Distracted driving has been widely acknowledged as a leading cause of road accidents, and in-vehicle technology, particularly touchscreens, has intensified the issue. As these devices become more embedded in vehicle design, understanding their influence on driver behaviour and performance becomes essential for improving road safety (Bassani et al., 2023). An in-depth analysis of fatal road crashes in Norway reported that driver inattention to the primary driving task was a contributing factor in approximately 29% of all fatal accidents, with mobile phone distraction accounting for 2–4% and other in-vehicle secondary tasks contributing to about 8% of fatal crashes (Sundfør et al., 2019). While official accident statistics rarely distinguish between touchscreen interfaces and traditional physical controls, experimental and naturalistic studies consistently show that visually demanding in-vehicle interactions, particularly those requiring prolonged glances away from the road, are associated with elevated crash risk (Li et al., 2024; Mehrotra et al., 2023). In this regard, a considerable amount of research has been conducted to examine the impact of distraction due to secondary tasks on drivers’ visual attention and eye behaviour. With high risk associated with distraction condition, most researchers prefer driving simulator studies to maintain safe driving environment. Grahn and Kujala (2020) examined the influence of touchscreen size, user interface design, and subtask boundaries on the visual demand and distraction associated with in-vehicle secondary tasks. Through driving simulator experiments, the authors identified that while larger screens marginally decrease glance durations, the design of the user interface plays a more substantial role. Applications specifically designed for automotive use, particularly those incorporating features like speech-to-text and read-aloud functionality, were found to significantly reduce visual distraction compared to conventional smartphone applications. These results from Grahn and Kujala (2020) highlight the importance of well-designed task structures and interface elements. In another study, Kim and Song (2014) investigated the usability and safety of various touch gestures, such as flicking, panning, and pinching, in comparison to tapping for IVIS during driving. Their findings, based on the visual occlusion method, indicated that panning is both safe and practical for driving-related tasks. In contrast, flicking required refinement to minimize distraction, while pinching was associated with elevated visual demands and wrist discomfort, underscoring potential ergonomic challenges. Ebel et al. (2023) examined visual demand during touchscreen interactions in vehicles. The author identified key factors such as design elements, driving conditions, and specific scenarios that influence how drivers allocate their visual attention. The study highlighted the roles of interaction type, vehicle speed, and levels of driving automation in contributing to driver distraction. Ezzati Amini et al. (2023) found the impact of mobile phone distraction, specifically writing and reading text messages, on driver behaviour using a car driving simulator. The research examines driver responses under three scenarios: normal driving, intervention with fixed timing, and distraction-based interventions. Eye-tracking glasses revealed significant shifts in drivers' gaze patterns towards the intervention display during distraction. Key findings highlight increased lateral positioning deviation and reduced longitudinal acceleration during pedestrian collision risks and non-critical events under distraction. Ban and Park (2024) investigated the influence of nondriving-related task (NDRT) touchscreen location and task difficulty on driver performance, eye gaze behaviour, and workload during SAE Level 3 automated driving. Their driving simulator study found that touchscreen location significantly affected both NDRT and take-over task performance, with the lower right location being more effective for NDRT tasks but less suitable for take-over tasks. Moreover, task difficulty impacted the perceived workload during take-over scenarios. These findings offer valuable guidance for designing touchscreen interfaces and NDRT tasks in conditionally automated vehicles.

Beyond interface design and interaction characteristics, increasing attention has also been directed toward the role of driver training and familiarisation in mitigating distraction and supporting safer engagement with increasingly complex in-vehicle technologies. Zhang et al. (2025) compared three driver training formats: (i) conventional text-based instructions using owners’ manual, (ii) knowledge-based materials with visual aids, and (iii) simulator-based hands-on practice, to assess drivers’ understanding and safe engagement with vehicle automation. The authors concluded that all training types of significantly improved system comprehension, but the knowledge-based format produced the largest gains in both understanding, safe usage and more frequent use of Advanced driver assistance systems. Similarly, Murtaza et al. (2024) compared video-based and text-based (manual) training methods using driving simulator to evaluate user performance of driver assistance systems. Authors highlighted that video-based training leads to higher accuracy and faster reaction times compared with manual-based learning. Moreover, participants trained with videos also displayed more consistent responses, indicating stronger mental models. In another study, Krampell et al. (2020) evaluated a short, scenario-based training intervention designed to instigate more accurate mental models human-vehicle automation interaction. Participants who received short training showed significant improvements in system comprehension, clearer distinctions between driver vs. automation roles, and more appropriate supervision behaviour, indicating that targeted training can enhance users’ readiness for automated driving. Zahabi et al. (2021) evaluated two training approaches, video-based and demonstration-based using a driving simulator with 20 older adults. The findings indicated that demonstration-based training was more effective in reducing perceived workload, whereas video-based training showed a slight advantage for female participants by lowering cognitive load and enhancing knowledge acquisition. Despite these training-specific effects, both approaches led to similar increases in trust in vehicle automation, with no statistically significant differences between the two methods. In simulator-based study, Pradhan et al. (2023) examined three training approaches (i) visualization, (ii) text-based user manual, and (iii) Sham (control) training with 24 novice ADAS users. The results demonstrated significant improvements in mental model scores following training, particularly for the visualization and manual groups, whereas the sham training showed no effect. However, despite these gains in conceptual understanding, training did not lead to significant improvements in real-time response accuracy or reaction times. These findings suggest that while training can enhance user knowledge, such improvements may not immediately translate into more effective system interaction during driving. Noble et al. (2019) compared manual-based and interactive multimedia training for vehicle autopilot use and found minimal differences in knowledge and no significant effects on driver behaviour or attitudes. Familiarity increased mainly through hands-on experience rather than brief training. Eye-glance analysis showed longer off-road glances during automation, particularly among younger drivers, leading the authors to conclude that experience is more critical than short training for safe system use.

In general, previous studies have highlighted the critical role of visual distraction and training in connection with driver behaviour and performance. The present study examines the potential effects of an instructor-led short pre-drive training session on touchscreen interface usage and its possible influence on drivers’ visual attention. An earlier study by Hazoor et al. (2026) examined the same experimental context using conventional driving-performance metrics such as mean speed, standard deviation of speed, lane position, and standard deviation of lateral position. That work concluded that short pre-drive training did not significantly influence these performance measures. While those findings provided valuable insights into longitudinal and lateral control under touchscreen interaction, they did not capture the underlying visual behaviour that drives such performance outcomes. The present study extends this research question by analysing eye-tracking data to explore drivers’ visual attention patterns during touchscreen use. This complementary perspective is critical because visual distraction is a primary mechanism through which touchscreen interactions affect safety. By focusing on gaze behaviour, this study aims to uncover whether short pre-drive training influences drivers’ ability to allocate attention efficiently, thereby providing deeper understanding beyond what driving-performance measures alone can reveal.

This study aims to examine the effectiveness of an instructor-led short pre-drive training session on the use of in-vehicle touchscreen in reducing driver distraction. Using a driving simulator combined with eye-tracking technology, the study seeks to evaluate whether such short training enhances drivers’ visual attention and mitigates distraction. The research question guiding this investigation is: Does a pre-drive short training session improve drivers’ visual attention and reduce distraction when interacting with in-vehicle touchscreen?

This study recruited 60 licensed drivers (24 females) from Norway, who volunteered via a social media platform (Facebook). Participants provided background information, including age, total years of driving experience, estimated annual driving distance, and any history of road traffic crashes. Eligibility criteria required participants to have normal vision or corrected-to-normal vision. To prevent bias, participants were not informed about the study's underlying hypothesis. Ethical clearance was obtained from the Norwegian Agency for Shared Services in Education and Research (SIKT), with approval documented under reference number 951733. Participants ranged in age from 23 to 55 years. Table 1 presents the demographic characteristics of the two groups, which were evenly divided (30 participants each) and balanced with respect to age, sex, and driving experience.

The experimental study was conducted using a fixed-base driving simulator at the TRAFIKKLAB, Nord University, Norway. The simulator, known as SimEASY, is operated via the SCANeR Studio™ platform developed by AV Simulation. The hardware setup features a driver cockpit including an automatic transmission, steering wheel, pedals, and an adjustable seat to replicate standard vehicle ergonomics. To simulate realistic driving sensations, the simulator is equipped with active force-feedback steering wheel, replicating tactile feedback such as wheel resistance and surface impacts. To immerse participants in a realistic driving environment, the simulation visuals were displayed across three 43-inch full HD LED monitors (1920 x 1080 resolution), arranged to provide a 140⁰ horizontal 27⁰ vertical field of view, operating at a refresh rate of 60 Hz. The visual display accurately replicates the interior of a car, including a functional dashboard and centre and side rear-view mirrors. A high-fidelity surround sound system generated environmental audio, including engine noise, traffic sounds, wind, and ambient background noise. Additionally, a seat-mounted vibration unit simulated vehicle vibrations and pavement roughness (Figure 1).

A 12-inch touchscreen interface was mounted on the right side of the steering wheel to perform the secondary tasks. This screen was placed 13 inches from the steering wheel centre to its outer edge, similar location of infotainment display as in actual vehicles. As shown in Figure 1, this setup allows easy access without requiring excessive arm extension (Ban & Park, 2024). The touchscreen simulated an in-vehicle infotainment system and offered a range of functions, including media playback, navigation, climate control, lighting adjustments, seat heating, and various other features typical of modern car dashboards. The interface design was adapted from an openly accessible online source to resemble current automotive user interface designs (Ebube, 2024). It is important to note that none of the participants had prior familiarity with the touchscreen interface employed in this study.

The simulated motorway alignment (Figure 2) used in this study was designed based on the Norwegian standards for geometric highway design (Statens Vegvesen, 2023). The roadway featured two lanes in each direction, each 3.5 m wide, with 1.5 m right shoulder, a 0.75 m left shoulder, and a 2 m central median. Safety barriers with a height 0.75 m were installed along the median and right roadside. The road signage and pavement markings adhered to the specifications outlined in the Norwegian Highway Code (Statens Vegvesen, 2021), with a posted speed limit of 90 km/h.

Participants began their drive from a lay-by area, after which they merged onto the main motorway via an entry ramp. The route consisted of three main motorway sections, incorporating two large horizontal curves (radius = 1000 m). Short tangents and low-radius curves were inserted between major sections to disrupt behavioural carryover. Data collected from these transitional sections were excluded from the analysis.

To support realistic curve negotiation and ensure smooth transitions between straight and curved roadway sections, spiral transition curves were integrated into the alignment design. The total length of the test route was limited to 13 kilometres, limiting the driving session under 20 minutes, reducing the likelihood of simulator-induced discomfort and sickness (Philip et al., 2003).

The scenario was intentionally designed without interaction with surrounding vehicles or traffic events (e.g., intersections or traffic lights). The level of service was set to category A (free-flow traffic) to ensure highly controlled driving conditions across all participants. This approach minimized variability in visual workload from traffic negotiation, thereby allowing the study to isolate the specific effects of touchscreen interaction and pre-drive training on eye-movement behaviour (Board et al., 2022). The use of curves and straight segments provided continuous but stable driving demands sufficient to maintain lane-keeping and speed-control tasks without introducing confounding safety-critical events.

Eye-tracking was employed to quantify drivers’ visual attention during secondary task performance and to examine how short pre-drive training affected interaction with the in-vehicle touchscreen. For each distraction event, following core components were considered: Time of Interest (TOI), which defines the period of task engagement; Areas of Interest (AOI), which specify the visual regions analysed; and eye-movement metrics including Fixation Count (FC), Total Fixation Duration (TFD), and the Conditional Probability Matrix (CPM), which together provide a comprehensive assessment of visual attention allocation and gaze transitions.

The eye movement data were recorded using Argus science wearable eye-tracker (Figure 3a), the ET-Vision software for data recording and the ET-Analysis software of the Argus Science was used to analyse the recorded gaze data. In this study, the eye movement data were analysed during all distraction events along road tangents (straight sections), i.e., use of in-vehicle touchscreen.

Upon arrival at the lab, participants were introduced to the objectives and procedures of the study, along with a brief overview of the driving simulator and its key components. The process followed a structured protocol to ensure consistency and ease of participation throughout the experiment.

Initial Briefing:

Familiarization Drive:

Preparation for the Experimental Drive:

Main Driving Session:

In addition to the simulator sickness questionnaire and feedback on provided training, subjective workload assessments (NASA Task Load Index) and objective driving performance variables (i.e., speed and lateral lane position variation) were also collected during the experiment (Hazoor et al., 2026). However, the present article focuses on eye-tracking based measures of visual attention to maintain a clear scope aligned with the primary research question.

This study examined the effect of short pre-drive training on drivers’ gaze behaviour during interaction with an in-vehicle touchscreen interface. Visual attention metrics, including FC, TFD, and conditional gaze transition probabilities across predefined AOIs, were compared between two participant groups: trained and untrained drivers. All participants completed the same set of experimental tasks (Table 2).

The analysis was restricted to tangent (straight) road segments, while data collected during curved road sections (Tasks 3a and 3b) were excluded because the limited number of observations did not allow for a reliable comparison between straight and curved roadway conditions. Group differences in gaze behaviour metrics were evaluated using one-way analysis of variance (ANOVA). Statistical significance was assessed at a 95% confidence level (p < 0.05).

Table 3 presents the FC for each AOI, comparing Group 1 (untrained participants) and Group 2 (trained participants) with respect to secondary tasks. Overall, both groups showed the highest FC on the touchscreen, except for the undistracted (UD) driving condition, followed by the road AOI during secondary tasks.

For the Road AOI, both groups exhibited similar patterns, without any significant difference among two groups during tasks involving touchscreen interaction. In the case of undistracted condition, number of fixations were higher for group 1 (UD: G1 = 42.18 vs. G2 = 37.37).

For the Touchscreen AOI, participants in both groups demonstrated substantial visual engagement during touchscreen-related tasks (e.g., Task 1a, 1b, and 1c). Group 1 recorded slightly higher FC values during these tasks (e.g., Task 1a: G1 = 33.71 vs. G2 = 30.94); however, these difference were not statistically significant (p > .05). Task 4 deviated from this pattern, with trained drivers (G2) recording higher FC (G1 = 25.93 vs. 29.42). Moreover, in the case of task repetition (Task 1a, 1b, and 1c), a reduction in FC were observed, particularly for Group 2.

FC on the Message Area AOI remained relatively low across all tasks, with comparable levels of attention observed in both groups. For the Centre Mirror and Left Mirror AOIs, FC were minimal across tasks for both groups, reflecting limited visual engagement with mirrors during secondary task performance. The highest mirror-related FC values were observed during the undistracted driving condition. Similarly, fixations on the Speedometer were low in both groups, with the highest counts occurring in the undistracted condition (G1 = 6.89; G2 = 7.33). During task-based conditions, Group 2 occasionally showed slightly more frequent fixations (e.g., Task 1a: G2 = 1.94 vs. G1 = 1.14). To evaluate whether the differences in visual attention between groups were statistically significant, one-way ANOVA tests were conducted for each AOI. The results showed no statistically significant differences between the trained and untrained groups across all AOIs (p > .05).

One of the limitations of this study is that it was conducted in a controlled environment, without significant interaction with surrounding vehicles. Although this setup improved internal validity by reducing visual demand, it limits the realism of the driving conditions compared to real-world setting. As a result, the findings mainly reflect driver behaviour in low-demand traffic scenarios. Future studies should examine the effects of training in environments with multiple vehicles, intersections, varied road geometry conditions and other events to better understand how short training on the use of touchscreen influence driver visual behaviour. Moreover, the broad range of participant ages and driving experience may have introduced variability in visual behaviour. While group matching was applied to balance participant characteristics across, individual differences related to age and driving experience may still have influenced the results. Future research with more narrowly defined samples would allow for a more detailed examination of how these factors interact with touchscreen distraction and short pre-drive training effects. On the other hand, the pre-drive training consisted of an observational demonstration rather than interactive hands-on practice. While this approach ensured standardized exposure across participants, it may have limited the extent to which procedural familiarity and motor learning could develop compared to active training methods. Future studies should examine interactive training protocols that allow drivers to practice touchscreen tasks prior to driving to better assess how different training modalities influence distraction and performance as well as participants opinion about the provided training.

The lack of statistically significant differences in eye-tracking measures parallels the earlier findings based on driving-performance metrics suggesting that short pre-drive training may have limited impact on both behavioural and visual domains (Hazoor et al., 2026). However, the complementary nature of these datasets highlights an opportunity for integrated analysis. Combining driving-performance indicators (e.g., speed and lane position variability) with visual attention metrics could provide a deeper understanding of how short pre-drive training influences multitasking behaviour. Future research should consider integrating eye-tracking and driving-performance data to examine the interplay between visual distraction and vehicle control. This multimodal approach could reveal whether changes in gaze behaviour correspond to measurable differences in driving stability, thereby offering a more comprehensive understanding of training effects.

Although baseline (non-distracted) driving was included in the analysis, the primary objective of this study was to compare trained and untrained drivers during secondary task engagement; baseline analysis therefore served as a complementary reference. Data for baseline driving was acquired within the same drive session may have influenced visual behaviour even without secondary tasks. While this does not affect the main between-group comparisons during distracted driving, future studies could include a separate baseline driving to better isolate training effects under undistracted conditions.

Finally, future research should consider collecting additional participant-level information, such as familiarity with touchscreen interface, the frequency of car-sharing and usage of in-vehicle touchscreens. These factors could influence drivers’ familiarity with interface systems and their visual attention patterns, providing deeper insight into the effectiveness of short pre-drive training and individual differences in distraction risk.