Fig 1.
Off-axis pupil diameter ratio.
Plot based on model given by Mathur et al. [18]
Fig 2.
Illustration of microsaccade detection in a single participant’s gaze recording during calibration.
The small dots in (a) show gaze points prior to processing, i.e., detection of fixations or of microsaccades. Following processing, microsaccades in (b) are highlighted by larger, darker dots and thicker connecting segments. Notice that all small, raw gaze points that were not part of a fixation have now been removed. Thus any remaining small dots are members of fixation clusters that may or may not contain microsaccades. The point of this illustration is to show that microsaccades are detectable at locations far beyond the central fixation point. In the replication of Siegenthaler et al.’s [19] study, however, the participant’s gaze was held fixed at a central fixation point, hence the microsaccades depicted here during calibration were not used in the analysis of the experiment.
Fig 3.
Time course of each experimental trial.
Note that because we wanted to present the experimental procedure as accurately as possible the fixation point and text entry text are hardly visible on this figure due to its size. In the actual experimental settings both were clearly visible for participants on the computer screen. The actual fixation point was defined as a circle with radius of 5 pixels and color defined as [0.2,0.2,0.2] in the RGB color scale. That color can be described as an increment of 0.2 above the gray background. Note that each value in the RGB space in PsychoPy is defined by the range between -1 and 1. The same color was applied to the text on the “Answer entry” screen.
Table 1.
Internal consistency of cognitive load measures in response to task difficulty.
The table presents Cronbach’s α overall and within each task. The reliability coefficient could not be calculated for the BCPD measure in the Control tasks as it was treated as the baseline. Note the meaning of abbreviations in the table: MM (Microsaccades Magnitude), MR (Microsaccade Rate), PV-M (Microsaccade Peak Velocity—Magnitude), CPD (Intra-Trial Change in Pupil Dilation), BCPD (Inter-Trial Change in Pupil Dilation).
Fig 4.
Manipulation check and questionnaire data.
Results on response accuracy and subjective evaluation of effort. Means are plotted on trial type and Time-On-Task. The error bars represent ± 1 SE for the means.
Table 2.
Dependent variables’ descriptive statistics overall and for each task: Means and standard errors (SE).
Note the meaning of abbreviations in the table: CA (Correct Answers), MM (Microsaccades Magnitude), MR (Microsaccade Rate), MPV-M (Microsaccade Peak Velocity—Magnitude), CPD (Intra-Trial Change in Pupil Dilation), BCPD (Inter-Trial Change in Pupil Dilation).
Fig 5.
Microsaccadic peak velocity vs. magnitude (main sequence).
Main panel: the scatter plot represents detected microsaccades with amplitude indicated on the abscissa and peak velocity on the ordinate. The line drawn through is a linear fit through the scatter plot. Bottom and side panels: microsaccade amplitude and peak velocity distributions (N = 13 subjects).
Fig 6.
Microsaccade magnitude, rate in response to task difficulty and Time-On-Task.
Means are plotted for trial type versus Time-On-Task. Error bars represents ± 1 SE for the means.
Fig 7.
Pupil response to task difficulty and Time-On-Task.
Plots show mean change in pupil diameter versus Time-On-Task; error bars represent ± 1 SE.
Table 3.
Regression coefficients (β) and standard errors (in parentheses) of two multinomial logistic regression analyses.
In the first (Difficult vs. Control) and the second (Easy vs. Control) analysis, Control task served as reference. The intercept coefficient for the Difficult task was not statistically significant, β0 = 0.39, SE = 0.26, z = 1.49, p > 0.05, but the intercept was significantly different from zero for the Easy task, β0 = 0.65, SE = 0.25, z = 2.63, p < 0.01. Statistical significance of all β coefficients was checked with a Wald z-test; p-values are marked with asterisks (**p < 0.001, *p < 0.05).