS then samples lengths from this fitted distribution and is scored against the same trials' length bins. The AUC/NSS/EMD results are therefore in-sample measures of how well a distribution fitted to the pooled corpus matches the same corpus, not held-out predictions. The Discussion's 'zero-shot generalisation' claim is contradicted by construction.

  2. self citation load bearing [Introduction, paragraph beginning 'This seemingly inevitable dependence...' and Results, 'Model structure']
    "Following the finding that the distribution of relative saccade lengths remains stable across tasks, observers, and stimuli20, we propose and formalise a minimalist Markov chain eye movement model that rests solely on empirical regularities of eye movements relative to the preceding eye movement."

    Reference 20 is the present author's own prior paper (Fabian, Cognition 257, 106079, 2025). The central premise—a context-invariant relative-saccade-length regularity—is imported from that self-citation, and the Methods state that the present corpus and distribution estimation use that paper's data and preprocessing. The model's universality is therefore not independently derived here; it is assumed from a same-author prior result and then 'validated' on the same data that produced that result. This is load-bearing because removing ref. 20 removes the only justification for treating the fitted length distribution as a universal motor prior.

1 more flagged steps
  1. renaming known result [Results, 'Model evaluation' and Methods, 'Performance evaluation']
    "To evaluate whether BURRITOS assigns higher saliency to the exact locations where fixations occur, we employ the Normalised Scanpath Saliency (NSS) 37. ... NSS denotes the mean zbi over all empirical saccades, with positive NSS indicating that observed saccades fall in bins above the model’s mean probability."

    NSS is introduced as a spatial saliency metric ('assigns higher saliency to the exact locations where fixations occur'), but in Methods it is computed as the mean z-score zbi over saccade-length bins; AUC and EMD likewise use only the bin index bi of observed lengths. The angle distribution, inhibition-of-return term, and fixation positions are never scored. Consequently, the claim that BURRITOS 'describe[s] gaze patterns' and 'spatially aligns its predictions' is a renaming of a length-histogram comparison: the spatial scanpath claim is definitionally reduced to the fitted length distribution, so the reported performance is not evidence about gaze patterns in space.

full rationale

The paper's central novelty is a context-invariant transition kernel for gaze. The kernel is not derived from first principles; it is the empirical relative-saccade-length distribution of ref. 20, which is the same author's prior work and is estimated from the same 36-dataset corpus used for evaluation. Consequently, the reported AUC/NSS/EMD values are in-sample fits of a fitted marginal distribution, not held-out predictions. There is some non-forced content: BURRITOS is compared with a uniform baseline, and the trial-length correlations are not guaranteed; an arbitrary fit could fail these comparisons. However, the comparisons do not rescue the universality claim because all three metrics score only saccade-length bins, and the angle/position components of the model are never tested. The paper's naming of a length-bin z-score as NSS, a spatial saliency measure, further lets a length-histogram match stand in for 'gaze pattern' prediction. The self-citation of ref. 20 is load-bearing: the existence of a task-invariant relative-length distribution is the premise, and the present validation re-uses the data behind that premise. This is partial circularity: the 'prediction' reduces to re-sampling a fitted distribution and checking how well it matches the corpus it was fitted on, but the comparison against uniform and the correlation analyses provide some non-tautological signal. Score 6.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The model rests on a small number of fitted empirical distributions and hand-chosen parameters. The relative saccade length kernel, the angle mixture, and the inhibition-of-return constants are all either fitted to the evaluation corpus or chosen ad hoc. The invented neural entity, a default visual attention network, is an interpretation with no direct evidence in this paper.

free parameters (5)
  • Power-law parameters for relative saccade lengths > 1 = scale 6.73, exponent 2.03
    Fitted to pooled eye-tracking data using Levenberg-Marquardt; no error bars or held-out validation reported (Methods, Distribution estimation).
  • Gaussian parameters for relative saccade lengths < 1 = 0.023, 1.13, 0.59
    The text calls this an exponential function but the formula is a Gaussian; fitted by linear least squares on the same corpus (Methods, Distribution estimation).
  • Von Mises mixture weights and concentrations = not reported
    The angle distribution is described as a mixture of four von Mises distributions with weights wf, wo, wr and concentrations kappa; numeric values are not given, so the model cannot be exactly reproduced.
  • Inhibition-of-return range and strength = 40 degrees, 70%
    The authors state these are arbitrary and chosen by hand (Methods, Model structure).
  • Initial saccade length = 1 pixel
    Arbitrary initialization needed to start the Markov chain (Methods, Model structure).
assumptions (4)
  • domain assumption The distribution of relative saccade lengths is invariant across tasks, observers, and stimuli.
    This is the core regularity from the author's prior work (ref 20), assumed before model construction and used to justify the transition kernel (Introduction; Methods, Distribution estimation).
  • domain assumption The next saccade depends only on the previous saccade length and angle (Markov property).
    BURRITOS is defined as a first-order Markov chain; no memory or higher-order dependencies are modeled (Methods, Model structure).
  • domain assumption The empirical angle distribution and the chosen von Mises mixture apply across all viewing contexts.
    The angle distribution is fitted to pooled eye movements and applied unchanged to reading, search, scene perception, and children (Methods, Model structure).
  • domain assumption The 36 public datasets are representative of human gaze behavior.
    The corpus is a convenience sample with acknowledged geographic and stimulus biases (Methods, Data).
invented entities (1)
  • Context-invariant motor prior / default visual attention network
    purpose: Postulated neural substrate that would generate the baseline gaze behavior the model captures.
    No neuroimaging or physiological evidence is presented in this paper; the fMRI citation (ref 35) shows eye movements can be predicted from brain activity but does not establish a default network.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Baseline behaviour in human vision." pith.science (2026). https://pith.science/paper/C7DS2YTZ

@misc{pith2026250714573,
  author       = {Pith},
  title        = {Pith review of: Baseline behaviour in human vision},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C7DS2YTZ}},
  note         = {Machine review of arXiv:2507.14573}
}
read the original abstract

Humans perceive their visual environment by directing their eyes towards relevant objects. The deployment of visual attention depends substantially on the stimulus's properties, higher cognitive processes, and biases and constraints of the visual system. Numerous models describe people's eye movements depending on the performed task or the viewed content. However, there is no universal, context-invariant model of human gaze behaviour. Here we show that statistical regularities can be utilised to model human gaze behaviour regardless of task, observer, and content. Using a context-agnostic eye movement model, we were able to describe human gaze behaviour better than a uniform random model in various viewing situations. Using a fixed transition kernel, the model can describe gaze patterns during reading, visual search, and scene perception, as well as for both adults and children. Thus, contrary to current belief, human gaze patterns follow a baseline behaviour, making them comparable across contexts. Since gaze behaviour is directly related to brain structure, our results provide the first evidence for the existence of an underlying, context-invariant motor prior in the human visual system.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

77 extracted references · 75 canonical work pages

  1. [1]

    Capabilities and Limitations of Peripheral Vision

    Rosenholtz, R. Capabilities and Limitations of Peripheral Vision. Annu. Rev. Vis. Sci. 2, 437–457 (2016)

    work page 2016

  2. [2]

    & Jüttner, M

    Strasburger, H., Rentschler, I. & Jüttner, M. Peripheral vision and pattern recognition: A review. Journal of Vision 11, 13 (2011)

    work page 2011

  3. [3]

    Interaction effects in parafoveal letter recognition

    Bouma, H. Interaction effects in parafoveal letter recognition. Nature 226, 177–178 (1970)

    work page 1970

  4. [4]

    Eye movements: The past 25years

    Kowler, E. Eye movements: The past 25years. Vision Research 51, 1457–1483 (2011)

    work page 2011

  5. [5]

    Land, M. F. Motion and vision: why animals move their eyes. Journal of Comparative Physiology A 185, 341–352 (1999)

    work page 1999

  6. [6]

    Buswell, G. T. How People Look at Pictures: A Study of the Psychology and Perception in Art. 198 (University of Chicago Press, Chicago, 1935)

    work page 1935

  7. [7]

    Yarbus, A. L. Eye Movements and Vision. (Plenium, New York, 1967)

    work page 1967

  8. [8]

    Eye movements in reading and information processing: 20 years of research

    Rayner, K. Eye movements in reading and information processing: 20 years of research. Psychological Bulletin 124, 372–422 (1998). 27

    work page 1998

Show all 77 references
  1. [9]

    & König, P

    Kollmorgen, S., Nortmann, N., Schröder, S. & König, P. Influence of Low-Level Stimulus Features, Task Dependent Factors, and Spatial Biases on Overt Visual Attention. PLOS Computational Biology 6, e1000791 (2010)

    2010

  2. [10]

    Anderson, T. J. & MacAskill, M. R. Eye movements in patients with neurodegenerative disorders. Nature Reviews Neurology 9, 74–85 (2013)

    2013

  3. [11]

    & Säljö, R

    Gegenfurtner, A., Lehtinen, E. & Säljö, R. Expertise Differences in the Comprehension of Visualizations: a Meta-Analysis of Eye-Tracking Research in Professional Domains. Educational Psychology Review 23, 523–552 (2011)

    2011

  4. [12]

    & Niebur, E

    Itti, L., Koch, C. & Niebur, E. A model of saliency-based visual attention for rapid scene analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence 20, 1254–1259 (1998)

    1998

  5. [13]

    Wolfe, J. M. & Horowitz, T. S. What attributes guide the deployment of visual attention and how do they do it? Nature Reviews Neuroscience 5, 495–501 (2004)

    2004

  6. [14]

    S., Gatys, L

    Kümmerer, M., Wallis, T. S., Gatys, L. A. & Bethge, M. Understanding low-and high-level contributions to fixation prediction. in Proceedings of the IEEE international conference on computer vision (ICCV) 4789–4798 (2017). doi:10.1109/ICCV.2017.513

    2017 doi

  7. [15]

    & Ullman, S

    Koch, C. & Ullman, S. Shifts in Selective Visual Attention: Towards the Underlying Neural Circuitry. in Matters of Intelligence: Conceptual Structures in Cognitive Neuroscience (ed. Vaina, L. M.) 115–141 (Springer Netherlands, Dordrecht, 1987). doi:10.1007/978-94-009-3833-5_5

    1987 doi

  8. [16]

    Williams, C. C. & Castelhano, M. S. The Changing Landscape: High-Level Influences on Eye Movement Guidance in Scenes. Vision 3, (2019)

    2019

  9. [17]

    Henderson, J. M. & Hayes, T. R. Meaning-based guidance of attention in scenes as revealed by meaning maps. Nature Human Behaviour 1, 743–747 (2017)

    2017

  10. [18]

    Tatler, B. W. The central fixation bias in scene viewing: Selecting an optimal viewing position independently of motor biases and image feature distributions. Journal of Vision 7, 4–4 (2007)

    2007

  11. [19]

    Posner, M. I. & Cohen, Y. Components of visual orienting. in Attention and performance X: Control of language processes (eds. Bouma, H. & Bouwhuis, D.) 531–556 (Lawrence Erlbaum, London, 1984)

    1984

  12. [20]

    Exploring power-law behavior in human gaze shifts across tasks and populations

    Fabian, T. Exploring power-law behavior in human gaze shifts across tasks and populations. Cognition 257, 106079 (2025). 28

    2025

  13. [21]

    Bruce, N. D. B., Wloka, C., Frosst, N., Rahman, S. & Tsotsos, J. K. On computational modeling of visual saliency: Examining what’s right, and what’s left. Vision Research 116, 95–112 (2015)

    2015

  14. [22]

    Eye movements and attention in reading, scene perception, and visual search

    Rayner, K. Eye movements and attention in reading, scene perception, and visual search. The quarterly journal of experimental psychology 62, 1457–1506 (2009)

    2009

  15. [23]

    & Fischer, M

    Rayner, K. & Fischer, M. H. Mindless reading revisited: Eye movements during reading and scanning are different. Perception & Psychophysics 58, 734–747 (1996)

    1996

  16. [24]

    & Raney, G

    Rayner, K. & Raney, G. E. Eye movement control in reading and visual search: Effects of word frequency. Psychonomic Bulletin & Review 3, 245–248 (1996)

    1996

  17. [25]

    & Itti, L

    Borji, A. & Itti, L. State-of-the-Art in Visual Attention Modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence 35, 185–207 (2013)

    2013

  18. [26]

    & Bethge, M

    Kümmerer, M. & Bethge, M. State-of-the-art in human scanpath prediction. arXiv preprint arXiv:2102.12239 (2021)

    2021 arXiv

  19. [27]

    & Wallis, T

    Kümmerer, M., Bethge, M. & Wallis, T. S. A. DeepGaze III: Modeling free-viewing human scanpaths with deep learning. Journal of Vision 22, 7–7 (2022)

    2022

  20. [28]

    & Ferraro, M

    Boccignone, G. & Ferraro, M. Modelling gaze shift as a constrained random walk. Physica A: Statistical Mechanics and its Applications 331, 207–218 (2004)

    2004

  21. [29]

    A., Barthelmé, S

    Engbert, R., Trukenbrod, H. A., Barthelmé, S. & Wichmann, F. A. Spatial statistics and attentional dynamics in scene viewing. Journal of Vision 15, 14–14 (2015)

    2015

  22. [30]

    Reichle, E. D. Computational Models of Reading: A Handbook. (Oxford University Press, 2021)

    2021

  23. [31]

    Eye Movements in Reading: Models and Data

    Rayner, K. Eye Movements in Reading: Models and Data. Journal of Eye Movement Research 2, (2008)

    2008

  24. [32]

    & Geisel, T

    Brockmann, D. & Geisel, T. The ecology of gaze shifts. Neurocomputing 32–33, 643–650 (2000)

    2000

  25. [33]

    Tatler, B. W. & Vincent, B. T. The prominence of behavioural biases in eye guidance. Visual Cognition 17, 1029–1054 (2009)

    2009

  26. [34]

    Zettersten, M. et al. Peekbank: An open, large-scale repository for developmental eye-tracking data of children’s word recognition. Behavior Research Methods 55, 2485–2500 (2023)

    2023

  27. [35]

    O’Connell, T. P. & Chun, M. M. Predicting eye movement patterns from fMRI responses to natural scenes. Nature Communications 9, 5159 (2018). 29

    2018

  28. [36]

    W., Baddeley, R

    Tatler, B. W., Baddeley, R. J. & Gilchrist, I. D. Visual correlates of fixation selection: effects of scale and time. Vision Research 45, 643–659 (2005)

    2005

  29. [37]

    J., Iyer, A., Itti, L

    Peters, R. J., Iyer, A., Itti, L. & Koch, C. Components of bottom-up gaze allocation in natural images. Vision Research 45, 2397–2416 (2005)

    2005

  30. [38]

    & Guibas, L

    Rubner, Y., Tomasi, C. & Guibas, L. J. The Earth Mover’s Distance as a Metric for Image Retrieval. International Journal of Computer Vision 40, 99–121 (2000)

    2000

  31. [39]

    & Hochberg, Y

    Benjamini, Y. & Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society: Series B (Methodological) 57, 289–300 (1995)

    1995

  32. [40]

    Statistical Power Analysis for the Behavioral Sciences

    Cohen, J. Statistical Power Analysis for the Behavioral Sciences. (Routledge, New York, 1988). doi:10.4324/9780203771587

    1988 doi

  33. [41]

    Mosconi, M. W. & Sweeney, J. A. Sensorimotor dysfunctions as primary features of autism spectrum disorders. Science China Life Sciences 58, 1016–1023 (2015)

    2015

  34. [42]

    Pereira, M. L. F., Camargo, M. von Z. A., Aprahamian, I. & Forlenza, O. V. Eye movement analysis and cognitive processing: detecting indicators of conversion to Alzheimer’s disease. NDT 10, 1273–1285 (2014)

    2014

  35. [43]

    & Björnsdóttir, M

    Hollenstein, N., Barrett, M. & Björnsdóttir, M. The Copenhagen Corpus of Eye Tracking Recordings from Natural Reading of Danish Texts. in Proceedings of the Thirteenth Language Resources and Evaluation Conference (eds. Calzolari, N. et al.) 1712–1720 (European Language Resourc...

    2022

  36. [44]

    S., Eerland, A

    Eekhof, L. S., Eerland, A. & Willems, R. M. Readers’ Insensitivity to Tense Revealed: No Differences in Mental Simulation During Reading of Present and Past Tense Stories. Collabra: Psychology 4, 16 (2018)

    2018

  37. [45]

    & Ferguson, H

    Black, J., Barzy, M., Williams, D. & Ferguson, H. Intact counterfactual emotion processing in autism spectrum disorder: Evidence from eye-tracking. Autism Research 12, 422–444 (2019)

    2019

  38. [46]

    Lowder, M. W. & Gordon, P. C. Relative Clause Effects at the Matrix Verb Depend on Type of Intervening Material. Cognitive Science 45, e13039 (2021)

    2021

  39. [47]

    Mirault, J. et al. Parafoveal-on-foveal repetition effects in sentence reading: A co-registered eye- tracking and electroencephalogram study. Psychophysiology 57, e13553 (2020). 30

    2020

  40. [48]

    Cognitive constraints on advance planning of sentence intonation

    Ots, N. Cognitive constraints on advance planning of sentence intonation. PLOS ONE 16, e0259343 (2021)

    2021

  41. [49]

    & Meltzer-Asscher, A

    Kobo, O., Schonberg, T. & Meltzer-Asscher, A. Individual differences in (Hebrew) sentence processing using Eye-Tracking. (2020) doi:10.17605/OSF.IO/9ERJK

    2020 doi

  42. [50]

    A., Laurinavichyute, A

    Ziubanova, A. A., Laurinavichyute, A. K. & Parshina, O. Does early exposure to spoken and sign language affect reading fluency in deaf and hard-of-hearing adult signers? Frontiers in Psychology 14, (2023)

    2023

  43. [51]

    R., Bulls, L

    Koslov, S. R., Bulls, L. S. & Lewis-Peacock, J. A. Distinct monitoring strategies underlie costs and performance in prospective memory. Memory & Cognition 50, 1772–1788 (2022)

    2022

  44. [52]

    C., Walenchok, S

    Ratiu, I., Hout, M. C., Walenchok, S. C., Azuma, T. & Goldinger, S. D. Comparing visual search and eye movements in bilinguals and monolinguals. Attention, Perception, & Psychophysics 79, 1695–1725 (2017)

    2017

  45. [54]

    Zhang, H., Miller, K. F. & Jonides, J. How does mind-wandering affect distractor suppression? Visual Cognition 1–18 doi:10.1080/13506285.2024.2347605

    2024

  46. [55]

    Ferguson, H. J. & Cane, J. Tracking the impact of depression in a perspective-taking task. Scientific Reports 7, 14821 (2017)

    2017

  47. [56]

    Slim, M. S. & Hartsuiker, R. J. Moving visual world experiments online? A web-based replication of Dijkgraaf, Hartsuiker, and Duyck (2017) using PCIbex and WebGazer.js. Behavior Research Methods 55, 3786–3804 (2023)

    2017

  48. [57]

    & Yuval-Greenberg, S

    Amit, R., Abeles, D., Carrasco, M. & Yuval-Greenberg, S. Oculomotor inhibition reflects temporal expectations. NeuroImage 184, 279–292 (2019)

    2019

  49. [58]

    & Lew-Williams, C

    Byers-Heinlein, K., Morin-Lessard, E. & Lew-Williams, C. Bilingual infants control their languages as they listen. Proceedings of the National Academy of Sciences 114, 9032–9037 (2017)

    2017

  50. [59]

    Niedtfeld, I. et al. Enhanced memory for negative social information in borderline personality disorder. Journal of Abnormal Psychology 129, 480–491 (2020). 31

    2020

  51. [60]

    & Straube, T

    Bruchmann, M., Schindler, S. & Straube, T. The spatial frequency spectrum of fearful faces modulates early and mid-latency ERPs but not the N170. Psychophysiology 57, e13597 (2020)

    2020

  52. [61]

    & Straube, T

    Bruchmann, M., Schindler, S., Heinemann, J., Moeck, R. & Straube, T. Increased early and late neuronal responses to aversively conditioned faces across different attentional conditions. Cortex 142, 332–341 (2021)

    2021

  53. [62]

    & Davachi, L

    Clewett, D., Gasser, C. & Davachi, L. Pupil-linked arousal signals track the temporal organization of events in memory. Nature Communications 11, 4007 (2020)

    2020

  54. [63]

    The influence of social stigmas on observers’ eye movements

    Hermens, F. The influence of social stigmas on observers’ eye movements. Journal of Articles in Support of the Null Hypothesis 14, (2017)

    2017

  55. [64]

    C., Levy, R

    Meylan, S. C., Levy, R. P. & Bergelson, E. Children’s Expressive and Receptive Knowledge of the English Regular Plural. Proceedings of the Annual Meeting of the Cognitive Science Society 42, (2020)

    2020

  56. [65]

    Bogdanov, V. B. et al. Nature and the City: Audiovisual interactions in pleasantness and psychophysiological reactions. Applied Acoustics 193, 108762 (2022)

    2022

  57. [66]

    C., Sugarman, E., Horowitz, A

    Frank, M. C., Sugarman, E., Horowitz, A. C., Lewis, M. L. & Yurovsky, D. Using Tablets to Collect Data From Young Children. Journal of Cognition and Development 17, 1–17 (2016)

    2016

  58. [67]

    J., McMillan, B

    Mahr, T. J., McMillan, B. T. M., Saffran, J. R., Ellis Weismer, S. & Edwards, J. Anticipatory coarticulation facilitates word recognition in toddlers. Cognition 142, 345–350 (2015)

    2015

  59. [68]

    & Frank, M

    Yurovsky, D., Wade, A. & Frank, M. Online Processing of Speech and Social Information in Early Word Learning. in Proceedings of the 35th Annual Conference of the Cognitive Science Society (eds. Knauff, M., Pauen, M. & Sebanz, N.) 1641–1646 (Cognitive Science Society, Austin, TX, 2013)

    2013

  60. [69]

    & Frank, M

    Yurovsky, D. & Frank, M. C. Beyond naïve cue combination: salience and social cues in early word learning. Developmental Science 20, e12349 (2017)

    2017

  61. [70]

    & Frank, M

    Yurovsky, D. & Frank, M. C. Development of Social Attention. (2016)

    2016

  62. [71]

    W., Harding, S

    Boyer, T. W., Harding, S. M. & Bertenthal, B. I. The temporal dynamics of infants’ joint attention: Effects of others’ gaze cues and manual actions. Cognition 197, 104151 (2020)

    2020

  63. [72]

    & Haun, D

    Thiele, M., Hepach, R., Michel, C. & Haun, D. Infants’ Preference for Social Interactions Increases from 7 to 13 Months of Age. Child Development 92, 2577–2594 (2021). 32

    2021

  64. [73]

    Information theory and an extension of the maximum likelihood principle

    Akaike, H. Information theory and an extension of the maximum likelihood principle. in (eds. Petrox, B. N. & Caski, F.) 267–281 (Akademiai Kiado, Budapest, 1973)

    1973

  65. [74]

    Estimating the Dimension of a Model

    Schwarz, G. Estimating the Dimension of a Model. The Annals of Statistics 6, 461–464 (1978)

    1978

  66. [75]

    Generalized linear models

    McCullagh, P. Generalized linear models. European Journal of Operational Research 16, 285–292 (1984)

    1984

  67. [76]

    Applied Linear Regression

    Weisberg, S. Applied Linear Regression. (John Wiley & Sons, 2005)

    2005

  68. [77]

    A Method for the Solution of Certain Non-Linear Problems in Least Squares

    Levenberg, K. A Method for the Solution of Certain Non-Linear Problems in Least Squares. Quarterly of Applied Mathematics 2, 164–168 (1944)

    1944

  69. [78]

    Marquardt, D. W. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. Journal of the Society for Industrial and Applied Mathematics 11, 431–441 (1963)

    1963

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.