Effect of route familiarity on attentional resource allocation during real-world road driving

Abstract

Previous studies indicate that route familiarity can adversely affect drivers’ attentional states. However, few studies have examined how route familiarity affects attentional resource allocation using physiological measures. We investigated the effect of route familiarity on attentional resource allocation by measuring physiological indices during real-world road driving. Participants (n = 10) drove a vehicle on public roads on the same route over six days. We analyzed the eye-fixation-related brain potentials (EFRPs), auditory-evoked potentials (AEPs), and eye-blink duration from the electroencephalographic (EEG) and electrooculographic (EOG) signals as indices of attentional resource allocation. The results indicated that the P2 amplitude of the AEPs elicited by task-irrelevant auditory probes significantly increased on Days 3–4 and Days 5–6 compared to Days 1–2, suggesting that attentional resources allocated to cognitive processing decreased as route familiarity increased. The P1 amplitude of the EFRPs decreased across days, with a significant reduction from Days 1–2 to Days 3–4, suggesting that attentional resources allocated to visual processing also decreased as route familiarity increased. This study supports previous findings that route familiarity affects drivers’ attentional states, providing new objective physiological evidence. We also discussed the differences in drivers’ inattention caused by route familiarity compared to the driving duration.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-55010-0.

Introduction

Route familiarity and inattention

A road traffic crash, defined as a collision involving at least one vehicle in motion on a public or private road that results in at least one person being injured or killed¹, can result in property damage, injury, or loss of life². Road traffic crashes arise from multiple causes, including driving system malfunctions and inadequate road infrastructure³. Most studies suggest that human error is the primary cause of road traffic crashes⁴. A review by Petridou and Moustaki⁵ identified four categories of human factors in road traffic crashes. These factors included those that reduce capability on a long-term basis (e.g., inexperience, aging), those that reduce capability on a short-term basis (e.g., drowsiness, fatigue), those that promote risk-taking behavior with long-term impact (e.g., overestimation of capabilities), and those that promote risk-taking behavior with short-term impact (e.g., moderate ethanol intake). Based on this categorization, Bucsuházy et al.³ investigated human factors related to causes of more than 1800 traffic accidents through individual interviews. The results indicated that the most common contributing human factor in traffic accidents, regardless of age or gender, was inattention, one of those that reduce capability on a short-term basis³. Thus, identifying factors that cause temporary reductions in capability, particularly inattention, is essential to improving traffic safety.

Several studies^6–8 have demonstrated that route familiarity can impair drivers’ attention. For instance, in a study by Charlton and Starkey⁶, participants completed driving sessions in a driving simulator regularly over three months. They were asked to report anything interesting or unusual they noticed while driving. The results showed that participants reported fewer items relevant to safe driving as they became more familiar with the route. Another simulator study by Martens and Fox⁷, which examined the effect of repeated exposure to a road environment on eye movements over several days, reported similar results. They indicated a decrease in glance duration at traffic signs with an increasing number of days. Furthermore, in a study by Yanko et al.⁸, participants driving in a simulator were instructed to step on the brake pedal to avoid a collision. The results revealed that participants familiar with the route had longer reaction times to a collision than those unfamiliar with the route. These findings align with the well-established observation that drivers are more likely to crash near their homes^9,10.

Although a close relationship between route familiarity and inattention has been suggested, further studies are needed to investigate two key aspects of this relationship. First, only a few studies have investigated how route familiarity affects drivers’ states in terms of attentional resource allocation. Driving is a complex task that involves several types of information processing such as visual and cognitive processing, which requires the allocation of psychological energies called attentional resources¹¹. Thus, for understanding the influence of route familiarity in detail, it is important to estimate attentional resources allocated to visual and cognitive processing, as well as total attentional resources while driving to systematically examine information processing that underlies driving behavior and find a way to increase safety level of traffic environment. Previous studies^6–8 examined the effect of route familiarity on drivers’ attentional states through indices such as glance duration, driving behaviors, and detection performance. These indices can reflect variation of general mental state, but are not sufficient for understanding the nature of attentional resources in detail, so existing evidence is insufficient to clarify how route familiarity affects attentional resource allocation.

The second limitation of previous studies examining the effect of route familiarity on inattention is that, in most cases, drivers’ attentional state was assessed while driving in a simulator rather than on a real-world road. Although previous studies have suggested that driving simulator studies can help understand drivers’ behavior¹², drivers’ attentional states can differ between real-world and simulated environments because the potential consequences of a crash differ between the two environments. For instance, in a previous study¹³ that compared state of drivers between in a driving simulator and in a real vehicle, physiological stress reactions occurred only in real vehicles and feeling of presence was higher in real vehicle than in the simulator. Therefore, studies conducted in real-world road environments are necessary to clarify how route familiarity affects drivers’ attentional states.

Physiological measures of attentional resource allocation

Kimura et al.¹⁴ proposed that attentional resources allocated to visual and cognitive processing, as well as total attentional resources when driving, can be estimated by combining three physiological measures: (1) eye-fixation-related potentials (EFRPs), (2) auditory-evoked potentials (AEPs) elicited by task-irrelevant auditory probes, and (3) eye-blink durations. EFRPs are a type of event-related potentials (ERPs) obtained by averaging electroencephalographic (EEG) signals time-locked to the offset of saccadic eye movements (i.e., the onset of eye fixations). The P1 of EFRPs is a positive-going component with a peak at around 90 ms (Yagi, 1979). Studies have reported that the P1 amplitude in EFRPs increases with attention to visual information^15,16. Thus, the P1 of EFRPs is a helpful measure of the attentional resources allocated to visual processing during driving^17,18.

Kimura et al.¹⁴ also proposed that the task-irrelevant probe technique, in which a person drives a vehicle while ignoring auditory stimuli (i.e., task-irrelevant auditory probes), could help estimate attentional resource allocation. In this technique, AEPs are obtained by averaging EEG signals time-locked to the onset of task-irrelevant auditory probes. Because the capacity of attentional resources is limited¹¹, reduced allocation to the primary task (driving) can increase residual resources available for task-irrelevant auditory probes, thereby increasing AEP amplitudes. Thus, amplitudes of AEPs to task-irrelevant auditory probes are negatively correlated with attentional resources allocated to the main task (e.g., Kramer et al.¹⁹). This phenomenon applies when the main task is driving. For instance, Takeda et al.²⁰ reported that the P2 amplitude of AEPs, a positive-going component with a peak latency of approximately 180 ms²¹, was smaller when participants reported that the driving route was challenging. Furthermore, it is suggested that the P2 component is associated with voluntary rather than involuntary attention^20,22, indicating that the P2 component of AEPs could be an index of voluntary allocation of attentional resources to cognitive processing during driving. It has also been suggested that the N1 component of AEPs, a negative deflection with a peak latency of approximately 90 ms²¹, is associated with involuntary attentional resource allocation to relatively earlier information processing^20,23.

Moreover, eye-blink duration is a physiological measure that can be obtained with EOG. Previous studies have shown that longer eye-blink duration is correlated with lower arousal^24–26. Because total attentional resources are closely related to arousal level¹¹, we estimated total attentional resources during driving by measuring eye-blink duration (e.g., Wu²⁷, Wu²⁸).

Since current technology of measurement and analysis of EEG and EOG still needs delicate process, it is difficult to monitor EFRPs, AEPs, and eye-blink durations real-time and adapt the result to a driver monitoring system at this point unlike driving or eye behaviors²⁹. Nonetheless, the combined use of these three measures enables us to disentangle individual amounts of attentional resources allocated to visual and cognitive processing and further consider voluntary and involuntary attentional resource allocation separately.

The present study

Purpose

Building on these findings, this study investigated the effect of route familiarity on attentional resource allocation during real-world road driving by measuring physiological indices. Participants (N = 10) drove a vehicle along the same, approximately 60 km, route in Tsukuba City and Tsuchiura City in Japan over six days. To assess attentional resource allocation, we recorded EEG and EOG signals while driving and analyzed EFRPs, AEPs, and eye-blink durations as physiological measures of attentional resource allocation. Furthermore, subjective ratings (sleepiness and mental workload) and driving behavior (vehicle speed and acceleration of steering wheel angle) were also recorded as additional indices. The acceleration of the steering wheel angle was analyzed as an index of visuomotor control because an increase in steering wheel angle acceleration is considered an indicator of steering control instability³⁰.

Hypotheses

As route familiarity increases, drivers are expected to allocate fewer attentional resources to visual information processing, which should be reflected in reduced P1 amplitude in EFRPs. At the same time, cognitive processing demands are expected to shift, leading to increased P2 amplitude in AEPs. Finally, overall attentional load is anticipated to decline with familiarity, which should manifest as longer eye-blink durations during driving. Based on this reasoning, we proposed the following hypotheses that align with prior findings.

H1

P1 amplitude of EFRPs is expected to decrease as driving days increase.

H2

P2 amplitude of AEPs is expected to increase as the number of driving days increases.

H3

Eye-blink duration is expected to increase as driving days increase.

Additional purpose

Driving for a long time is a major cause of road traffic crashes (e.g., Chu³¹, Ting et al.³², Walker and Trick³³). It is often difficult to remain entirely focused on driving for extended periods, and drivers may consequently divert their attention from driving³³. Based on Petridou and Moustaki’s⁵ categorization of human factors related to road traffic crashes, driving duration, like route familiarity, can be considered those that reduce capability on a short-term basis. Therefore, driving duration is also worth considering, alongside route familiarity, as an attentional resource allocation factor. This study instructed participants to drive to the halfway point and then return to the starting point following the same path. Thus, we also compared physiological indices (EFRPs, AEPs, and eye-blink duration), subjective ratings (sleepiness and mental workload), and driving behavior (vehicle speed and acceleration of steering wheel angle) between outward and return journeys to investigate the effect of driving duration on attentional resource allocation. We hypothesized that a decrease in the P1 of EFRPs, an increase in the P2 of AEPs, and an increase in eye-blink duration would be observed on the return journey compared to the outward journey if drivers become less attentive due to longer driving duration.

Methods

Participants

Adults with normal hearing and normal or corrected-to-normal vision participated in the experiment (N = 10, 5 females; age range, 24–67 years; mean, 48.5 years). They were recruited through a staffing agency. They lived in Tsukuba City or the surrounding area, thus they have some knowledge about environments of the driving route in this experiment. All participants were licensed drivers who reported driving more than three days per week. They were paid to participate in the experiment through the staffing agency. This study was approved by the National Institute of Advanced Industrial Science and Technology (AIST) Safety Ethics Committee and was performed in accordance with relevant guidelines and regulations. The experiment was conducted after each participant provided written informed consent.

Procedure

Participants took part in the experiment for six days. On each experimental day, they sat in an experimental vehicle (a Toyota Prius) after electrodes were attached. Half of the participants began driving at 9:15 AM; the others began at 1:45 PM. They were asked to avoid smoking, drinking, eating, sleeping, or talking while driving. A male experimenter rode in the rear seat of the vehicle to provide the driver with lane and speed instructions and to monitor electrophysiological signals.

We conducted 60 experimental runs (six per participant). Among these, two runs, on Day 2 and Day 3, had missing data for one participant on each day due to equipment malfunctions.

Driving route

The driving route’s starting point and halfway points were respectively the parking lot of the National Institute of Advanced Industrial Science and Technology in Tsukuba City and the Tsuchiura Kita interchange of the Joban Expressway in Tsuchiura City. Participants were required to drive the experimental vehicle from the starting point to the halfway point (about 30 km one-way) and then return to the starting point along the same route. The driving route consisted of two parts: an urban road (approximately 13 km one-way) and an expressway (approximately 17 km one-way). Mean of driving duration was 32.7 min (standard error [SE] = 0.5) for outward journey of an urban road, 31.5 min (SE = 0.4) for return journey of an urban road, and 32.8 min (SE = 0.5 min) for expressway. The participants entered the expressway at the Tsukuba Chuo interchange on the Ken-Ō Expressway during the outward journey and exited at the same interchange on the return journey. They were instructed to go with the flow of traffic (i.e., not to go against the traffic, not to weave unpredictably, and to follow the movement of the traffic around them).

We focused on drivers’ attentional states during urban road travel because they encounter diverse traffic situations and must process greater amounts of visual and cognitive information. Thus, data acquired on urban roads during the outward and return journeys were analyzed and included in the statistical analyses.

Moreover, driving on expressways is known to decrease arousal and increase boredom^34,35, which can be critical for reducing attention. When investigating the effect of route familiarity on attentional states, it is essential to consider the potential interaction with decreased arousal or increased boredom. Therefore, we included an expressway in the driving route. However, it is assumed that an effect of route familiarity is slight when driving in expressway, since attentional resources are less needed in expressway compared to in urban road irrespective of route familiarity¹⁷. Thus, we excluded the expressway’s data from the main analysis in the present study. As a supplementary analysis, we examined the effect of route familiarity on attentional resource allocation in expressway (available as Methods and results, Table S1, and Table S2 in Supplementary file).

Auditory probes

To measure AEPs, task-irrelevant auditory probes were discretely presented during driving via speakers (Sennheiser, HD6 MIX) placed on each side of the driver’s headrest. As in previous studies^2036,37;, the stimulus sequence consisted of 12 pure tones of 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, or 1600 Hz, which were presented with equal probabilities in random order with the restriction that tones with the same frequency did not appear successively. The duration of each tone was 50 ms (including 10-ms rise and fall times). The stimulus interval (onset to onset) was 400, 500, 600, 700, or 800 ms (mean = 600 ms). The sound pressure level was approximately 75 dB.

Measures and analysis

Physiological responses

Each driver’s EOGs and EEGs were recorded with a digital amplifier (BrainProducts, BrainAmp Standard system) attached to silver–silver chloride electrodes. The horizontal EOG was recorded from the outer left and right canthi, and the vertical EOG was recorded using electrodes placed above and below the right eye. EEGs were recorded from 23 scalp sites (Fp1, Fp2, F7, F3, Fz, F4, F8, FCz, T3, C3, Cz, C4, T4, CPz, T5, P3, Pz, P4, T6, Poz, O1, Oz, and O2 according to the extended international 10–20 system). The left earlobe (A1) and AFz were set as the reference and ground electrodes, respectively. The impedance of all the electrodes was kept below 10 kΩ. The signals were digitized at a sampling rate of 1000 Hz. The time constant was set at 10 s. The EEG signals were re-referenced to mathematically averaged earlobes (A1 and A2) and then band-pass filtered offline at 0.1–30 Hz with a second-order Butterworth filter.

EFRPs. The offset of saccadic eye movements relative to horizontal EOG signals was visually inspected by an experimenter blind to the experimental condition. Then, EFRPs time-locked to the saccadic eye movement offset were calculated by averaging the EEG signals. Very small saccades do not reflect new visual information processing, and excessively large saccades are often triggered by information unrelated to driving. Therefore, based on a previous study¹⁷, we chose saccadic eye movements for EFRP calculation, with saccade sizes ranging from 25 to 400 µV. The EEG signals were averaged to obtain EFRPs for each journey in each participant. Epoch averaging was set from − 100 to 400 ms relative to the saccadic eye movement offset. Before averaging, artifacts derived from blinks and eye movements were removed using independent component analysis implemented in EEGLAB version 14.1.2b³⁸. Following the procedures used in previous studies^14,17, components related to blinks and eye movements were identified by visual inspection. As a result, the number of averaged epochs in each journey was, on average, 2832.4 (SE = 76.2) in the outward journey and 2641.6 (SE = 88.2) in the return journey.

We calculated the EFRPs at the midline occipital site because typical EFRPs are shown at this site¹⁷. In the present study, grand-average EFRPs (collapsed across all participants and journeys) showed that the peak latency of the EFRPs’ P1 component was 85 ms relative to the saccadic eye movements’ offset. Thus, we calculated P1 as the mean amplitude between 85 ± 25 ms (i.e., 60–110 ms). Following previous studies^14,17, we estimated the amplitude of P1 at the Oz site by subtracting the amplitude of N0, a component that precedes P1 (i.e., the P1 minus N0 amplitude difference). Since the N0 component’s peak latency was − 5 ms relative to the offset of saccadic eye movements, we calculated N0 as the mean amplitude between − 5 ± 25 ms (i.e., − 30 to 20 ms).

AEP. The EEG signals were averaged to obtain AEPs for each participant during the journey. Epochs were averaged from − 100 to 400 ms relative to the onset of the auditory probes. Similar to the EFRP analysis, we used independent component analysis³⁸ to remove blink- and eye-movement-related artifacts before averaging. As a result, the number of averaged epochs in each journey was, on average, 3307.1 (SE = 39.2) in the outward journey and 3161.6 (SD = 40.7) in the return journey. The grand-average AEPs (collapsed across all participants and journeys) showed a negative deflection at approximately the FCz site, with a peak latency of 80 ms (i.e., N1), and a positive deflection at approximately the same site, with a peak latency of 192 ms (i.e., P2). The mean amplitudes of N1 and P2 were calculated at FCz within the 80 ± 12 ms (i.e., 68 to 92 ms) time window and the 192 ± 12 ms (i.e., 180 to 204 ms) time window, respectively (baseline correction: −100 to 0 ms) using the quantification method used in previous studies^14,20,36.

Eye-blink duration. Eye-blink duration was calculated from vertical EOG signals. Time points of eye blinks were defined as the peaks of eye-blink-related EOG deflections. Following previous studies^17,24,27,28, the eye-blink initiation for each eye blink was defined as the time point when the rising slope of the eye-blink-related EOG deflection reached 50% of the peak amplitude. Similarly, the eye-blink termination was defined as the time point when the falling slope of the eye-blink-related EOG deflection reached 50% of the peak amplitude. The interval between these initiation and termination points was defined as eye-blink duration.

Subjective ratings

Participants were asked to rate their subjective sleepiness and mental workload. The subjective level of sleepiness was measured using the Japanese version³⁹of the Karolinska Sleepiness Scale (KSS⁴⁰;, consisting of a 9-point scale: 1 = very alert, 3 = alert, 5 = neither alert nor sleepy, 7 = sleepy, 9 = very sleepy (fighting sleep). The subjective level of mental workload was assessed using the Rating Scale of Mental Effort (RSME⁴¹;, which comprises a vertical axis ranging from 0 to 150 and nine descriptive indicators ranging from 3 (not effortful) to 114 (awfully effortful).

To measure baseline of subjective sleepiness, participants filled out the KSS before driving. Regarding subjective level of sleepiness and mental workload during driving, we conducted subjective ratings when participants returned to the starting point each day after completing the drive, since making subjective ratings while driving could be dangerous. Then, they recalled subjective sleepiness and mental effort felt during each part of the journey, outward and return.

Driving behaviors

The driving behavior was recorded at a sampling frequency of 10 Hz. Driving behavior when the vehicle’s speed was under 10 km/h was excluded from the analysis. We analyzed mean vehicle speed, which can be an index of demand control during driving (e.g., Sato et al.⁴². In addition, we also examined the mean acceleration of the steering wheel angle, calculated as an absolute value, because it has been suggested as an indicator of steering control instability³⁰.

Other measurements

In addition to the above measures, we also acquired videos of the drivers’ appearances and road environments, body pressure distribution on the driver’s seat, electrodermal activity (EDA), and electrocardiographic (ECG) signals. These data will be reported elsewhere.

Statistical analysis

The physiological responses (i.e., P1 amplitude of EFRPs, N1 and P2 amplitudes of AEPs, and eye-blink duration), subjective ratings (KSS and RSME scores), and driving behavior (vehicle speed and acceleration of steering wheel angle) were calculated for each journey. Then, to decrease variations caused by factors unrelated to route familiarity, such as weather and road traffic, they were averaged for the following three periods: Days 1–2, Days 3–4, and Days 5–6. We conducted a two-factor repeated measures analysis of variance (ANOVA) with these data: 3 Days (Days 1–2 vs. Days 3–4 vs. Days 5–6) × 2 Journeys (outward journey vs. return journey). Post hoc tests were conducted using Bonferroni correction when the main effect or an interaction was significant. The significance level was set at 0.05.

Results

Physiological responses

EFRP. The grand-average EFRPs and the mean P1 amplitude are shown in Fig. 1. The ANOVA revealed a significant main effect of Day, F (2, 45) = 5.31, p = 0.0085, ηp² = 0.19. There was neither a significant main effect of Journey (p = 0.9950) nor a significant interaction (p = 0.9978). Post hoc tests showed that the P1 amplitudes were significantly smaller on Days 3–4 than on Days 1–2 (Bonferroni-corrected p = 0.0033).

Fig. 1.

(A) Grand-average EFRPs at the Oz site for each Day. (B) Grand-average EFRPs at the Oz site for each Journey. (C) Topographical map for the P1 time window of the grand-average EFRPs. (D) Mean P1 amplitudes of the EFRPs at the Oz site. Error bars indicate one standard error. Asterisks indicate significant differences (**Bonferroni-corrected p < 0.01) shown in post-hoc tests following the ANOVA.

AEP. The grand-average AEPs and the mean N1 amplitude and mean P2 amplitude of AEPs are shown in Fig. 2. Regarding the N1 amplitude (Fig. 2D), an ANOVA revealed a significant main effect of Journey, F (1, 45) = 55.45, p < 0.0001, ηp² = 0.55, indicating that the N1 amplitude was significantly smaller, or less negative, during the return journey than in the outward journey. There was neither a significant main effect of Day (p = 0.4799) nor a significant interaction (p = 0.4352).

Fig. 2.

(A) Grand-average AEPs at the FCz site for each Day. (B) Grand-average AEPs at the FCz site for each Journey. (C) Topographical map for the N1 time window of the grand-average AEPs. (D) Mean N1 amplitudes of the AEPs at the FCz site. (E) Topographical map for the P2 time window of the grand-average AEPs. (F) Mean P2 amplitudes of the AEPs at the FCz site. Error bars indicate one standard error. Asterisks indicate significant differences (*Bonferroni-corrected p < 0.05) shown in post-hoc tests following the ANOVA.

Regarding the P2 amplitude (Fig. 2F), an ANOVA revealed a significant main effect of Day, F(2, 45) = 8.93, p = 0.0005, ηp² = 0.28. There was neither a significant main effect of Journey (p = 0.1155) nor a significant interaction (p = 0.5783). Post hoc tests showed that the P2 amplitudes were significantly greater on Days 3–4 and Days 5–6 than on Days 1–2 (Days 1–2 vs. Days 3–4: Bonferroni-corrected p = 0.0258; Days 1–2 vs. Days 5–6: Bonferroni-corrected p = 0.0108).

Eye-blink duration. The mean eye-blink durations are shown in Fig. 3. An ANOVA revealed a significant main effect of Journey, F (1, 45) = 13.58, p = 0.0006, ηp² = 0.23, indicating that the eye-blink durations in the return journey were significantly longer than those in the outward journey. There was neither a significant main effect of Day (p = 0.5794) nor a significant interaction (p = 0.8752).

Fig. 3.

Mean eye-blink durations. Error bars indicate one standard error.

Subjective ratings

Sleepiness. Mean KSS scores are shown in Fig. 4A. The ANOVA revealed a significant main effect of Journey, F (1, 45) = 5.85, p = 0.0197, ηp² = 0.12, indicating that subjective sleepiness levels in the return journey were significantly higher than those in the outward journey. There was neither a significant main effect of Day (p = 0.4569) nor a significant interaction (p = 0.4389). As an additional analysis, we conducted a repeated ANOVA with Days for KSS scores reported before driving to examine whether there was any difference in baseline of subjective sleepiness level among experimental days. The results of the ANOVA indicated that there was no significant effect of Days (p = 0.3306).

Fig. 4.

(A) Mean scores of subjective sleepiness. (B) Mean scores of subjective mental workload. Error bars indicate one standard error.

Mental workload. The mean scores of the RSME are shown in Fig. 4B. There was neither a significant main effect of Day (p = 0.8499), nor a significant main effect of Journey (p = 0.5653), nor a significant interaction (p = 0.2297).

Driving behaviors

Vehicle speed. The mean vehicle speed is presented in Fig. 5A. An ANOVA revealed a significant main effect of Journey, F (1, 45) = 11.44, p = 0.0015, ηp² = 0.20, indicating that vehicle speed in the return journey was significantly higher than that in the outward journey. There was no significant main effect of Day (p = 0.8698) nor a significant interaction (p = 0.6882).

Fig. 5.

(A) Mean speed of vehicle. (B) Mean acceleration of the steering wheel angle. Error bars indicate one standard error.

Acceleration of the steering wheel angle. The mean acceleration of the steering wheel angle is shown in Fig. 5B. An ANOVA revealed a significant main effect of Journey, F (1, 45) = 5.60, p = 0.0223, ηp² = 0.12, indicating that the acceleration of the steering wheel angle in the return journey was significantly greater than that in the outward journey. There was no significant main effect of Day (p = 0.3665) nor a significant interaction (p = 0.8428).

Discussion

We investigated the effect of route familiarity on attentional resource allocation during real-world road driving using physiological measures. Participants drove a vehicle along the identical route over six days. EFRPs, AEPs, and eye-blink duration were analyzed from EEG and EOG signals as indices of attentional resource allocation.

Route familiarity and attentional resource allocation

Based on previous reports that route familiarity has a negative impact on drivers’ attentional states^6,7; Yanko et al.⁸, we assumed that attentional resources allocated to visual and cognitive processing, as well as total attentional resources, would decrease as days passed and the route became more familiar. As mentioned in the Introduction, the P1 of EFRPs reflects the attentional resources allocated to visual processing when driving^17,18. Therefore, we hypothesized that the P1 amplitude of EFRPs would decrease over later experimental days as route familiarity increased. The results revealed that the P1 amplitude of EFRPs was significantly smaller on Days 3–4 than on Days 1–2, indicating that participants may have allocated less attention to visual information as route familiarity increased, which partly supported our hypothesis. However, the difference in the P1 amplitude of EFRPs between Days 1–2 and Days 5–6 was not significant, indicating that the day-to-day changes in P1 amplitude were inconsistent. These results suggest that route familiarity exerts a modest effect on visual attentional resources.

We also found that the P2 amplitude of AEPs increased significantly on Days 3–4 and 5–6 compared to Days 1–2. The P2 amplitude of AEPs is negatively associated with attentional resources required for cognitive processing²⁰. Therefore, this result suggests that attentional resources allocated to cognitive processing decreased with increasing route familiarity, which aligns with our hypothesis regarding the P2 of AEPs. Unlike the P1 component of the EFRPs, the P2 amplitudes showed stable day-to-day increases. These findings suggest that the effect of route familiarity on attentional resource allocation to cognitive processing is more pronounced than to visual processing. In addition, there was no significant effect of route familiarity on the N1 amplitude of AEPs, which reflects involuntary attentional resource allocation. Given that the P2 reflects voluntary attentional resource allocation, the increase in route familiarity may disrupt voluntary rather than involuntary attention.

A previous study by Sugimoto et al.³⁶ reported similar results for the P2 component of AEPs. In their study, participants drove a vehicle along the same route over 6 days, using different operating systems (i.e., one-pedal vs. two-pedal). The results indicated that the P2 amplitude of AEPs was larger in the second half of the drive than in the first half, regardless of the operating system. The consistent results between the present and the previous study by Sugimoto et al. support the correlation between route familiarity and the decrease in attentional resources allocated to cognitive processing.

We hypothesized that total attentional resources would also decrease as route familiarity increased, and that eye-blink duration would increase in later experimental days compared to earlier ones. However, the results did not support this hypothesis. There was no significant effect of the Day on the eye-blink duration, indicating that route familiarity did not significantly affect total attentional resources in the present study. Considered alongside the EFRP and AEP findings, these results suggest that route familiarity may decrease the voluntary allocation of attentional resources to visual and cognitive processing, but it does not affect total attentional resources.

We did not find a significant effect of Day on subjective sleepiness level or mental workload. The result on the subjective sleepiness level is consistent with the finding of no effect of Day on eye-blink duration. On the other hand, the result for subjective mental workload does not perfectly align with that for P2 amplitude in AEPs, suggesting that attentional resources allocated to cognitive processing during driving decrease in proportion to increases in route familiarity, perhaps because the driving route in the present study did not require much effort. Indeed, this assumption is supported by the RSME scores, which ranged from 10.3 to 90.7, whereas the original RSME scale ranges from 0 to 150. This finding suggests that the physiological index (i.e., the P2 amplitude) may be more sensitive to attentional resource allocation than subjective ratings, at least when a high level of effort is not required while driving.

Regarding driving behavior, there was no significant effect of Day on vehicle speed, nor the acceleration of the steering wheel angle. It has been reported that drivers tend to slow down driving speeds when they try to decrease task demands while driving (e.g., Sato et al.⁴². Thus, the result regarding vehicle speed in the present study suggests that participants did not show adaptive behavior to decrease demands by slowing down speed, even when attentional resources were reduced by high route familiarity. Further, a previous study³⁰ suggested that an increase in the rate of acceleration of the steering wheel angle is associated with steering control instability. Therefore, the present result indicates that route familiarity did not affect steering control instability. Although an increase in route familiarity reflects cognitive processing, as shown in the P2 amplitude, it did not affect driving behaviors such as driving speed and steering stability, suggesting that the detrimental effect of route familiarity on traffic safety^9,10 may occur through a decrease in cognitive processing rather than through visuomotor control instability.

Driving duration and attentional resource allocation

We also investigated the effect of driving duration on attentional resource allocation, as the driving duration is a crucial factor in increasing drivers’ inattention^31–33, by examining the effect of the outward vs. return journeys on measures assessing the route familiarity effect.

Results indicated no significant effect of Journey on the P1 amplitude of the EFRPs or the P2 amplitude of the AEPs, suggesting that resources allocated to visual or cognitive processing do not change with increasing driving duration. This finding does not support our hypothesis that P1 of EFRPs decreases and P2 of AEPs increases during the return journey, when drivers become inattentive due to longer driving duration. However, the results regarding eye-blink duration supported our hypothesis: eye-blink duration was significantly longer during the return journey than the outward journey, suggesting that overall attentional resources decreased as driving duration increased. Furthermore, we found that subjective sleepiness was higher during the return journey than during the outward journey. Subjective sleepiness ratings also supported the eye-blink findings, as a longer eye-blink duration is closely associated with decreased arousal^24–26. Taken together, the journey-related results suggest that resources allocated to visual or cognitive processing do not decrease as the driving duration increases; however, total attentional resources allocated to these tasks decrease.

Furthermore, the acceleration of the steering wheel angle was significantly higher on the return journey than on the outward journey. This finding suggests that increased driving duration may reduce total attentional resources and accuracy of visuomotor control. Moreover, speed of vehicle was higher on the return journey than on the outward journey. Given that slowing down speed has a function to decrease demands of driving⁴² as mentioned above, the present result suggests that there was no adaptive behavior to decrease demands by slowing down speed in response to attentional resources that were reduced by long driving duration; rather it suggests that driving behavior of participants could become careless with increasing driving duration. Although the N1 amplitude of the AEPs decreased (became less negative) in the return journey compared to the outward journey, the interpretation of this result is controversial. The eye-blink duration was longer in the return journey, and therefore, the decrease in the N1 amplitude may be attributable to reduced total attentional resources rather than a deficit in involuntary attention.

Relevance to road safety

As mentioned in the Introduction, inattention while driving is a crucial factor in increasing the risk of road traffic crashes. The study showed that both route familiarity and driving duration negatively impact drivers’ attentional states, as measured using physiological methods, including EFRPs, AEPs, and eye-blink duration. Simultaneously, we found that the effects of route familiarity and driving duration on attentional resource allocation differed. Route familiarity decreased attentional resources allocated to cognitive processing and possibly to visual processing, whereas a long driving duration decreased the total attentional resources. This finding is also supported by the absence of a significant interaction between Day and Journey for any index, indicating that the effects of route familiarity and driving duration were independent. Thus, even though drivers may appear similarly inattentive, route familiarity and the driving duration should be considered separately. Route familiarity may lead to detection errors due to reduced cognitive processing, which can be mitigated by V2X (Vehicle-to-Everything) technology. In contrast, the influence of driving duration may induce instability in steering control, which can be prevented with lane-keeping assistant systems.

Meanwhile, it is noted that alternative interpretation regarding the effect of route familiarity on road safety should be considered. Although inattention is generally regarded to have a negative impact on traffic safety, it can be also considered as adaptation process to environments. For instance, decreased visual processing of static environments (e.g., traffic sign) when route familiarity is enough high can be interpreted as appropriate driving behavior from another point of view (i.e., drivers reduce wasteful processing). Thus, it is hard to conclude that route familiarity is necessarily related to increase of traffic accidents from the results of the present study. Future studies need to examine this issue for deeper understanding of the influence of route familiarity.

Limitations of the study

The present study has specific limitations. First, our sample size of 10 was relatively small. Therefore, the results should be interpreted with caution. For example, the results of the present study showed that route familiarity had an impact on voluntary allocation of attentional resources (as indicated by the P2 amplitude of AEPs) but not on involuntary allocation (as indicated by the N1 amplitude of AEPs) nor total attentional resources (as indicated by the eye-blink duration), however more data is needed to ensure the null results. Further studies using larger samples are needed to confirm the present results.

Second, the study was conducted on actual roads, and participants were instructed to travel in the direction of traffic flow, which is both a strength and a limitation, as conditions other than route familiarity, such as traffic or weather, could not be controlled, as is the case in driving simulator studies. These uncontrolled real-world factors may also have influenced the association between route familiarity and attentional resource allocation. Further studies that replicate the present paradigm and collect more data on real-world roads under various conditions may help address this limitation.

The third limitation of the present study is related to the effect of driving duration. We compared drivers’ physiological responses between outward and return journeys to investigate the effect of driving duration on attentional resource allocation, as an additional purpose of the present study. However, the outward and return journey should not be easily regarded as the repetition of the same route (e.g., a right turn in the outward journey becomes a left turn in the return journey). Thus, further studies using more sophisticated experimental designs are needed to complement the results of the present study.

Conclusions

We investigated the effect of route familiarity on attentional resource allocation using physiological indices during real-world road driving. The results indicated that route familiarity significantly reduces the attentional resources allocated to cognitive processing and possibly to visual processing, suggesting that route familiarity contributes to increased driver inattention. The present study provides objective physiological evidence supporting previous findings that route familiarity has a significant effect on drivers’ safety, as demonstrated by physiological measures in real-world driving.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ANOVA

Analysis of variance

AEP

Auditory-evoked potential

EEG

Electroencephalogram

EFRP

Eye-fixation-related brain potential

EOG

Electrooculogram

ERP

Event-related potential

KSS

Karolinska Sleepiness Scale

RSME

Rating Scale Mental Effort

SE

Standard error

Author contributions

D. Choi: Writing – Original Draft, Visualization. Y. Takeda: Conceptualization, Supervision, Writing – Review & Editing. M. Akamatsu: Supervision, Writing – Review & Editing. M Murayama: Project Administration, Writing – Review & Editing. T. Shibata: Resources, Data Curation, Writing – Review & Editing. M. Yamamoto: Resources, Writing – Review & Editing. K. Sasai: Project Administration, Writing – Review & Editing.

Funding

This study was supported through collaborative research between Sumitomo Riko Co., Ltd. and AIST.

Data availability

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.