Twenty participants completed the study. No significant differences were found between the age and gender parameters of the sample population.
Data analysis
In this study, data processing involved several key steps. First, we pre-processed the eye-tracking data by removing outliers and correcting for any calibration errors in the measurements. The analysis of eye-tracking metrics (e.g., gaze duration, fixation count, saccadic movements) was performed using Tobii Pro Lab software, which provides real-time data for these parameters during the cognitive tasks. Descriptive statistics, including means and standard deviations, were calculated for each of the eye-tracking metrics across the tasks. Inferential statistics, including correlation analysis and regression models, were used to assess relationships between eye-tracking metrics, learning styles, and academic performance (GPA) (Table 1). All analyses were conducted using SPSS 24.0 (SPSS, Chicago, IL, USA), with a significance level set at p < 0.05.
The participants’ visual learning preference scores were calculated, and the visual dimensions (strong, moderate and neutral) were assigned. The eye-tracking metrics for each participant were then aggregated by category, and correlation analyses were conducted to explore the relationship between the learning style dimensions and gaze behaviors (Figs. 2 and 3).
Comparison of (a) Pupil Size (mm) and (b) Task Duration (s) across Tasks
Achievement based and visual ILS dimension distribution of participants
Descriptive statistics
Descriptive statistics were first computed to provide an overview of the participants’ characteristics and the distribution of key variables. Table 2 presents the means, standard deviations, and ranges for the key eye-tracking metrics (gaze duration, fixation count, and saccadic movements) across all tasks, as well as for the learning styles and GPA variables.
The average fixation duration across all tasks was 350 ms (SD = 25 ms), with saccadic movements accounting for 6% of the total gaze time. The pupil diameter varied slightly, with an average diameter of 3.2 mm (SD = 0.3 mm). Participants showed a predominant preference for visual learning, with 70% categorized as strong visual learners. The remaining 30% exhibited moderate or verbal preferences. The average GPA of the participants was 3.1 (SD = 0.4), with a range from 2.4 to 3.8.
There was a moderate preference among 33% and a strong preference for visual learning among 56% of the female students, which indicated a right-skewed distribution. None of the female participants had neutral preference (0%) and 1 participant (11%) had verbal learning preference. In contrast, 27% of the male students had neutral, 45% had moderate, and 9% had strong preference for visual learning. Two participants (18%) had verbal learning preference.
The raw eye-tracking data were processed using Tobii Pro Lab (Analyzer Edition). Fixations were defined as any gaze point held for at least 50 ms and saccades were defined as rapid eye movements between fixations. The fixation percentage, average pupil diameter, task duration and number of total eye movement were extracted from the collected data. Average fixation (94%) and saccadic movement (6%) percentages were identical during each task (Fig. 2).
The analysis of pupil diameter and fixation percentage across four cognitive tasks revealed no significant differences in these parameters. The lack of variation in pupil diameter across tasks suggested that the cognitive demands of each test elicited comparable levels of mental effort. The consistency of fixation percentages across the four tasks indicated that participants allocated their visual attention similarly, regardless of the task type (Fig. 3).
Figure 4 illustrated the distribution of participants categorized by their Felder-Soloman Learning Style Index (ILS) dimension on the visual-verbal spectrum, cumulative GPA, and achievement-based groups (Explorer, Achiever, Scholar). Additionally, gender differences were visualized with separate markers for male and female participants. Seventeen participants fell within the visual learning spectrum, particularly in the moderate and strong visual learning ranges (values ≥ + 4). Three participants (1 Female, 2 Male) exhibited verbal learning tendencies (values ≤ -4). Female participants dominated the strong visual learner category, particularly in the higher GPA groups (Achiever and Scholar). Male participants were distributed more evenly across the GPA categories but exhibited fewer instances of strong visual preference. In the Explorer group, visual learning preferences were distributed across moderate and strong categories. Among Achievers, participants showed a concentration in moderate visual learning preference. The Scholar group had fewer participants, predominantly with strong visual learning preference.
Comparison of pupil diameter and fixation percentage across four cognitive tasks (a) TMT-A, b TMT-B, c VSAT, and d Stroop test
Regression analysis results indicated that pupil size and fixation percentage are predictors of cognitive engagement so that both pupil size (coefficient = 0.45) and fixation percentage (coefficient = 0.67) were significant predictors of cognitive engagement, as reflected in the statistical significance (p-values < 0.05) of the model.
Hypothesis testing
Visual Learning Preference and Eye-Tracking Metrics: This hypothesis tested whether the students with strong visual learning preferences exhibit longer fixation durations and more frequent saccadic movements compared to those with verbal or moderate preferences. The results revealed that the visual learners had an average fixation duration of 375 ms (SD = 28 ms), while verbal learners had a shorter average of 320 ms (SD = 20 ms). A one-way ANOVA was conducted to compare fixation durations across learning style groups (strong visual, moderate visual, verbal). The results indicated a significant main effect of learning style on fixation duration, f (2, 18) = 5.32, p = 0.013. Post-hoc comparisons using Tukey’s HSD test revealed that visual learners had significantly longer fixation durations than verbal learners (p = 0.04). No significant differences were found between moderate and strong visual learners (p = 0.65).
Visual Learning Preference and Academic Performance: This hypothesis tested whether the visual learners would have higher GPAs compared to verbal learners, particularly among female students. The mean GPA for visual learners was 3.3 (SD = 0.3), while the mean GPA for verbal learners was 2.8 (SD = 0.5). A t-test was conducted to compare GPAs between visual and verbal learners. The results indicated a significant difference in GPA, t (18) = 2.41, p = 0.026, with visual learners achieving higher GPAs than verbal learners. Gender differences were examined within each learning style, but no significant interaction between gender and learning style on GPA was found (p = 0.12).
Cognitive Load Across Tasks: This hypothesis tested whether the cognitive demands of different tasks would not significantly alter the relationship between learning style and cognitive task performance and the visual learners would show consistent eye-tracking metrics (e.g., fixation percentage) across various tasks, indicating stable cognitive processing strategies. Across tasks, visual learners maintained an average fixation percentage of 94%, with no significant variation across TMT-A, TMT-B, VSAT, and Stroop tests. A repeated-measures ANOVA was conducted to examine the differences in eye-tracking metrics (fixation percentage and pupil diameter) across the four tasks for visual learners. The results showed no significant differences in fixation percentage across tasks, f (3, 21) = 1.45, p = 0.25, suggesting that the cognitive processing strategies remained consistent across tasks.
link