How task demands influence scanpath similarity in a sequential number-search task

Richard Dewhurst, Tom Foulsham, H.M. Jarodzka, Roger Johansson, Kenneth Holmqvist, Marcus Nyström

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

More and more researchers are considering the omnibus eye movement sequence—the scanpath—in their studies of visual and cognitive processing (e.g. Hayes, Petrov, & Sederberg, 2011; Madsen, Larson, Loschky, & Rebello, 2012; Ni et al., 2011; von der Malsburg & Vasishth, 2011). However, it remains unclear how recent methods for comparing scanpaths perform in experiments producing variable scanpaths, and whether these methods supplement more traditional analyses of individual oculomotor statistics. We address this problem for MultiMatch (Jarodzka et al., 2010; Dewhurst et al., 2012), evaluating its performance with a visual search-like task in which participants must fixate a series of target numbers in a prescribed order. This task should produce predictable sequences of fixations and thus provide a testing ground for scanpath measures. Task difficulty was manipulated by making the targets more or less visible through changes in font and the presence of distractors or visual noise. These changes in task demands led to slower search and more fixations. Importantly, they also resulted in a reduction in the between-subjects scanpath similarity, demonstrating that participants’ gaze patterns became more heterogenous in terms of saccade length and angle, and fixation position. This implies a divergent strategy or random component to eye-movement behaviour which increases as the task becomes more difficult. Interestingly, the duration of fixations along aligned vectors showed the opposite pattern, becoming more similar between observers in 2 of the 3 difficulty manipulations. This provides important information for vision scientists who may wish to use scanpath metrics to quantify variations in gaze across a spectrum of perceptual and cognitive tasks.
Original languageEnglish
Pages (from-to)9-23
Number of pages15
JournalVision Research
Volume149
DOIs
Publication statusPublished - Aug 2018

Fingerprint

Eye Movements
Saccades
Noise
Research Personnel

Keywords

  • Scanpaths
  • Eye movements
  • MultiMatch
  • search

Cite this

Dewhurst, Richard ; Foulsham, Tom ; Jarodzka, H.M. ; Johansson, Roger ; Holmqvist, Kenneth ; Nyström, Marcus. / How task demands influence scanpath similarity in a sequential number-search task. In: Vision Research. 2018 ; Vol. 149. pp. 9-23.
@article{b95307d0fbd046e589f319f40640fcbc,
title = "How task demands influence scanpath similarity in a sequential number-search task",
abstract = "More and more researchers are considering the omnibus eye movement sequence—the scanpath—in their studies of visual and cognitive processing (e.g. Hayes, Petrov, & Sederberg, 2011; Madsen, Larson, Loschky, & Rebello, 2012; Ni et al., 2011; von der Malsburg & Vasishth, 2011). However, it remains unclear how recent methods for comparing scanpaths perform in experiments producing variable scanpaths, and whether these methods supplement more traditional analyses of individual oculomotor statistics. We address this problem for MultiMatch (Jarodzka et al., 2010; Dewhurst et al., 2012), evaluating its performance with a visual search-like task in which participants must fixate a series of target numbers in a prescribed order. This task should produce predictable sequences of fixations and thus provide a testing ground for scanpath measures. Task difficulty was manipulated by making the targets more or less visible through changes in font and the presence of distractors or visual noise. These changes in task demands led to slower search and more fixations. Importantly, they also resulted in a reduction in the between-subjects scanpath similarity, demonstrating that participants’ gaze patterns became more heterogenous in terms of saccade length and angle, and fixation position. This implies a divergent strategy or random component to eye-movement behaviour which increases as the task becomes more difficult. Interestingly, the duration of fixations along aligned vectors showed the opposite pattern, becoming more similar between observers in 2 of the 3 difficulty manipulations. This provides important information for vision scientists who may wish to use scanpath metrics to quantify variations in gaze across a spectrum of perceptual and cognitive tasks.",
keywords = "Scanpaths, Eye movements, MultiMatch, search",
author = "Richard Dewhurst and Tom Foulsham and H.M. Jarodzka and Roger Johansson and Kenneth Holmqvist and Marcus Nystr{\"o}m",
year = "2018",
month = "8",
doi = "10.1016/j.visres.2018.05.006",
language = "English",
volume = "149",
pages = "9--23",
journal = "Vision Research",
issn = "0042-6989",
publisher = "Elsevier Ltd",

}

How task demands influence scanpath similarity in a sequential number-search task. / Dewhurst, Richard; Foulsham, Tom; Jarodzka, H.M.; Johansson, Roger; Holmqvist, Kenneth; Nyström, Marcus.

In: Vision Research, Vol. 149, 08.2018, p. 9-23.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - How task demands influence scanpath similarity in a sequential number-search task

AU - Dewhurst, Richard

AU - Foulsham, Tom

AU - Jarodzka, H.M.

AU - Johansson, Roger

AU - Holmqvist, Kenneth

AU - Nyström, Marcus

PY - 2018/8

Y1 - 2018/8

N2 - More and more researchers are considering the omnibus eye movement sequence—the scanpath—in their studies of visual and cognitive processing (e.g. Hayes, Petrov, & Sederberg, 2011; Madsen, Larson, Loschky, & Rebello, 2012; Ni et al., 2011; von der Malsburg & Vasishth, 2011). However, it remains unclear how recent methods for comparing scanpaths perform in experiments producing variable scanpaths, and whether these methods supplement more traditional analyses of individual oculomotor statistics. We address this problem for MultiMatch (Jarodzka et al., 2010; Dewhurst et al., 2012), evaluating its performance with a visual search-like task in which participants must fixate a series of target numbers in a prescribed order. This task should produce predictable sequences of fixations and thus provide a testing ground for scanpath measures. Task difficulty was manipulated by making the targets more or less visible through changes in font and the presence of distractors or visual noise. These changes in task demands led to slower search and more fixations. Importantly, they also resulted in a reduction in the between-subjects scanpath similarity, demonstrating that participants’ gaze patterns became more heterogenous in terms of saccade length and angle, and fixation position. This implies a divergent strategy or random component to eye-movement behaviour which increases as the task becomes more difficult. Interestingly, the duration of fixations along aligned vectors showed the opposite pattern, becoming more similar between observers in 2 of the 3 difficulty manipulations. This provides important information for vision scientists who may wish to use scanpath metrics to quantify variations in gaze across a spectrum of perceptual and cognitive tasks.

AB - More and more researchers are considering the omnibus eye movement sequence—the scanpath—in their studies of visual and cognitive processing (e.g. Hayes, Petrov, & Sederberg, 2011; Madsen, Larson, Loschky, & Rebello, 2012; Ni et al., 2011; von der Malsburg & Vasishth, 2011). However, it remains unclear how recent methods for comparing scanpaths perform in experiments producing variable scanpaths, and whether these methods supplement more traditional analyses of individual oculomotor statistics. We address this problem for MultiMatch (Jarodzka et al., 2010; Dewhurst et al., 2012), evaluating its performance with a visual search-like task in which participants must fixate a series of target numbers in a prescribed order. This task should produce predictable sequences of fixations and thus provide a testing ground for scanpath measures. Task difficulty was manipulated by making the targets more or less visible through changes in font and the presence of distractors or visual noise. These changes in task demands led to slower search and more fixations. Importantly, they also resulted in a reduction in the between-subjects scanpath similarity, demonstrating that participants’ gaze patterns became more heterogenous in terms of saccade length and angle, and fixation position. This implies a divergent strategy or random component to eye-movement behaviour which increases as the task becomes more difficult. Interestingly, the duration of fixations along aligned vectors showed the opposite pattern, becoming more similar between observers in 2 of the 3 difficulty manipulations. This provides important information for vision scientists who may wish to use scanpath metrics to quantify variations in gaze across a spectrum of perceptual and cognitive tasks.

KW - Scanpaths

KW - Eye movements

KW - MultiMatch

KW - search

U2 - 10.1016/j.visres.2018.05.006

DO - 10.1016/j.visres.2018.05.006

M3 - Article

VL - 149

SP - 9

EP - 23

JO - Vision Research

JF - Vision Research

SN - 0042-6989

ER -