Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 86 of the 91 references it cites were found in a registry or an index.

Parsing eye-tracking data of variable quality to provide accurate fixation duration estimates in infants and adults

Open access
Wass, S. V., Smith, T. J., Johnson, M. H. · 2013 · Behavior Research Methods

Abstract

Researchers studying infants' spontaneous allocation of attention have traditionally relied on hand-coding infants' direction of gaze from videos; these techniques have low temporal and spatial resolution and are labor intensive. Eye-tracking technology potentially allows for much more precise measurement of how attention is allocated at the subsecond scale, but a number of technical and methodological issues have given rise to caution about the quality and reliability of high temporal resolution data obtained from infants. We present analyses suggesting that when standard dispersal-based fixation detection algorithms are used to parse eye-tracking data obtained from infants, the results appear to be heavily influenced by interindividual variations in data quality. We discuss the causes of these artifacts, including fragmentary fixations arising from flickery or unreliable contact with the eyetracker and variable degrees of imprecision in reported position of gaze. We also present new algorithms designed to cope with these problems by including a number of new post hoc verification checks to identify and eliminate fixations that may be artifactual. We assess the results of our algorithms by testing their reliability using a variety of methods and on several data sets. We contend that, with appropriate data analysis methods, fixation duration can be a reliable and stable measure in infants. We conclude by discussing ways in which studying fixation durations during unconstrained orienting may offer insights into the relationship between attention and learning in naturalistic settings.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3758/s13428-012-0245-6

Cites

86 of 86 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Frequency distribution showing fixation durations returned by a standard dispersal algorithm from 3 sample infant participants
Fig. 1 Frequency distribution showing fixation durations returned by a standard dispersal algorithm from 3 sample infant participantsCC BY · PubMed Central
Fig. 2 Sample data plots. Data were taken from 3 infant participants. Sample 1 shows high-quality tracking, sample 2 shows low precision, and sample 3 shows flicker. The exact way in which these terms are used are defined in detail in the text
Fig. 2 Sample data plots. Data were taken from 3 infant participants. Sample 1 shows high-quality tracking, sample 2 shows low precision, and sample 3 shows flicker. The exact way in which these terms are used are defined in detail in the textCC BY · PubMed Central
Fig. 3 Quantifying how fixation duration varies as a function of data quality. The relationship between fixation durations (as analyzed using standard dispersal algorithms) and data quality is evaluated in two ways: a flicker, the mean duration of raw data segments, And b precision, the variance in reported point of gaze (POG)
Fig. 3 Quantifying how fixation duration varies as a function of data quality. The relationship between fixation durations (as analyzed using standard dispersal algorithms) and data quality is evaluated in two ways: a flicker, the mean duration of raw data segments, And b precision, the variance in reported point of gaze (POG)CC BY · PubMed Central
Fig. 4 Quantifying performance of a standard dispersal-based algorithm relative to hand-coding as a function of data quality. a Relationship between flicker and proportion agreement between automatic and hand-coding. b Relationship between precision and proportion agreement between automatic and hand-coding. c Relationship between flicker and mean fixation duration, as processed using both hand-coding and standard dispersal algorithms. d Relationship between precision and mean fixation duration, as processed using both hand-coding and standard dispersal algorithms
Fig. 4 Quantifying performance of a standard dispersal-based algorithm relative to hand-coding as a function of data quality. a Relationship between flicker and proportion agreement between automatic and hand-coding. b Relationship between precision and proportion agreement between automatic and hand-coding. c Relationship between flicker and mean fixation duration, as processed using both hand-coding and standard dispersal algorithms. d Relationship between precision and mean fixation duration, as processed using both hand-coding and standard dispersal algorithmsCC BY · PubMed Central
Fig. 5 Evaluating the performance of a standard dispersal-based algorithm using simulations of the effect of flicker and precision. a Initial sample of clean data. The x- and y-coordinates are plotted, followed by a velocity plot (with the velocity threshold marked in blue). The fixations being returned by the standard dispersal algorithm are marked below the text as orange bars. b The same data distorted to simulate the effect of flickeriness. The orange bars show the fixations that are returned when these data are processed using standard dispersal algorithms. c The same data distorted to simulate the effect of low precision in reported point of gaze (POG). d) Frequency distribution showing the fixation durations returned in the three conditions. e Number of fixations identified in the three conditions. f Mean fixation duration returned by the algorithm in the three conditions
Fig. 5 Evaluating the performance of a standard dispersal-based algorithm using simulations of the effect of flicker and precision. a Initial sample of clean data. The x- and y-coordinates are plotted, followed by a velocity plot (with the velocity threshold marked in blue). The fixations being returned by the standard dispersal algorithm are marked below the text as orange bars. b The same data distorted to simulate the effect of flickeriness. The orange bars show the fixations that are returned when these data are processed using standard dispersal algorithms. c The same data distorted to simulate the effect of low precision in reported point of gaze (POG). d) Frequency distribution showing the fixation durations returned in the three conditions. e Number of fixations identified in the three conditions. f Mean fixation duration returned by the algorithm in the three conditionsCC BY · PubMed Central
Fig. 6 Sample plots of eyetracker data obtained from typically developing 11-month-old infants viewing static (still images) that illustrate the issues identified in Fig. 5. In each case, time (in seconds) has been plotted along the x-axis. Vertically from top to bottom each data plot shows missing data, raw x- and y-coordinates, and fixations identified by a standard dispersal algorithm (drawn as red bars). a Instance of a participant showing high flicker in the contact with the eyetracker, with multiple incomplete fixations being stored. b Samples from 2 participants, one of whom shows high precision data (with accurate identification of fixations), the other of whom shows low precision data (with inaccurate identification of fixations). On these plots, the velocity derived from the x- and y-coordinates has been plotted, along with the velocity threshold (drawn as an orange line). In sample 2, it can be seen that the velocity threshold is being surpassed due to random sampling error between iterations, leading to the false identification of fixations
Fig. 6 Sample plots of eyetracker data obtained from typically developing 11-month-old infants viewing static (still images) that illustrate the issues identified in Fig. 5. In each case, time (in seconds) has been plotted along the x-axis. Vertically from top to bottom each data plot shows missing data, raw x- and y-coordinates, and fixations identified by a standard dispersal algorithm (drawn as red bars). a Instance of a participant showing high flicker in the contact with the eyetracker, with multiple incomplete fixations being stored. b Samples from 2 participants, one of whom shows high precision data (with accurate identification of fixations), the other of whom shows low precision data (with inaccurate identification of fixations). On these plots, the velocity derived from the x- and y-coordinates has been plotted, along with the velocity threshold (drawn as an orange line). In sample 2, it can be seen that the velocity threshold is being surpassed due to random sampling error between iterations, leading to the false identification of fixationsCC BY · PubMed Central
Fig. 7 Sample processed data plot. The time (in seconds) is plotted along the x-axis, top of the screen, and the number of iterations along the x-axis, bottom of the screen. Vertically from top to bottom, we have plotted the following: missing data (above the time axis); rough (i.e. raw) data, with separate x- and y-coordinates for left and right eyes; smoothed data (with interpolated sections marked in red); fixations (tentative fixations marked as a thick green line, and fixations that have passed all of our verification checks marked in orange); bad data flags (absent in this sample); and velocity (calculated from raw data in black and from smoothed data in red). The velocity threshold of 35ºs-1 is drawn as a thin blue line. On the right hand side of the screen, the participant and trial numbers are drawn (in this case ‘4011_23’), along with the stimulus the infant was viewing, with their gaze positions (left eye, blue; right eye, red) drawn on the screen. The letters a–d below the bottom axis refer to examples we describe in the text
Fig. 7 Sample processed data plot. The time (in seconds) is plotted along the x-axis, top of the screen, and the number of iterations along the x-axis, bottom of the screen. Vertically from top to bottom, we have plotted the following: missing data (above the time axis); rough (i.e. raw) data, with separate x- and y-coordinates for left and right eyes; smoothed data (with interpolated sections marked in red); fixations (tentative fixations marked as a thick green line, and fixations that have passed all of our verification checks marked in orange); bad data flags (absent in this sample); and velocity (calculated from raw data in black and from smoothed data in red). The velocity threshold of 35ºs-1 is drawn as a thin blue line. On the right hand side of the screen, the participant and trial numbers are drawn (in this case ‘4011_23’), along with the stimulus the infant was viewing, with their gaze positions (left eye, blue; right eye, red) drawn on the screen. The letters a–d below the bottom axis refer to examples we describe in the textCC BY · PubMed Central
Fig. 8 Same layout as that in Fig. 7. This is a sample of lower quality than that shown in Fig. 7. Instances in which our false fixation and saccade detection flags have fired are drawn as colored squares above the velocity plot. Tentatively identified fixations are marked as green bars; those fixations that have passed all our post hoc validity checks and have been stored as reliable fixations are marked as orange bars
Fig. 8 Same layout as that in Fig. 7. This is a sample of lower quality than that shown in Fig. 7. Instances in which our false fixation and saccade detection flags have fired are drawn as colored squares above the velocity plot. Tentatively identified fixations are marked as green bars; those fixations that have passed all our post hoc validity checks and have been stored as reliable fixations are marked as orange barsCC BY · PubMed Central
Fig. 9 Frequency distributions for fixation durations obtained from three different age groups, comparing the results obtained from a standard dispersal algorithm with those obtained from our algorithms. a Six-month-olds. b Twelve-month-olds. c) Adults. The averages of the fixations obtained from each age group are shown in Table 1
Fig. 9 Frequency distributions for fixation durations obtained from three different age groups, comparing the results obtained from a standard dispersal algorithm with those obtained from our algorithms. a Six-month-olds. b Twelve-month-olds. c) Adults. The averages of the fixations obtained from each age group are shown in Table 1CC BY · PubMed Central
Fig. 10 Quantifying how fixation duration varies as a function of data quality. The relationship between fixation durations as analyzed using our new algorithms and data quality, evaluated in two ways: a flicker, the mean duration of raw data segments, and b precision, the variance in reported point of gaze (POG). This recreates the analyses shown in Fig. 3
Fig. 10 Quantifying how fixation duration varies as a function of data quality. The relationship between fixation durations as analyzed using our new algorithms and data quality, evaluated in two ways: a flicker, the mean duration of raw data segments, and b precision, the variance in reported point of gaze (POG). This recreates the analyses shown in Fig. 3CC BY · PubMed Central
Fig. 11 Quantifying performance of our algorithms relative to hand-coding as a function of data quality. Individual data-points show the results from individual participants whose data have been parsed using different processing techniques. a Relationship between flicker and proportion agreement between automatic and hand-coding. b Relationship between precision and proportion agreement between automatic and hand-coding. c Relationship between flicker and mean fixation duration as processed using both hand-coding and our algorithms. d Relationship between precision and mean fixation duration as processed using both hand-coding and our algorithms. This recreates the analyses shown in Fig. 4
Fig. 11 Quantifying performance of our algorithms relative to hand-coding as a function of data quality. Individual data-points show the results from individual participants whose data have been parsed using different processing techniques. a Relationship between flicker and proportion agreement between automatic and hand-coding. b Relationship between precision and proportion agreement between automatic and hand-coding. c Relationship between flicker and mean fixation duration as processed using both hand-coding and our algorithms. d Relationship between precision and mean fixation duration as processed using both hand-coding and our algorithms. This recreates the analyses shown in Fig. 4CC BY · PubMed Central
Fig. 12 Evaluating the performance of our algorithm using simulations of the effect of flicker and precision. We reprocessed the same data shown in Fig. 5, but using our own algorithms rather than the standard dispersal algorithms. a Repeats the analysis shown in Fig. 5d). b Repeats the analysis shown in Fig. 5e. c Repeats the analysis shown in Fig. 5f
Fig. 12 Evaluating the performance of our algorithm using simulations of the effect of flicker and precision. We reprocessed the same data shown in Fig. 5, but using our own algorithms rather than the standard dispersal algorithms. a Repeats the analysis shown in Fig. 5d). b Repeats the analysis shown in Fig. 5e. c Repeats the analysis shown in Fig. 5fCC BY · PubMed Central
Fig. 13 Frequency distributions of fixation durations of high-quality adult viewing data from two participants viewing the same stimuli on two eyetrackers. Viewing data have then been processed twice: first, using the manufacturer-supplied fixation detection algorithms, and second, using the new algorithms we present here. a Eyelink 1000, participant 1. b Eyelink 1000, participant 2. c Tobii TX300, participant 1. d Tobii TX300, participant 2
Fig. 13 Frequency distributions of fixation durations of high-quality adult viewing data from two participants viewing the same stimuli on two eyetrackers. Viewing data have then been processed twice: first, using the manufacturer-supplied fixation detection algorithms, and second, using the new algorithms we present here. a Eyelink 1000, participant 1. b Eyelink 1000, participant 2. c Tobii TX300, participant 1. d Tobii TX300, participant 2CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 91 references

Research works printed but not linked: 4

  1. Dekker, T.M., Smith, T.J., Mital, P.K. & Karmiloff-Smith, A. (under review). The debate on childhood screen exposure: An in-depth eye-tracking comparison of infant DVDs. Manuscript under review.
  2. Scerif, G., Cornish, K., Wilding, J., Driver, J., & Karmiloff-Smith, A. (2005). Early attention difficulties in fragile X syndrome: Underlying mechanisms and developmental trajectories. Journal of Cognitive Neuroscience, 220-221.
  3. Shic, F., Chawarska, K. and Scassellati, B. (2008). The incomplete fixation measure, Proceedings of the 2008 symposium on Eye tracking research & applications, Savannah, Georgia: ACM, 2008, pp. 111-114.
  4. Wass, S. V. (2011). Gaze-contingent concentration training for infants. PhD Thesis. Birkbeck, University of London.

Cited by

See all 154

Metadata

Authors
Wass, S. V., Smith, T. J., & Johnson, M. H.
Affiliations
University of London (GB)
Corresponding author
Wass, S. V.
Venue
Behavior Research Methods
Published
2012-09-06
Volume
45
Issue
1
Pages
229-250
ISSN
  • 1554-351X (ISSN-L)
  • 1554-3528
OpenAlex
W2072416967
Cited by
154
References
91
Citations by year
2026: 11 · 2025: 10 · 2024: 5 · 2023: 9 · 2022: 16 · 2021: 13 · 2020: 14 · 2019: 18 · 2018: 19 · 2017: 10 · 2016: 4 · 2015: 7 · 2014: 15 · 2013: 2
FWCI
14.10
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Behavior Research Methods (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • BIROn (Birkbeck, University of London) (acceptedVersion)
  • University of the Arts London Research Online (University of the Arts London) (acceptedVersion)
Topics
  • Physical Sciences › Computer Science › Human-Computer Interaction › Gaze Tracking and Assistive Technology
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Visual Attention and Saliency Detection
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Visual perception and processing mechanisms
Keywords
fixation duration, data quality, artifact removal, gaze direction, attention allocation, interindividual variability, reliability statistics
MeSH terms
Adult, Algorithms, Analysis of Variance, Attention, Differential Threshold, False Positive Reactions, Fixation, Ocular, Humans, Infant, Models, Biological, Saccades, Software Design, Time Factors, Reproducibility of Results, Artifacts, Eye Movement Measurements

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Parsing eye-tracking data of variable quality to provide accurate fixation duration estimates in infants and adults. https://app.cochat.ai/papers/10.3758/s13428-012-0245-6
The paper
Wass, S. V., Smith, T. J., & Johnson, M. H. (2013). Parsing eye-tracking data of variable quality to provide accurate fixation duration estimates in infants and adults. Behavior Research Methods. https://doi.org/10.3758/s13428-012-0245-6

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.