How does hands-free cognitive distraction influence cycling behaviour and perceived safety?

Full text

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
Available online 2 July 2024
1369-8478/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
How does hands-free cognitive distraction inuence cycling
behaviour and perceived safety?
Mette Møller
a
,
*
, Frauke Luise Berghoefer
b
, Mark Vollrath
b
a
Technical University of Denmark, Department of Technology, Management and Economics, Building 358, DK-2800 Kgs Lyngby, Denmark
b
Technische Universit¨
at Braunschweig, Trafc and Engineering Psychology, Gaußstraße 23, 38106 Braunschweig, Germany
ARTICLE INFO
Keywords:
Cycling
Cycling safety
Cognitive distraction
Secondary tasks
Bike simulator
ABSTRACT
Previous studies using a survey or eld observational approach show that secondary task
engagement negatively impacts cyclists’ performance. Prohibiting handheld phone use, while
allowing hands-free use via headphones, is applied to reduce safety critical impact of secondary
task engagement. Hands-free secondary task engagement limits visual-motor distraction, but
cognitive distraction might still impact cycling performance. Therefore, the purpose of this study
was to investigate the behavioural effects of hands-free cognitive secondary task engagement
while riding alone on different kinds of cycle paths as well as when overtaking other cyclists and
reacting to typical events like stopping at a trafc light or evading a pedestrian or an obstacle on
the cycle way. Using the cycling simulator at the Department of Trafc and Engineering Psy-
chology at the TU-Braunschweig, a mixed design was used with three levels of secondary task
engagement as the independent between-factor: no task (NT), podcast task (PC), acoustic speech
task (AS). Additionally, three types of lane markings of the cycle way and three events were
varied as within-factors. N =58 participants (36 female, 22 male) completed the experiment. In
none of the situations and none of the parameters examined, an effect of secondary task
engagement was found, although cyclists subjectively felt more distracted. This was not due to a
lack of sensitivity to the parameters measured, as the types of infrastructure signicantly inu-
enced cycling and overtaking behaviour. In line with multiple resource theory, results suggest
that secondary tasks requiring cognitive but not visual-motor resources can be done while cycling
without adverse effects on behaviour. However, this might also be an effect of task difculty and
may change when the cycling task or the secondary task becomes more complex. Thus, additional
studies including more complex trafc situations, are relevant.
1. Introduction
Cycling is associated with environmental benets (Xia et al., 2013) and the individual health benets of cycling generally outweigh
the possible negative effects of increased exposure to air pollution (Woodward & Samet, 2016). Therefore, bicycling is promoted as a
sustainable means of daily transport and a key element to ensure sustainability in the transport sector. However, the risk of being
seriously injured or killed in a road trafc crash is comparably high for cyclists. In Denmark, for instance, the risk of being seriously
injured or killed in a crash is 13 times higher per km cycled for cyclists compared to drivers (Christiansen & Warnecke, 2018). Although
* Corresponding author.
E-mail address: mette@dtu.dk (M. Møller).
Contents lists available at ScienceDirect
Transportation Research Part F:
Psychology and Behaviour
journal homepage: www.elsevier.com/locate/trf
https://doi.org/10.1016/j.trf.2024.06.026
Received 13 March 2024; Received in revised form 28 June 2024; Accepted 28 June 2024

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
523
motor trafc is the primary source of danger for cyclists, distraction can facilitate unsafe behaviour among cyclists and should,
therefore, be examined.
Previous studies indicate that secondary task engagement impacts performance and is associated with increased unsafe behaviours
among cyclists. Thus, Terzano (2013) found that people who engage in activities such as using a mobile phone, iPod, or smoking while
cycling were more likely to engage in behaviours such as riding in the wrong direction on a one-way cycle path, failing to slow down
and look for crossing trafc or cycling unexpectedly slow when entering an intersection thereby forcing approaching trafc to brake.
De Waard et al. (2014) found larger variation in lateral position and worse performance in visual detection among cyclists engaging in
secondary tasks compared to cyclists not engaging in secondary tasks, and Jiang et al. (2021) found secondary task engaging cyclists to
xate slower and accelerate faster.
Secondary task engagement has also been found to be associated with an increased risk of crash and near-crash involvement. Thus,
Goldenbeld et al. (2012) found frequent use of electronic devices while cycling to be associated with an increased likelihood of crash
involvement, and engaging with a smartphone, e.g., receiving a call, texting, or looking for information while cycling has been found to
be associated with increased engagement in violations and associated near-crashes (e.g., Useche et al., 2018, Useche et al., 2024).
Similarly, De Angelis et al. (2020) and Terzano (2013) found cyclists who engage in secondary tasks to more frequently engage in
behaviours that force other road users to perform crash avoidance manoeuvres. However, as in all correlational studies, it is unclear
whether engaging in secondary tasks is the cause for these dangerous behaviours or just another example of the general tendency of
some cyclists to behave more dangerously.
Although some previous studies indicate that secondary task engagement among cyclists has a detrimental effect on cycling
performance, other studies nd that cyclists compensate for this detrimental effect by different kinds of behavioural adjustments
during secondary task engagement, e.g., reducing riding speed (e.g., de Waard et al., 2014; Kircher et al., 2015) and engaging in the
tasks in less demanding situations (e.g., Brandt et al., 2021; Kircher et al., 2015). Among drivers such compensatory behaviours have
been shown to outweigh the detrimental effects of secondary task engagement in some situations thereby lowering the increased risk of
crash involvement due to secondary task engagement (for an overview see Vollrath et al., 2016; Parnell et al., 2020).
Many previous studies about secondary task engagement among cyclists focus on handheld secondary task engagement such as
phoning or texting (e.g., de Waard et al., 2010; 2014; Jiang et al., 2021). However, hands-free secondary task engagement such as
hands-free phone use is legal in many countries. However, hands-free phone use is among the most frequent type of secondary task
engagement among cyclists (Brandt et al., 2021; Ethan et al., 2016; Huemer et al., 2019). With a general increase in cycling and phone
use the prevalence of hands-free phone use while cycling is likely to increase too, but knowledge about the impact of hands-free phone
use and potential impact on safety is still limited. A study by de Waard et al. (2011) found that phone use reduced cycling speed and
response to auditory cues, but no difference between handheld and hands-free (via in-earbuds) phone use was found. As auditory cues
can be argued to resemble nonvisible safety–critical events approaching from behind, it remains relevant to assess the behavioural
impact of hands-free cognitive secondary task engagement regarding potentially safety–critical events occurring in front of the cyclists,
as the perception of such events does not depend on auditive cues.
Wickens’ (2002) multiple resources theory provides a theoretical and empirically supported explanation of the detrimental effect of
secondary task engagement. According to the theory the effect of secondary task engagement mirrors the overlap in resources needed
in both tasks. In general, the larger the overlap and demand between the resources needed, the larger the detrimental effect is expected
to be. More specically, the well-documented detrimental effect of texting on driver behaviour (e.g., Dingus et al., 2016; Victor et al.,
2015) as well as cyclists’ behaviour (Jiang et al., 2021) results from the driving task and the secondary task competing for the same
resources in terms of visual-spatial information and manual reactions. This also applies to the detrimental effect of hands-free phone
Fig. 1. The bicycle simulator used in the study.
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
524
use on auditory cues (de Waard et al., 2011). Thus, engaging hands-free in secondary tasks limits the detrimental visual-motor effect on
cycling behaviour but reduces access to auditory cues and can be expected to increase the cognitive demand, thereby negatively
impacting the cognitive processes related to safe cycling such as hazards perception and related behavioural response. However, so far,
no study has focused specically on this aspect of hands-free secondary task engagement, which should therefore be studied.
Based on the above, the objective of this study was to investigate the behavioural effects of engagement in hands-free cognitive
secondary tasks while cycling with a particular focus on overtaking behaviour and behavioural response in potentially critical situ-
ations. The study followed a controlled experimental approach using a bicycle simulator. Based on the above we expected to nd a
behavioural effect of secondary task engagement either in terms of compensatory behaviours, e.g., slower riding speed, or in terms of
crash-increasing behaviours, e.g., slower reaction time or shorter overtaking distance. We further expected the behavioural effect to be
stronger for more cognitively demanding tasks than baseline and less cognitively demanding tasks.
2. Method
2.1. Bicycle simulator
We conducted the study in a bicycle simulator (see Fig. 1) at the Department of Trafc and Engineering Psychology at the
Technische Universit¨
at Braunschweig, Germany. The mock-up of the simulator consists of a lady’s mountain bike mounted on a
platform. This platform is static but allows tilts to the side, making cycling more realistic. The bicycle is surrounded by twelve
monitors, which enable a 360◦view. A fan in front of the bicycle provides a headwind depending on the cycling speed. Ambient
sounds, such as passing motor vehicles or bicycle tyres on the ground are provided by Bluetooth headphones with noise cancellation.
The simulator runs with the simulation software SILAB 7.0 (Würzburg Institute for Trafc Sciences, 2022) which simulates the
environment and records the cycling data.
No validation study has been conducted in the simulator but the cycling behaviour in the simulator, e.g., the cycling speed and the
standard deviation of lateral position (SDLP), is similar to the behaviour found naturalistic and other simulator studies (e.g., O’Hern
et al., 2017; Kircher et al., 2018), and can therefore be considered to be close to natural.
2.2. Scenarios and parameters
All scenarios were realised on a straight section of about 300 m length with a cycle lane of 2 m width along an arterial road. The
width of the cycle lane followed German recommendations for cycling infrastructure (FGSV, 2010) and was chosen to create a realistic
setting. Furthermore, in all scenarios we simulated medium adjacent motor trafc with passenger cars and vans passing the participant
Fig. 2. The cycle lane markings in the overtaking scenarios (top) and the special events (bottom).
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
525
every 6 s with a speed of 50 km/h. The scenarios can be divided into two groups: overtaking scenarios and special event scenarios.
In the overtaking scenarios, a simulated cyclist appeared in front of the participants. This simulated cyclist cycled at a speed of 9
km/h and cycled close to the edge of the cycle lane so that it could easily be overtaken. The cycle lane was marked in three different
ways: a solid white demarcation line, a buffer zone, and protected by bollards on a buffer zone (see Fig. 2, top). These scenarios were
used to examine baseline cycling behaviour and overtaking behaviour and, to see how this is inuenced by trafc infrastructure. Each
of these scenarios was presented three times to obtain a stable estimate of cycling behaviour in these situations.
To describe baseline cycling behaviour we recorded the cycling behaviour starting at 10 m after the overtaking was completed and
until the end of the section. Specically, we measured average speed [km/h], mean lateral position [m], and standard deviation of lane
position (SDLP, m).
To examine differences in the overtaking behaviour, we used as dependent variables the velocity [km/h] when overtaking the other
cyclists, the lateral distance to the overtaken cyclists [m] and the location of the overtaking manoeuvre [m] in the scenario. As the
simulated cyclists appeared at the same location and cycled the same speed across all participants, the location of the overtaking
manoeuvre indicates how fast this simulated cyclist was approached by the participant. For the analyses, for each of the three scenarios
(road markings) a mean was computed for each participant over the three trials of this scenario.
The special event scenarios included a trafc light, a crossing pedestrian, and a road work blocking half of the cycle lane (see Fig. 2,
bottom). When the participant approached the trafc light, it turned yellow for about 1.8 s (s), and then red. The participant had to
wait for about 16 s until the trafc light rst became red and yellow and then green again, and the participant could start cycling. In
this scenario, we examined whether the secondary task led to differences in the time to standstill [s] from the moment the trafc light
became yellow before red, the distance [m] to the trafc light at standstill, and the reaction time [s] to start cycling when the trafc
light became yellow before green.
The crossing pedestrian rst walked on the footpath and then suddenly stepped onto the cycle lane in front of the participant so that
the participant had to move to the left to pass the pedestrian. As dependent variables in this scenario we examined the distance when
passing the pedestrian [m] and the speed when passing him [km/h]. As the same behaviour was required at the roadworks, we used the
same dependent variables to describe how the cyclist managed this situation. All special events occurred on a section with a cycle lane
marked with a solid white demarcation line, as this represents the typical and most often used type of cycle lane in Germany.
Before starting the test ride, participants could practice cycling and overtaking in a practice trial. The practice trial was like the
overtaking scenario but without cycle lane markings. The trial was explicitly instructed as a practice trial to the participants.
2.3. Secondary tasks
The study included three secondary task conditions: no task (NT), podcast task (PC), or acoustic speech task (AS). The NT condition
served as baseline. In the PC condition, participants chose one of three podcasts and listened to it during the test ride via Bluetooth
headphones. To ensure listening, they were informed that they would be quizzed on the podcast content after the test ride. In the AS
condition, participants listened to a voice via the Bluetooth headphones saying different colour names and a rule how to reorder the
colour names. The participants’ task was to say the correct new order of the colour names out loud. Thus, the AS involved acoustic
input, visual-spatial representation, and speech output. For example, if the participant heard “red, yellow, blue, yellow before red”,
they were to say the new correct order out loud which would be “yellow, red, blue”. The AS was originally developed and successfully
used in a study on secondary task engagement among drivers (see Vollrath & Totzke, 2005). To ensure engagement in the task, the
experimenter informed the participants that they would get feedback on their performance after the test ride.
2.4. Experimental design
The main independent variable was the type of secondary task comparing a control group (NT) to a group with a podcast task (PC)
and an acoustic speech task (AS) in a between-subjects design. For basic cycling behaviour and behaviour when overtaking the cyclist,
the effect of different lane markings (line marking, buffered, protected) was additionally examined as a second independent within-
subject variable. Each of these three markings was cycled three times in a xed but quasi-random order to prevent expectation effects
of the participants. The parameters were averaged over these three trials of each lane marking for each participant. The three special
events were inserted after the third, the fth and the last of these trials (see Fig. 3). The assignment of the specic event scenario to its
position in the xed order was done with the Latin square. For the three special scenarios (trafc light, crossing pedestrian and road
work) only the effect of secondary task could be analysed.
2.5. Procedure
The experimenter welcomed the participants and briefed them about the procedure. After agreeing to participation and data
privacy, participants were randomly assigned to one of the three secondary task conditions: NT, PC, or AS and introduced to the bicycle
Fig. 3. Sequence of scenarios: Practice trial (PR), overtaking (OT) and special event (SE).
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
526
simulator. On a test course, participants practised turning and braking, and on a city ride they practised handling the bicycle simulator
in trafc situations. In both familiarization rides, participants cycled as long as needed to feel comfortable cycling in the simulator.
Most participants needed about 20 min.
All participants cycled all scenarios as described above. The scenarios were connected via a short section on a residential street.
After about 300 m on the test segment, participants turned right into a residential street and right again into the next test segment.
Participants were informed that they could overtake the cyclist if they wanted to, but they did not have to. If they did not overtake,
they followed the simulated cyclist until turning right into the residential street. The participants were not informed about the
occurrence of the special events, thus, they behaved individually and spontaneously. The complete test ride lasted about 25 min.
Afterwards, participants lled out a demographic questionnaire and evaluated the overtaking scenarios and each of the event scenarios
regarding their perceived safety, cycling performance, and distraction on a 15-point Likert scale, respectively. In the PC condition,
participants then did the podcast content quiz. In both the PC and AS conditions participants received their scores based on the number
of correct answers. They were thanked and compensated for their participation. The complete experiment lasted about 1.5 h.
2.6. Participants
Participants were invited via mailing lists and online forums at the Technische Universit¨
at Braunschweig and via the participant
pool at the Department of Trafc and Engineering Psychology (a list of previous participants who agreed to be invited in future
studies). Participants should be able to ride a bike and know the German trafc regulations. They were compensated with
€
5 per half
hour, or by credit points (psychology students). The study was approved by the ethical commission at the Institute of Psychology at the
Technische Universit¨
at Braunschweig.
In total, 60 participants completed the study. One participant overtook the slow cyclist in very few situations and was excluded
from the sample due to a lack of comparability to all other participants, who overtook the cyclist in every trial. Due to technical
problems, data from one participant was missing. Thus, data from 58 participants (36 female, 22 male) were included in the analyses.
Participants’ age ranged from 18 to 43 years (M =23.5 years, SD =4.0 years). Most participants (76 %) cycled several times per week
or more. About half of the participants (52 %) listened to music via headphones often or always while cycling, whereas 36 % never did.
Other activities such as listening to podcasts or phoning (handheld or hands free) were rarely done while cycling by the participants.
2.7. Data analysis
The data analysis was done in several steps. To examine differences in the reaction to the special events, we ran multivariate
analyses of variance (MANOVAs) separately for each special event with the secondary task condition as between-subject factor.
To examine whether the secondary task inuences the overtaking behaviour and normal cycling behaviour, we ran two 3 x 3 mixed
MANOVAs with the secondary task as between subject factor and the marking type as within subject factor.
Similarly, the subjective ratings were analysed in a 3 x 4 mixed MANOVA with the secondary task as between subject factor, the
events (overtaking, trafc light, pedestrian, and work zone) as within subject factor, and as subjective safety, subjective performance,
and subjective distraction as dependent variables. All analyses ran with the statistic software IBM SPSS Version 28.
3. Results
3.1. The effect of secondary task on normal cycling behaviour
To test whether participants paid attention to the secondary tasks, we recorded their correct answers in the podcast quiz and in the
AS task. If participants gave no reply in the AS task, this was recorded as missing and, thus, incorrect reply. On average, participants in
the PC condition answered 7 out of 10 questions correctly. Participants in the AS condition replied in 79 % of trials correctly, in 14 %
incorrectly, and missed the reply in 7 % of trials. In general, we can assume that participants engaged in the two secondary tasks.
In the 3 x 3 mixed MANOVA we found a signicant main effect of marking in average speed and lateral position, but not in SDLP,
and no main effect for secondary task and no signicant interaction in any of the dependent variables. Table 1 shows the results of the
Table 1
Results of the MANOVA for the effects of secondary task and marking for normal cycling behaviour.
Source Measure Num df Denum df F p Partial
η
2
Secondary task Speed 2 55 0.481 0.621 0.017
SDLP 2 55 0.572 0.567 0.020
LP 2 55 0.027 0.973 0.001
Marking Speed
a
1.63 89.40 3.521 0.043 0.060
SDLP
a
1.69 93.04 1.860 0.167 0.033
LP 2 110 37.615 0<.001 0.406
Marking * Secondary task Speed
a
3.25 89.40 0.416 0.757 0.015
SDLP
a
3.38 93.04 0.486 0.715 0.017
LP 4 110 0.560 0.692 0.020
a
corrected by Greenhouse-Geisser correction.
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
527
univariate test of the MANOVA.
Fig. 4 shows the average speed and the mean lateral position for the three marking types during normal cycling. Cyclists seem to go
a bit faster on the track with the line and slowest with the buffer. However, the difference is only marginal, and the effect size is rather
weak. The stronger effect, in contrast, is found for lateral position. The stronger the separation from the cars, the more to the right the
cyclists go and thus keep more distance to this protective strip.
3.2. Effect of the secondary tasks on overtaking behaviour
The results of the 3 x 3 mixed MANOVA showed a signicant main effect of lane marking for the location when overtaking, for the
overtaking distance, and for the overtaking speed. No signicant main effect for the secondary task or interaction effect was found.
Table 2 shows the results of the MANOVA.
Fig. 5 shows the effect of lane marking for the location of overtaking, the overtaking speed and the overtaking distance for each
marking type. The other cyclist was being overtaken about 20 m earlier with the buffer and protected lane, but the speed when
overtaking was a little slower then as compared to overtaking with the line. The overtaking distance was closer on the protected lane,
which is similar to the effect of driving more to the right with the protected lane (see above). Cyclists seem to tend to keep more of a
distance from this protection.
3.3. Effect of secondary task on cycling behaviour in special events
When reacting to the trafc light, participants took 8.3 s on average to come to a standstill after the trafc light had turned yellow.
They came to standstill about 8.6 m in front of the trafc light, which is 4.2 m in front of the stopping line right before the trafc light.
When the light changed from red to yellow, it took them 1.7 s to start cycling. In the MANOVA, none of the measures regarding the
trafc light showed signicant differences between the three distraction groups. Thus, there was no effect found of engaging in these
secondary tasks as compared to cycling without it.
When passing a pedestrian, participants kept an average minimum passing distance of about 1.3 m and an average speed of 17 km/
h. The results of the MANOVA did not show signicant differences between the distraction types. Thus, neither the listening to the
podcast nor the acoustic speech task resulted in changes in cycling behaviour as compared to the control group.
The average minimum distance when passing the road work was 2.4 m and the average speed was 18 km/h. The MANOVA with
these two measures showed no signicant differences between the secondary task groups. Again, neither of the two acoustic tasks
changed cycling behaviour. Table 3 gives the results of the three MANOVAs, each with the parameters of the three special events.
3.4. Effect of secondary task on subjective measures
The results of the 3 x 4 mixed MANOVA (distraction group x event type) showed a signicant main effect for the scenario in
perceived safety, perceived performance, and perceived distraction. Furthermore, we found a signicant main effect for the secondary
task in perceived distraction. The interaction was not signicant. Table 4 shows the results of the MANOVA.
Fig. 6 shows the subjective ratings of safety, performance and distraction by scenario and distraction type. Perceived safety was
largest in the trafc light scenario, followed by the overtaking situation. Both the pedestrian entering the road and the work zone
reduced perceived safety. The effects of the situations on perceived quality of performance are comparable to that of perceived safety.
Concerning perceived distraction, it is interesting to see that both the cyclist to be overtaken as well as the pedestrian led to a larger
rating of perceived distraction. More importantly, both the AS task and the PC increased perceived distraction, with the acoustic speech
task leading to the largest perceived distraction.
4. Discussion
In a bicycle simulator, we aimed to examine the behavioural effect of engagement in hands-free cognitive secondary tasks while
Fig. 4. Average speed [km/h] and lateral position [m] on the road segment after overtaking the cyclists. Lateral position describes the deviation
from the centre with positive values referring to a deviation to the right. Error bars represent 95% condence intervals.
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
528
cycling focussing on normal cycling behaviour, overtaking, and reaction to sudden events, thus covering a broad range of possible
manoeuvres while cycling. We varied the complexity of the cognitive secondary task by using a podcast condition (PC) where the
participants only had to listen (and remember) to a podcast, thus basically requiring only speech perception. In the second condition,
the acoustic speech task (AS), participants had to understand an instruction about a spatial arrangement of three colours, had to
mentally recongure that and then present the answer verbally, thus requiring speech perception, mental visual processing, and
Table 2
Results of the MANOVA examining the effect of secondary task and marking for the overtaking behaviour.
Source Measure Num df Denum df F p Partial
η
2
Secondary task Location 2 55 0.867 0.426 0.031
Distance 2 55 0.446 0.642 0.016
Speed 2 55 0.901 0.412 0.032
Marking Location 2 110 228.505 0<.001 0.806
Distance 2 110 16.761 0<.001 0.234
Speed 2 110 3.738 0.027 0.064
Marking * Secondary task Location 4 110 1.627 0.172 0.056
Distance 4 110 0.310 0.871 0.011
Speed 4 110 0.393 0.813 0.014
Fig. 5. Location of overtaking [m], lateral distance to the overtaken cyclist [m], and speed when overtaking [km/h] separately for the three
marking types. Error bars represent 95% condence intervals.
Table 3
Results of the three MANOVAs examining the effect of distraction type in the three critical scenarios.
Measure Num df Denum df F p partial
η
2
Trafc light
Time to stand 2 56 1.530 0.225 0.052
Locations at stand 2 56 0.614 0.545 0.021
RT to start 2 56 0.230 0.796 0.008
Pedestrian
Min. passing distance 2 55 0.307 0.737 0.011
Passing speed 2 55 0.671 0.515 0.024
Road work
Min. passing distance 2 54 2.447 0.096 0.083
Passing speed 2 54 0.014 0.986 0.001
Table 4
Results of the MANOVA examining the effect of secondary task and scenario for the subjective ratings.
Source Measure df num df denum F p Partial
η
2
Secondary Task Safety 2 55 0.808 0.451 0.029
Performance 2 55 1.816 0.172 0.062
Distraction 2 55 18.591 0<.001 0.403
Scenario Safety 3 165 23.453 0<.001 0.299
Performance
a
2.50 137.57 20.605 0<.001 0.273
Distraction
a
2.57 141.39 7.349 0<.001 0.118
Scenario * Secondary Task Safety 6 165 0.344 0.912 0.012
Performance
a
5.00 137.57 0.527 0.756 0.019
Distraction
a
5.14 141.39 1.802 0.114 0.061
a
corrected by Greenhouse-Geisser correction.
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
529
speech production. This variation was successfully varying the amount of distraction at least from the subjective perspective of the
participants, as the subjective ratings of distraction showed. However, in none of the different scenarios and in none of the parameters
measured, an effect of this cognitive distraction could be found. In all conditions, participants cycled with a similar speed at a similar
position of the cycle way and reacted to events with a similar reaction time and in a similar manner with regard to longitudinal and
lateral control. Thus, on an objective or behavioural level, neither the simple nor the complex cognitive secondary task was found to
signicantly change cycling behaviour in its different facets examined here.
This result contrasts with some other studies assessing the effect of secondary task engagement on cycling behaviour who identied
detrimental effects such as poor visual scanning (e.g., de Waard et al., 2010, 2014; Jiang et al., 2021), reduced riding speed, greater
variation in lateral position, auditory signals being missed (e.g., de Waard et al., 2014) as well as an increase in crash/near crash
involvement (e.g., Terzano, 2013). However, unlike our study the previous studies mainly included handheld secondary task
engagement, or a comparison between handheld and hands-free phone use. The detrimental effects found in the earlier studies can be
explained by the multiple resource theory (Wickens, 2002) as the consequence of an overlap in the visual-motor resources needed for
dual-task performance.
The AS task was previously found to negatively inuence lane-keeping quality for car drivers (Vollrath & Totzke, 2005). However,
this was only the case during car following in curves, which was the study’s most demanding driving situation. Considering results
from the previous studies on hands-free secondary task engagement among drivers in general and the use of the AS task in particular (e.
g., Vollrath & Totzke, 2005; Dingus et al., 2016; Vollrath et al., 2021), it is likely that the lack of a behavioural effect of the cognitive
tasks was due to the fact that the cycling tasks encountered in our study were too simple. The overtaking and special events included in
the study were designed to realistically mirror normal everyday cycling while also creating some variation and potentially dangerous
events during the ride. The subjective assessment of the perceived safety and distraction in relation to the special events conrmed that
the special events were perceived as expected as the crossing pedestrian got the lowest score on perceived safety and the highest score
on perceived disturbance due to distraction from the secondary task, while the trafc light got the highest score on perceived safety and
the lowest score on distraction. From a psychological perspective, however, these cycling tasks are either on the skill-based level of
behaviour or on the rule-based level. They are, therefore, mainly performed automatically and do not require many cognitive re-
sources, which may explain the lack of impact on cycling behaviour.
In addition, although the secondary task engagement was not self-paced, it was not unexpected either. On the contrary, it was
ongoing throughout the ride and thus expected, thereby possibly not creating the perceived need for spare capacity to handle un-
expected input or engage in compensatory behaviour prior to voluntary task engagement or post system-paced secondary task
engagement identied by Kircher et al. (2015). Support for this interpretation is found in our results, which show that although the
three conditions (NT, PC, AS) varied in perceived distraction and safety, there was no difference in self-assessed performance level. In
other words, the level of distraction created by the secondary task engagement and the level of complexity or unexpectedness of the
riding task were possibly too low to inuence the self-assessed capacity to handle the bicycle, creating no need for behavioural ad-
justments. A follow-up study that includes a more complex and demanding trafc situation and more unexpected events could be
relevant.
As mentioned in the method section, we used cycling behaviour starting at 10 m after the overtaking as the baseline cycling
behaviour. We considered 10 m to be sufcient to stabilise and normalise cycling speed, lateral position, and steering after an
overtaking manoeuvre. It is possible, that the cycling behaviour might be inuenced after the overtaking manoeuvre, thereby
somehow inuencing the results. However, we believe that it is very unlikely.
Another possible explanation for the lack of effects could be that the behavioural parameters measured were not sensitive enough to
demonstrate negative effects on cycling. However, in line with previous studies (e.g., Bernardi & Rupi, 2015; Flügel et al., 2019;
Schleinitz et al., 2017) we found an effect of infrastructure design on these parameters in normal cycling behaviour as well as on
overtaking behaviour. Participants rode a bit faster when the cycle path was separated from the road by a line, and more to the right at
the protected bike lanes. This also led to a smaller distance when overtaking another cyclist at the protected bike lane. Thus, cyclists
seem to perceive the bollards of the protected bike lane as potential hazards to avoid making behavioural adjustments to reduce the
risk of collision. Besides this effect of infrastructure, it also shows that the behaviour parameters are sensitive to showing effects. From
this point of view, a lack of sensitivity is not the explanation for not nding negative effects of hands-free cognitive secondary tasks.
Fig. 6. Subjective rating for perceived safety, performance, and distraction by scenarios and secondary tasks.
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
530
Finally, no critical information about the trafc situation was provided auditively as our three critical situations were all created
based on visual information. Therefore, an increase in response time to auditory stop signals or overall perception of auditory cues
found in previous studies (e.g., de Waard et al., 2011; Stelling-Konsczak et al., 2017) could not be assessed due to the study design. In
addition, normal headphones, earbuds, etc., reduce the sound and noises from the environment so that cyclists cannot hear very well,
which may inuence behaviour in different ways. In our study, the trafc sounds were provided via the headphones. Hence, the
cyclists were not at all impaired in hearing the surrounding sounds but could cycle normally (not compensatory) which may also
contribute to the lack of cognitive strain and perceived need to make behavioural adjustments.
5. Conclusion
From the multiple resource theory of Wickens (2002) one would expect visual-manual secondary tasks to negatively inuence
cycling behaviour as both require the same cognitive resources. With cognitive distraction caused by acoustic stimuli and speech
reactions, this should only lead to negative effects on cycling if cycling also requires cognitive resources. For car driving, this pattern
was found in a study using an acoustic speech task (Vollrath & Totzke, 2005). Moreover, recent studies on telephoning while driving
also show no negative effects on crash risk but even reductions of crash risk when phoning in at least some types of crashes (Victor
et al., 2015; Young, 2017). Our study is the rst to examine cognitive distraction in a cycling simulator in different cycling scenarios. In
none of the scenarios and in none of the behaviour parameters, a negative effect of listening to a podcast or engaging in an acoustic
speech task was found, although these parameters were well able to show the effects of different infrastructures encountered. While
this may change if the cycling scenarios become very complex (e.g., involving different and multiple trafc participants, unclear
cycling situations etc.) at least in this standard cycling situation cognitive distraction does not lead to dangerous situations or even a
decrease in the quality of cycling behaviour. This is in line with the idea of multiple resources, which can be used in parallel for
different concurrent tasks if the two tasks use different resources. Thus, the focus of prevention and legal measures against distraction
while cycling should be on secondary tasks involving visual-manual distraction.
CRediT authorship contribution statement
Mette Møller: Writing – original draft, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis,
Conceptualization. Frauke Luise Berghoefer: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualiza-
tion. Mark Vollrath: Writing – review & editing, Supervision, Methodology, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to
inuence the work reported in this paper.
Data availability
The data that has been used is condential.
Acknowledgement
The research was supported by Otto Mønsteds Foundation, grant no. 22-70-1788.
References
Bernardi, S., & Rupi, F. (2015). An analysis of bicycle travel speed and disturbances on off-street and on-street facilities. Transportation Research Procedia, 5, 82–94.
https://doi.org/10.1016/j.trpro.2015.01.004
Brandt, R. K., Haustein, S., & Møller, M. (2021). Cyclists’ phone use in relation to proximate environmental characteristics – A qualitative study. Journal of Transport &
Health, 23, Article 101283. https://doi.org/10.1016/j.jth.2021.101283’
Christiansen, H. & Warnecke, M. L. (2018). Risiko i trakken 2007-2016 (in Danish). Available at: https://orbit.dtu.dk/les/146185353/Risiko_i_trakken_2007_16.
pdf.
De Angelis, M., Fraboni, F., Puchades, V. M., Prati, G., & Pietrantoni, L. (2020). Use of smartphone and crash risk among cyclists. Journal of Transportation Safety &
Security, 12, 178–193. https://doi.org/10.1080/19439962.2019.1591559
de Waard, D., Schepers, P., Ormel, W., & Brookhuis, K. (2010). Mobile phone use while cycling: Incidence and effects on behaviour and safety. Ergonomics, 53, 30–42.
https://doi.org/10.1080/00140130903381180
de Waard, D., Edlinger, K. M., & Brookhuis, K. A. (2011). Effects of listening to music, and of using a handheld and handsfree telephone on cycling behaviour.
Transportation Research Part F, 14, 626–637. https://doi.org/10.1016/j.trf.2011.07.001
de Waard, D., Lewis-Evans, B., Jlijs, B., Tucha, O., & Brookhuis, K. (2014). The effects of operating a touch screen smartphone and other common activities
performance while bicycling on cycling behaviour. Transportation Research Part F, 22, 196–206. https://doi.org/10.1016/j.trf.2013.12.003
Dingus, T. A., Guo, F., Antin, J. F., Perez, M., Buchanan-King, M., & Hankey, J. (2016). Driver crash risk factors and prevalence evaluation using naturalistic riving
data. PNAS, 113, 2636–2641. https://doi.org/10.1073/pnas.1513271113
Ethan, D., Basch, C. H., Johnson, G. D., Hammond, R., Chow, C. M., & Varsos, V. (2016). An analysis of technology-related distracted biking behaviors and helmet use
among cyclists in New York City. Journal of Community Health, 138–145. https://doi.org/10.1007/s10900-015-0079-0
FGSV (Forschungsgesellschaft für Straßen- und Verkehrswesen. (2010). Empfehlungen für Radverkehrsanlagen (ERA). FGSV: Arbeitsgruppe Straßenentwurf.
Flügel, S., Hulleberg, N., Fyhri, A., Weber, C., & Ævarsson, G. (2019). Empirical speed models for cycling in the Oslo road network. Transportation, 46, 1395–1419.
https://doi.org/10.1007/s11116-017-9841-8
M. Møller et al.

Transportation Research Part F: Psychology and Behaviour 104 (2024) 522–531
531
Goldenbeld, C., Houtenbos, M., Ehlers, E., & de Waard, D. (2012). The use and risk of portable electronic devices while cycling among different age groups. Journal of
Safety Research, 43, 1–8. https://doi.org/10.1016/j.jsr.2011.08.007
Huemer, A. K., Gercek, S., & Vollrath, M. (2019). Secondary task engagement in German cyclists – An observational study. Safety Science, 120, 290–298. https://doi.
org/10.1016/j.ssci.2019.07.016
Jiang, K., Yang, Z., Feng, Z., Sze, N. N., Yu, Z., Huang, Z., & Chen, J. (2021). Effects of using mobile phones while cycling: A study from the perspectives of
manipulation and visual strategies. Transportation Research Part F, 83, 291–303. https://doi.org/10.1016/j.trf.2021.10.010
Kircher, K., Ahlstrom, C., Palmqvist, L., & Adell, E. (2015). Bicyclists‘ speed adaption strategies when conducting self-paced vs. system-paced smartphone tasks in
trafc. Transportation Research Part F, 28, 55–64. https://doi.org/10.1016/j.trf.2014.11.006
Kircher, K., Ihlstr¨
om, J., Nygårdhs, S., & Ahlstrom, C. (2018). Cyclist efciency and its dependence on infrastructure and usual speed. Transportation Research Part F,
54, 148–158. https://doi.org/10.1016/j.trf.2018.02.002
O’Hern, S., Oxley, J., & Stevenson, M. (2017). Validation of a bicycle simulator for road safety research. Accident Analysis and Prevention, 100, 53–58. https://doi.org/
10.1016/j.aap.2017.01.002
Parnell, K. J., Rand, J., & Plant, K. L. (2020). A diary study of distracted driving behaviours. Transportation Research Part F, 74, 1–14. https://doi.org/10.1016/j.
trf.2020.08.003
Schleinitz, K., Petzoldt, T., Franke-Bartholdt, L., Krems, J., & Gehlert, T. (2017). The German naturalistic cycling study – comparing cycling speed or riders of different
e-bikes and conventional bicycles. Safety Science, 92, 290–297. https://doi.org/10.1016/j.ssci.2015.07.027
Stelling-Konczak, A., van Wee, G. P., Commandeur, J. J. F., & Hagenzieker, M. (2017). Mobile phone conversations, listening to music and quiet (electric) cars: Are
trafc sounds important for safe cycling? Accident Analysis and Prevention, 106, 10–22. https://doi.org/10.1016/j.aap.2017.05.014
Terzano, K. (2013). Bicycle safety and distracted behaviour in The Hague, the Netherlands. Accident Analysis and Prevention, 57, 87–90. https://doi.org/10.1016/j.
aap.2013.04.007
Useche, S. A., Alonso, F., Montoro, L., & Esteban, C. (2018). Distraction of cyclists: How does it inuence their risky behaviors and trafc crashes? PeerJ, 6, e5616.
Useche, S.A., Alonso, F., Boyko, A, Buyvol, P., D. Castaneda, I., Cendales, B., Cervantes, A., Echiburu, T., Faus, M., Gene-Morales, J., Gnap, J., Gonzalez, V., Ibrahim,
M.K.A., Janstrup, K.H., Makarova, I., Mikusova, M., Møller, M., O’Hern, S., Orozco-Fontalvo, M., Shubenkova, K., Siebert, F.I., Soto, J.J., Stephens, A.N., Wang,
Y., Willberg, E.S., Wintersberger, P., Zeuwts, L., Zulkipli, Z.H., & McIlroy, R.C.(2024). Yes, size does matter (for cycling safety)! Comparing behavioral and safety
outcomes in S, M, L, and XL cities from 18 countries. Journal of Transport Geography, 114, 103754. DOI: 10.1016/j.jtrangeo.2023.103754.
Victor, T., B¨
arman, J., Boda, C.N., Dozza, M., Engstr¨
om, J., Flannagan, C., Lee, J.D., & Markkula, G. (2015). Analysis of Naturalistic Driving Study Data: Safer Glances,
Driver Inattention, and Crash Risk. SHRP 2 Safety Project S08A. https://dx.doi.org/10.17226/22297.
Vollrath, M., & Totzke, I. (2005). Secondary tasks while driving – effects and countermeasures. Advances in Transportation Studies - An International Journal Section B, 7,
67–80. Microsoft Word - Vollrath_ats.doc (researchgate.net.
Vollrath, M., Huemer, A. K., Teller, C., Likhacheva, A., & Fricke, J. (2016). Do German drivers use their smartphones safely? – Not really. Accident Analysis and
Prevention, 96, 29–38. https://doi.org/10.1016/j.aap.2016.06.003
Vollrath, M., Clifford, C., & Huemer, A. K. (2021). Even experienced phone users drive worse while texting – A driving simulator study. Transportation Research Part F,
78, 218–225. https://doi.org/10.1016/j.trf.2021.02.007
Wickens, C. D. (2002). Multiple resources and performance prediction. Theoretical Issues in Ergonomics Science, 3, 159–177. https://doi.org/10.1080/
14639220210123806
Woodward, A., & Samet, J. (2016). Active transport: Exercise trumps air pollution, almost always. Preventive Medicine, 87, 237–238. https://doi.org/10.1016/j.
ypmed.2016.03.027
Würzburg Institute for Trafc Sciences GmbH (WIVW), Simulatorsoftware, SILAB 7.0, 2021, Veitsh¨
ochheim, Germany.
Xia, T., Zhang, Y., Crabb, S., & Shah, P. (2013). Cobenets of replacing car trips with alternative transportation: A review of evidence and methodological Issues.
Journal of environmental and public health. https://doi.org/10.1155/2013/797312. Article ID 797312.
Young, R. (2017). Removing Biases from Crash Odds Ratio Estimates of Secondary Tasks. A New Analysis of the SHRP 2 Naturalistic Driving Study Data (No. 2017–01-
1380). SAE Technical Paper.
M. Møller et al.