Content uploaded by Simone Morgagni
Author content
All content in this area was uploaded by Simone Morgagni on Jun 08, 2022
Content may be subject to copyright.
1
WAITING TIME PERCEPTION AND REAL-TIME SCHEDULE DISPLAY
Emile COLIN1, Guillaume LEMAITRE1, Simone Morgagni1, Virginie VAN WASSENHOVE2
1SNCF DTIPG, Saint-Denis, France
2CEA Neurospin, Saclay, France
Corresponding Author: Guillaume Lemaitre (Guillaume.lemaitre@sncf.fr)
Abstract
In mass transit areas, passengers usually wait for their train for a few minutes. There, the arrival time is usually
forecasted on passenger information systems as a countdown, resulting both from the theoretical timetables
and the real-time tracking of trains on the network. However, because the precision of the tracking is not infinite,
small discrepancies may occur between the arrival time forecasted on these displays and the duration that
passengers actually wait, thus creating expectations in passengers that may be broken. The current study set
out to investigate how these discrepancies may affect passengers’ passage of time judgements (i.e. whether
time seems to flow slowly or fast). This on-line experiment simulated the waiting of a train on a train platform.
We manipulated the forecast for the arrival time of the next train, which could be indicated with certainty (“2
min”) or with a temporal uncertainty interval (“3 ± 1 minutes”) and the actual arrival of the train which could
arrive as predicted or earlier or later than the forecast. The results suggest that providing temporal uncertainty
with the forecast can make the passage of time feel faster when the train was late than when the forecast was
indicated with certainty. Overall, these results suggest that optimizing the way the forecast of the next train
arrival time is provided to passengers (by softening temporal expectations) could be used as a simple yet
effective tool to alleviate passenger’s experience of waiting on train platforms.
Keywords: Passage of time, waiting time, mass transit, passenger information systems
1. Introduction
Standing on a train platform, passengers often consult passenger information display systems to estimate how
long they must wait for the next train to come in. There, small discrepancies may occur between the arrival time
forecasted on these displays and the actual waiting time they will experience. This situation is common in mass
transit areas where departure times are displayed as minutes to come (i.e. countdown to next train) and not as
timetables. This can generate frustration among passengers as they may feel that a promise (a forecasted
waiting time) has been made and broken.
These discrepancies originate from technical limitations that are costly to address (tracking the position of the
train with particularly precise real-time updates). On suburban train lines in the Parisian area, the next train
arrival time (NTAT) displayed on passenger information system is forecasted in real time both from the
theoretical timetables and the actual position of the train, measured each time a train passes a beacon. For
example, when a train passes a beacon when leaving a station, a countdown is started in the next station, based
on the timetable. If the train is stopped or slowed down between beacons, the countdown will continue without
being updated, causing the last minute to stretch until the train finally reaches the next beacon. Other types of
discrepancies may occur depending on the density of beacons and the precise update strategies, resulting in a
countdown whose minutes may slightly stretch or shrink.
The study reported here originated from some initial observations suggesting that this behaviour of the
information systems may make passengers feel that time drags on or speeds up. The feeling of time speeding
up or slowing down is typical of many other everyday situations. For example, waiting is a situation that makes
time to pass more slowly (“a watched kettle never boils”) [1] [2]: this is in part because paying attention to time
makes it drag while distracting attention away from it makes it go faster [3]. Regarding more specifically waiting
in a public transport environment, waiting time is a good predictor of passengers’ satisfaction, with,
understandably, satisfaction being correlated with shorter waiting times [2], and especially the perceived or
subjective, rather than objective, waiting time [4].
2
The perception of time has been studied scientifically by multiple disciplines, including psychology, cognitive
sciences, and neurosciences [5]. Regarding our question, it is important to distinguish between passage of time
judgments (POTJ, i.e. the subjective speed at which time passes during a given time interval) and the subjective
estimation of duration (duration judgements DJ) [6] [7] [8]. POTJ and DJ have been found to be uncorrelated in
depressed patients [6], and direct comparisons of POTJ and DJ for durations of less than 2 minutes have found
no relationship between duration judgments and passage of time judgments for shorter durations [9], yet POTJ
and DJ seem to correlate more closely for durations between two and eight minutes [7]. Overall, these results
suggest that different mechanisms may be involved both in POTJ and DJ depending of the duration considered
(e.g. memory, internal clock models [10]).
Our project chose to focus on passage of time judgements, which are easier to relate to everyday
phenomenology of time and seems to correlate with passengers’ satisfaction. POTJ may be affected by different
factors. For example, time passes faster when attention is drawn away from the temporal aspects of a situation
[11]. The complexity of a task and its novelty are other predictors of POTJ [12]. More importantly for our project,
expectation appears to be an important factor affecting the perception of time [13]. For example, using
instructions to manipulate the time a person expects to wait was shown to influence POTJs: longer differences
between the instructed (i.e. expected) waiting time and the actual waiting time made POTJs slower, for durations
of a few minutes [14] [15].
Coming back to the practical situation of waiting on a train platform, this strongly suggests that the current
strategies used in NTAT may create a feeling in passengers that waiting times slows down, which in turn would
decrease their satisfaction.
This also suggests that working out how passenger information systems display the next train arrival time (NTAT)
could be used to alleviate the waiting time experience. For example, if the display could relax the temporal
expectation in passengers regarding the waiting time, these expectations would be less likely to be broken. The
goal of the work was precisely to explore alternative display strategies that do not create such discrepancies,
and study whether such strategies may improve the waiting experience.
To this aim, we conducted an online experiment akin to a video game in which participants waited on a virtual
platform for a train to arrive. The design of the video game made the waiting time frustrating. Next train arrival
time (NTAT) was indicated on a display, and some anecdotical pieces of information (“fun facts”) could be
presented next to the passenger information display. Critically, the display could indicate either a certain
duration (i.e. “2 minutes”) or qualify the precision of the estimation with an uncertainty interval (i.e. “3 ± 1
minutes”). The choice of these strategies resulted from pre-tests that had selected these options as producing
the most contrasted effects between different ways to indicate uncertainty (verbal indications, numeric
symbols, graphical representation). In the experiment, the train could arrive as predicted by or earlier, or later
than the forecasted arrival time. The experiment manipulated three variables: the actual arrival time of the train
(resulting in the train being early, on time, or late), the certainty of the forecast, and the presence or absence of
distraction (fun facts distracting participants’ attention) from time. The goal of the experiment was to test the
working hypothesis that distracting attention away from time and providing the uncertainty of the forecast
would yield faster POTJ when the train arrives later than the forecasted arrival time.
2. Materials and methods
2.1 Participants
228 participants (147 female, 80 male participants, mean age = 33.6 +/- 0.8 years old), between 18 and 67 years
old were recruited trough Prolific, an online recruitment platform
1
. All of them were native English speakers.
Participants were fully informed of the task requirement beforehand and signed an online consent form to take
part in the study. The experiment was approved by the Comité Ethique de la Recherche at the Université Paris-
Saclay (CER-18-CE22-0016-01 to V.vW.). Participants were randomly assigned to one of four groups, each group
corresponding to each of the two NTAT display conditions and the two distraction conditions (see below).
1
https://prolific.co/, last retrieved on January 2022
3
2.2 Procedure
The experiment was conducted online through the Gorilla website
2
. It alternated “game” and “waiting” phases.
In the game phases (lasting 3 minutes), participants had to identify items such as logos, animal species, tools,
flags, actors/actresses and give their answer by keyboard free input. A correct answer scored one point, and
they had to answer as fast as possible to get the highest possible score.
In the waiting phase participants watched a video of a train platform based on 3D model of the Saint-Michel
train station in Paris. The platform was empty of any passengers. The videos included always included a NTAT
display at the upper left corner of the video. There were two different NTAT displays, each assigned to a different
group of participants (“certain” or “uncertain”). A lower left panel displayed entertaining “fun facts” for half of
the participants, aiming at distracting their attention away from the NTAT (“Distracted” vs. “Attending” groups,
see Figure 1).
Figure 1. Screenshot of a video used in the experiment when a train boards the station, showing the NTAT display
in the upper left corner, and the “fun facts” panel in the lower left panel.
The task of the participants during the waiting phase was to count how many flashes would occur on screen. For
the “Attending” group, the flashes were located on the NTAT panel. For the “Distracted” group, the flashes were
located on the “fun facts” panel. The flashes and their location were designed to ensure that participants were
attentive during the waiting phase, and that their attention was spatially focused to the panel corresponding of
their group. We warned them that an incorrect number of red flashes would reset their score accumulated
during the game phases. Because they could only earn points during the game phase, any delay in the waiting
phase (caused for example by the train being late) reduced their ability to score points.
At the beginning of each trial, the display indicated that the forecasted arrival of next train was 2 minutes, either
stated with certainty (i.e. “2 minutes”) or qualified with an uncertainty interval (i.e. “3 ± 1 min”) and was update
every minute as time went by. After a certain amount of time (1 to 5 minutes, different at each trial, in random
order), a train arrived in the station: when the train arrived after 1 minute, it was early with respect to the initial
forecast, when it arrived after 2 minutes it was on time, when it arrived after 3 minutes it was late, etc. After
the train had arrived participants then judged the passage of time on a visual-analog scale: “During the task,
how fast did the time seem to be passing by?” (0 = “time passed very slowly”, 100 = “time passed very quickly”).
2.3 Stimuli
The experiment used a set of 200 videos: 5 arrival times (1 to 5 minutes) x 2 NTAT displays (certain vs.
uncertain forecast) x 10 flashing patterns (a different pattern per participant), with or without distraction. The
“fun facts” consisted of different panels of text, changing automatically over time.
For the game phases we selected 50 royalty-free images corresponding to a theme (vegetables, animals, tools,
flags, actors/actresses). The game phases were the same for all experimental conditions.
The videos included a NTAT display at the upper left corner of the video of the station environment. There
were two conditions of precision for the forecasted NTAT: certain (i.e. 2 min,) or uncertain (3 ± 1 min, see
2
https://gorilla.sc/, last retrieved on January 2022
4
Figure 2). The forecasted NTAT was updated every minute and would freeze at “1 minute” when the actual
train arrival time exceeded what was displayed (i.e. was late). Importantly, the number of visual changes in the
video (flashes and updates of NTAT) was kept constant for each across videos of the same durations. This
implies that the number of visual changes was the same across the factors Distraction and Display (see below),
thus avoiding a confounding effect of contextual changes [16].
Figure 2. The two different ways to display the forecasted next train arrival time used in the experiment. Left:
certain forecast. Right: uncertain forecast.
3. Results
The analysis first excluded 37 participants that had made more than one error when counting the flashes
during the waiting phase (16 %). We considered the NTAT Display (certain or uncertain) and the presence or
absence of Distraction as within-participant factors, and the Arrival time (1 to 5 minutes, corresponding to the
train being early, on time, or late with respect to the forecast) as a within-participant factor.
Because the exclusion of participants resulted in groups with unequal sizes and distributions that did not meet
the assumptions of parametric tests, we used a set of Wilcoxon rank sum tests to test the effect of the
different factors. We found no main effects of Display (W=99517, p=0.31) or Distraction on PoTJ (W=107468,
p=0.47). Arrival Time significantly affected PoTJ: each comparison of two successive arrival times yielded a
significant Wilcoxon test (p<0.01). Detailed analyses showed that whereas the comparisons of certain and
uncertain displays yielded non-significant tests in the absence of Distraction for every Arrival time (p>0.05),
Figure 3. Passage of time judgements (POTJ) as a function of the train arrival time and the type of display
used to indicate the forecasted arrival time (certain or uncertain). The dashed line represents the value POTJ
that corresponds to “neither slow or fast”.
5
this was not the case when there was Distraction and the train arrived after 3 minutes: the uncertain display
resulted in faster POTJs than the certain display. (W=645, p<0.05)
Erreur ! Source du renvoi introuvable. shows that the passage of time slowed down as the participants waited
longer for the train to arrive. The POTJs were about 50 (i.e. neither slow or fast) when the train arrived after 2
minutes, i.e. when it was on time with respect to the forecast. PoTJ were faster (>50) when the train arrived
early, and slower (<50) when the train arrived later than this forecast. This pattern is consistent with the idea
that POTJ are influenced by the discrepancy between participants’ expectations and their actual waiting
experience. This is also consistent with the idea that POTJ are correlated with perceived duration, since the
two factors are confounded in this experiment.
Looking more closely, the POTJ was thus faster for the uncertain display than for the certain display only when
the train arrived after 3 minutes (i.e. one minute late) and only was attention was drawn away from time by
the flashing fun facts. In other words, this result suggests that qualifying the forecast of the NTAT (by qualifying
the certainty of the forecast) may modulate the passage of time experienced by participants: when the train
arrived late compared to the forecast, qualifying the forecast made time passed less slowly than when the
forecast was indicated as certain, when attention was drawn away from time.
5. Discussion
The initial idea motivating this study was that the discrepancy between the expectations about the next train’s
arrival time and the actual duration waited by passengers on a train platform would cause the time to pass
more slowly or faster. Overall, the results of this study are consistent with this idea: participants reported that
time passed more slowly when they waited more than forecasted (i.e. POTJ >50), about normal when the
waited time corresponded to the forecast (POTJ = 50), and faster when they waited less than forecasted (POTJ
< 50). Additional work will be conducted to disentangle such effect from a mere effect of duration [8] or of the
number of flashes (i.e. change of visual information [16]) occurring during the waiting period, as these factors
were confounded in our design.
The results are also consistent with the initial hypothesis that the way the forecast is indicated could modulate
the passage of time: when participants’ attention was drawn away from the temporal information and the
train was 1-min late with respect to the forecast, qualifying the forecast with an uncertainty interval (“3 ± 1
min”) made participants feel that the time passed faster than when it was indicated with certainty (“2 min”).
Because the POTJs for the uncertain forecast were not different from those for the certain forecast (POTJs
were about 50 for the uncertain forecast both for a 2-min and 3-min waiting period), this suggests that the
“uncertain” forecast did not simply shift participants’ expectation towards 3-min, but rather expanded their
expectations : a train arriving after 3 minutes had the same effect as a train arriving after 2 minutes. The effect
of distraction was more puzzling. Conversely to previous studies [2], we did not show an overall effect of
distraction, and the effect of qualifying the forecast with uncertainty had an effect only when attention was
drawn away from the forecast. Maybe, this prevented participants to update their expectations over time and
reinforced the effect of the initial glance they may had had to the display.
Despite their limitations, these results show that the design of information displays may have an impact on
passenger’s perception of waiting time. This result has important practical implications for railway operators.
Whereas improving the precision of forecast would require costly technical solutions, our results show that
simply optimizing how information is displayed on passengers’ information systems could improve passengers’
experience of waiting on train platforms. Further work is of course required to confirm these results, work out
optimized displays, test these displays in real operational situations, and understand the relationship between
passage of time and customers’ satisfaction, but such work offer a rather economical solution for railway
operators to improve their passenger’s overall experience of railway journeys.
Acknowledgment
6
This project was funded by the Agence Nationale de la Recherche (ANR) under the grant ANR-18-CE22-0016
(Wildtimes project). The authors would like to thank Chloé Rémy for the design of the displays and Allan
Armougum for the 3D models of the station, and the Wildtimes consortium for their numerous inputs (Yvan
Nédelec in particular).
References
[1] E. A. Larson, “The time of our lives: The experience of temporality in occupation,” Can. J. Occup. Ther.,
vol. 71, no. 1, pp. 24–35, 2004, doi: 10.1177/000841740407100107.
[2] M. Van Hagen, M. Galetzka, and A. T. Pruyn, “Waiting Experience in Railway Environments,” J. Motiv.
Emot. Personal. Reversal Theory Stud., vol. 3, no. 1, pp. 41–55, 2014, doi: 10.12689/jmep.2014.305.
[3] I. Polti, B. Martin, and V. Van Wassenhove, “The effect of attention and working memory on the
estimation of elapsed time,” Sci. Rep., vol. 8, no. 1, pp. 1–11, 2018, doi: 10.1038/s41598-018-25119-y.
[4] A. Pruyn and A. Smidts, “Effects of waiting on the satisfaction with the service: Beyond objective time
measures,” Int. J. Res. Mark., vol. 15, no. 4, pp. 321–334, 1998, doi: 10.1016/s0167-8116(98)00008-1.
[5] P. Fraisse, Psychologie du temps. Presse Universitaires de France, 1967.
[6] S. Thönes and D. Oberfeld, “Time perception in depression : A meta-analysis,” J. Affect. Disord. J., vol.
175, pp. 359–372, 2015.
[7] S. Droit-Volet, P. Trahanias, and M. Maniadakis, “Passage of time judgments in everyday life are not
related to duration judgments except for long durations of several minutes,” Acta Psychol. (Amst)., vol.
173, pp. 116–121, 2017, doi: 10.1016/j.actpsy.2016.12.010.
[8] J. H. Wearden, “Passage of time judgements,” Conscious. Cogn., vol. 38, pp. 165–171, 2015.
[9] S. Droit-Volet and J. Wearden, “Passage of time judgments are not duration judgments: Evidence from
a study using experience sampling methodology,” Front. Psychol., vol. 7, p. 176, 2016.
[10] J. Wearden, “Origins and development of internal clock theories of psychological time,” Psychol. Fr., vol.
50, no. 1, pp. 7–25, 2005, doi: 10.1016/j.psfr.2004.10.002.
[11] J. Wearden, A. O’Donoghue, R. Ogden, and C. Montgomery, “Subjective duration in the laboratory and
the world outside,” in Subjective Time: The Philosophy, Psychology, and Neuroscience of Temporality, V.
Arstila and D. Lloyd, Eds. The MIT Press, 2014, pp. 287–306.
[12] E. Larson and A. Von Eye, “Predicting the perceived flow of time from qualities of activity and depth of
engagement,” Ecol. Psychol., vol. 18, no. 2, pp. 113–130, 2006, doi: 10.1207/s15326969eco1802_3.
[13] D. Cahoon and E. M. Edmonds, “The watched pot still won’t boil: Expectancy as a variable in estimating
the passage of time,” Bull. Psychon. Soc., vol. 16, no. 2, pp. 115–116, 1980, doi: 10.3758/BF03334455.
[14] R. Tanaka and Y. Yotsumoto, “Passage of time judgments is relative to temporal expectation,” Front.
Psychol., vol. 8, p. 187, 2017.
[15] Y. Nedelec, “The experience of Time in train travel: Duration Estimation or Passage of Time Judgement?,”
Master’s thesis, Université Paris Descartes, Paris, 2019.
[16] P. A. Vroon, “Effects of presented and processed information on duration experience,” Acta Psychol.
(Amst)., vol. 34, no. C, pp. 115–121, 1970, doi: 10.1016/0001-6918(70)90010-7.
