Conference PaperPDF Available

Interrupted by a Phone Call: Exploring Designs for Lowering the Impact of Call Notifications for Smartphone Users


Abstract and Figures

Mobile phones have evolved significantly in recent years from single-purpose communication devices to multipurpose computing devices. Despite this evolution, the interaction model for how incoming calls are handled has barely changed. Current-generation smartphones still use abrupt full-screen notifications to alert users to incoming calls, demanding a decision to either accept or decline the call. These full-screen notifications forcibly interrupt whatever activity the user was already engaged in. This might be undesirable when the user’s primary task was more important than the incoming call. This paper explores the design space for how smartphones can alert users to incoming calls. We consider designs that allow users to postpone calls and also to multiplex by way of a smaller partialscreen notification. These design alternatives were evaluated in both a small-scale controlled lab study as well as a large-scale naturalistic in-the-wild study. Results show that a multiplex design solution works best because it allows people to continue working on their primary task while being made aware that there is a caller on the line. The contribution of this work is an enhanced interaction design for handling phone calls, and an understanding of how people use it for handling incoming calls.
Content may be subject to copyright.
Interrupted by a Phone Call: Exploring Designs for Lower-
ing the Impact of Call Notifications for Smartphone Users
Matthias Böhmer
, Christian Lander
, Sven Gehring
, Duncan Brumby
, Antonio Krüger
DFKI GmbH, Saarbrücken, Germany;
UCL Interaction Centre, London, UK
Mobile phones have evolved significantly in recent years
from single-purpose communication devices to multi-
purpose computing devices. Despite this evolution, the in-
teraction model for how incoming calls are handled has
barely changed. Current-generation smartphones still use
abrupt full-screen notifications to alert users to incoming
calls, demanding a decision to either accept or decline the
call. These full-screen notifications forcibly interrupt what-
ever activity the user was already engaged in. This might be
undesirable when the user’s primary task was more im-
portant than the incoming call. This paper explores the de-
sign space for how smartphones can alert users to incoming
calls. We consider designs that allow users to postpone
calls and also to multiplex by way of a smaller partial-
screen notification. These design alternatives were evaluat-
ed in both a small-scale controlled lab study as well as a
large-scale naturalistic in-the-wild study. Results show that
a multiplex design solution works best because it allows
people to continue working on their primary task while be-
ing made aware that there is a caller on the line. The contri-
bution of this work is an enhanced interaction design for
handling phone calls, and an understanding of how people
use it for handling incoming calls.
Author Keywords
Smartphones; app usage; phone calls; interruptions.
ACM Classification Keywords
H.5.m Information interfaces and presentation (e.g., HCI):
People’s smartphones are used to support a large variety of
activities and tasks. Indeed, some have gone so far as to
suggest that the smartphone will become the primary com-
puter of choice for many users [28]. People can use their
smartphones for a variety of work and leisure activities,
from processing email and managing appointments in their
calendar, to listening to music and surfing the web. The
smartphone is, however, still fundamentally a tele-
phone. When a user receives a call, current-generation
smartphones tend to notify the user with a full-screen visual
notification. This notification abruptly forces the user to
stop whatever task they were previously occupied with and
attend to the call. For instance, a user might be partway
through entering the time and location of a meeting to a
calendar from an email. When a call is received, this activi-
ty must be suspended and returned to after deciding how to
handle the call. During that time, the user might have for-
gotten the location of the meeting and so have to look it up
again that is, if they remember to complete the task at all.
This vision of how people interact with their smartphones is
supported by the results of a recent large-scale in-the-wild
study: Leiva et al. [16] analyzed data from several thousand
users over an 18-month period. They found that smartphone
users are rarely interrupted by phone calls while they are
using other apps (at most 10% of daily app usage). But
when they are interrupted, it is massively disruptive and
increases the time it takes users to complete the task they
were working on prior to dealing with the call. Given the
disruption caused by incoming phone calls, we consider
whether there is potential for revisiting the design space for
how they are handled by smartphones.
When we consider how smartphones notify users of incom-
ing calls, it is quite clear that the basic interaction model
has not changed since the development of early mobile
phones. Figure 1 juxtaposes a Panasonic mobile phone from
circa 1999 (on the left) beside current generation phones. It
is clear that apart from the fact that hardware buttons have
been replaced with touchscreen buttons, the basic interac-
Figure 1. Only slight evolution occurred in phone call apps while
mobile phones evolved from mere phones to multifunctional tools.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. Copyrights for com-
ponents of this work owned by others than ACM must be honored. Ab-
stracting with credit is permitted. To copy otherwise, or republish, to post
on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from
CHI 2014, April 26 - May 01 2014, Toronto, ON, Canada
Copyright 2014 ACM 978-1-4503-2473-1/14/04$15.00.
tion model is the same: the user is alerted to an incoming
call and has to decide whether to accept it or decline it. This
is despite the fact that the current generation of
smartphones affords far greater functionality and support
for tasks than its earlier predecessor. Current call answering
screens allow declining calls with text messages, or setting
up reminders (see Figure 1). While this helps in catching up
with declined calls later, it still leaves the problem of full-
screen alerts interrupting a user’s concurrent app usage.
This paper revisits the design of mobile phone call UIs with
the goal to better handle interruptions caused by incoming
phone calls during app usage. We focus on cases where the
user is interrupted by an incoming phone call while they are
engaged in another ongoing activity on their smartphone.
We make three contributions: First we extend the activities
of call handling, explore the design space, and present two
implementations in that space. Second, we report the results
of a controlled lab study that evaluates the effectiveness of
these various design alternatives for mitigating the effects
of call interruptions. Finally, we describe a large-scale in-
the-wild study that was conducted following the release of a
call-handling app to an app store. From this study we
learned how our design was used in natural contexts.
There is a long tradition in the HCI community of studying
the effect that interruptions, such as handling an incoming
phone call, have on task performance [9,17,20]. We review
this work along with work that has sought to develop smart
systems to handle calls better. Finally, we discuss what few
attempts there have been to develop commercial apps to
tackle this problem.
Interruptions Are Disruptive
It is well understood that interruptions disrupt ongoing ac-
tivities and take time to recover from. Memory for goals
[3,27] has emerged as an important theoretical framework
for understanding how people re-engage with a task follow-
ing an interruption. The theory assumes that people use
their memory as well as salient cues in the environment to
help reconstruct what it was they were doing prior to being
interrupted. This process takes time, and is referred to as a
resumption lag.
There is evidence that incoming calls incur a significant
resumption lag for smartphone users. Leiva et al. [16] found
that after finishing a phone call, it took users up to 40 se-
conds longer to finish the task they were working on prior
to dealing with the call. This observation led Leiva et
al. [16] to suggest that current-generation smartphones
should move away from using an immediate full-screen
notification to signal an in-coming call, because this does
not give the user any time to prepare for the interruption.
Instead Leiva et al. suggest that a gradual overlay notifica-
tion should be used. The idea is that this would give the
user time to prepare for the call. Consistent with this idea,
Iqbal et al. [13] found that call pre-alerts can reduce the
impact that incoming calls have on driving performance
presumably because this gives people an opportunity to
think about whether or not it is a good time to take the call.
There is good evidence to suggest that given greater flexi-
bility and choice people will choose to defer an interrupting
task until after a task (or a subtask) has been completed
[5,14,17,20]. For instance, Fischer et al. [8] show that most
opportune moments for mobile interruptions are after epi-
sodes of device usage (e.g., after sending an SMS message).
Hence, if a user is working on their smartphone, they might
prefer to be given the opportunity to defer an incoming call
until after they have completed the task they are working
on. Some reports suggest that up to 30% of calls are missed,
often for intentional reasons [19]. Hence, there is scope to
consider alternative design options for handling calls that
reduce the level of demand on the person receiving a call.
Making Calls Less Disruptive
There has been a large body of work that has attempted to
design systems to reduce the disruption caused by incoming
phone calls. There have been numerous attempts to build
adaptive notification systems that manage when calls are
allowed. Iqbal et al.’s OASIS [14] system holds non-urgent
computer alerts until periods when users are interruptible.
Ter Hofte [26] has applied this idea to managing telephone
calls by building a predictive model that blocks calls to us-
ers when they are actively engaged in an activity. In a simi-
lar vein, Ho and Intille [12] presented a sensor-based strate-
gy for delaying call interruptions that are not time-sensitive
until a physical activity transition. Stamm et al. [23] have
also developed a system that calculates the cost of interrup-
tions on mobile phones so that they can be better scheduled.
However, they argue that modeling the scheduling of phone
call interruptions is not easy because of the synchronous
nature of telephone communication.
The sharing of context information has been put forward as
one way to overcome the issues imposed by the synchro-
nous nature of telephone communication. ContextCalls was
an early system proposed by Schmidt et al. [21] to mitigate
the problem of call interruptions by making the callee’s
context transparent to the caller and vice versa. Taking a
similar approach, TellingCalls by Granndhi et al. [10] con-
veys information about the call between caller and callee
(e.g. the call’s topic). Knittel et al. [15] have developed a
system that augments the personal address book in a phone
with information to help people make an informed decision
about whether they should call (and maybe interrupt) the
callee. Indeed, this idea has been realized in voice-over IP
systems, such as Skype, which allow users to make explicit
their availability. Ironically, though, Teevan and Hehmeyer
[25] found that people are actually more likely, rather than
less likely, to accept a call when their status is “busy” or
“do not disturb”. A possible explanation for this might be a
self-selection bias, such that only important calls are initiat-
ed to people with a busy status. Regardless, there is still the
problem that users are not very good about updating their
status with such context sharing systems.
Commercial Applications
There are a few commercial apps available on the app mar-
ket to help user manage calls better (e.g. SmallCall, A+
Call Manager). Simply silencing an incoming call has be-
come a feature on many devices. However, no scientific
insights have been generated from these apps. This paper
explores a more comprehensive design space going beyond
the isolated solutions of these apps.
In summary, little is known about the problem of task inter-
ruptions with the primary task running on the phone. This is
what this paper aims to address by presenting a design for
lowering the impact of call interruptions. Based on the first
observations of Leiva et al. [16] the goal of this paper is to
increase the understanding of phone call interruptions and
to propose new UIs to reduce their effects.
Analyzing current smartphone models (iPhone, different
Android devices, Windows Phone, N9) we found that they
have two shortcomings that may amplify the disruptiveness
of incoming call notifications:
1. Call apps by default use full-screen modal dialogs to
notify the user of incoming calls. This visually detaches
the user from his previous app and thus might lead to a
higher impact of the interruption.
2. Call apps only provide the user with two options: to
promptly either accept or decline an incoming call. This
unavoidable decision (accept vs. decline) might amplify
the interruption. Further, accepting the call pulls the us-
er’s attention further away from the previous app to the
phone conversation, and declining may have additional
negative side effects (e.g. social implications).
We tackle these two issues by revisiting and extending the
design space of phone call apps as follows. First, we in-
crease the user’s freedom in deciding when to pick up a call
by introducing the possibility of postponing an incoming
phone call. Second, we re-iterate on the design of user inter-
faces of phone call apps to mitigate the interruptive effect
of incoming calls while an app is being used. In particular,
we extend the design of current phone call UIs to allow for
a higher degree of multitasking and additional options to
handle incoming calls.
Figure 2 describes an activity diagram for handling calls
from incoming (I) to ending (E). The chart highlights the
new activity and transitions that we propose in this paper
(green). In addition to accepting (A) and declining (D) an
incoming call, we introduce postponing (P) the call. The
three activities A, D, and P relate to handling of notifica-
tions of incoming calls. While calls can be accepted or de-
clined only once, the postpone activity can be repeated sev-
eral times. The incoming call (I) and a postponed call (P)
might directly end (E) if the caller hangs up or the
voicemail answers the call.
Current Phone Call User Interfaces
Figure 3a sketches the design of currently predominant
phone apps, which we refer to as baseline UI. This design
only provides options for either accepting or declining the
incoming phone call. Some specific implementations pro-
vide additional options, e.g. shortcuts to sending messages
like “I am currently in a meeting” to the caller when de-
clining the call, or augmenting the call with additional in-
formation like the caller’s profile picture or birthday.
This baseline UI results in overheads in usage times of the
concurrent apps when interrupted [16], for instance, if a
user is writing an e-mail or updating their status on a social
network. As described earlier (see Figure 1), this interaction
model has barely changed since the development of early
mobile phones. This seems like a missed opportunity as
current-generation phone operating systems allow users to
multitask between different apps. Despite this potential,
current-generation smartphones do not allow users to simul-
taneously use an app while there is an incoming call.
Postponing Incoming Calls
A first improvement to mitigate the effect of call interrup-
tions is to give the callee a greater choice in deciding how
to handle the call. Besides the option to accept or decline a
call, we provide a third option: to postpone a call. Hence,
the user can return to his previous app without a need to
decide how to react to the incoming call. The approach of
postponing calls transfers a user’s ability to pick up the call
at will from landline phones to smartphones. Users benefit
from the increased flexibility and choice to defer the call
interruption. After some time, however, voicemail might
possibly answer the call. Postponing is not as determined as
a “decline and call-back” strategy because the caller will
not recognize the callee’s postpone action.
Figure 3b shows that the postpone option can be imple-
mented as an additional button besides accept and decline
within the full-screen notification for calls. When a call is
postponed, the phone call app should go into the back-
ground so that the user can continue working in his previ-
ous app. The caller will be kept waiting on the line. After a
certain time span the call app will come to the foreground
again, and again the user has the three options to accept,
postpone or decline the call. We will refer to this proposal
for a new call interaction design as the postpone UI.
The difference between 'letting it ring' and pressing 'post-
pone' is that the notification UI disappears when pressing
'postpone', and reappears automatically after some seconds.
Figure 2. Activity diagram of handling incoming calls.
For the caller, however, there is no difference and it will
keep ringing until he hangs up, or possibly the callee picks
up, or the voicemail answers.
Multiplexing Notifications
Multiplexing the primary and secondary tasks on a
smartphone’s limited screen real estate has been found to
provide the user with more control while being interrupt-
ed [2]. Therefore, a second approach to mitigate phone call
interruptions is to alter the visual appearance of call notifi-
cations. Rather than having a full-screen notification, we
propose to divide the mobile screen’s limited space into two
areas (see Figure 3c). The basic idea is to use a smaller area
of the screen to notify the user of an incoming call, rather
than using a full-screen notification. And again, the user has
the choice to either accept, decline, or postpone the call.
With less screen area used for the alert, the user can contin-
ue working on their primary task. For example, they might
want to finish writing a sentence in an email or tag a point
on a map. We will refer to this design as the multiplex UI.
We conducted a study to test the two proposed UI designs.
We were interested in how the interruption of an incoming
phone call would impact user experience and app usage.
Study Design
We followed a within-subject A/B/C design for the con-
trolled lab study. We tested the multiplex UI compared to
the postpone UI and the baseline UI. In each condition there
was a primary task people had to solve using some apps,
and a secondary task with a phone call that interrupted the
app usage of the primary task.
Twelve participants (six female) with mean age 25.8 years
(SD = 4.1 years, range: 19 - 31) were recruited from a local
university campus. Participants had a mix of technical and
non-technical backgrounds.
We implemented the presented design options for call noti-
fications as a prototype for Android phones, and we gave an
instrumented smartphone to the participants. Figure 3
shows screenshots of the three call-handling UIs we imple-
mented according to the design space introduced before.
The postpone duration was set to 5 seconds.
For the primary task we implemented common mobile use
cases inspired by the tasks of Cauchard et al. [6], who used
typical mobile apps like maps, contact list, and calendar.
Participants got a question and had to use three other apps
for answering the question, always by memorizing and
connecting pieces of information shown in the other apps.
Within the interruptive phone call our participants had to do
a word-generation task [24]: the caller would say five
words, and the participant had to think of and respond with
new words starting with the last letter of the given word
(which was in the participant’s mother tongue). As key-
words the peers would say “hello” and “bye” to signal start
and end of the task. The call was automatically initiated on
the caller’s phone when the participant started working on a
task and the interruption reached him after 6 seconds.
A pre-study revealed that users might intentionally post-
pone the call until they have solved the task, since there is
no hurry to accept the simulated call. To mitigate this ef-
fect, we first introduced a random time (between 20-30 se-
conds) after which the caller would hang up (as if the
voicemail answered the call). In addition, to motivate peo-
ple to perform well in both tasks, we gave an additional
award (20 EUR gift certificate) to the participant who per-
formed best in the whole study (correctly solved tasks, cor-
rectly created words, and fastest in both tasks).
We explained to our participants what the study was about.
We spent about 15 minutes to acquaint them with the tasks,
as well as the different designs: Participants learned how to
use the three different call UIs and how to solve the primary
tasks. They were also introduced to the secondary task on
the phone, both standalone and as an interruption during the
primary task. For the training we used questions and words
not used in the three main parts of the study.
After the training the study had three parts: In each part the
participants had to answer 20 questions in the primary task,
11 of them being interrupted by calls. During training the
participants were instructed to be the one to hang up after
each call. Each part was assigned to one condition (baseline
UI / postpone UI / multiplex UI) in a counterbalanced way.
After each part, the participants were asked to fill out a
questionnaire. One experimenter stayed with the partici-
pant, while a second experimenter carried out the call inter-
ruptions remotely. At the end of the study we asked for ad-
ditional demographic information. The experiment took
about 60 minutes to complete and participants received a 10
EUR gift certificate for their time.
We collected logging data from the device to measure how
people interacted with the UIs and the applications of the
primary task, and we collected measures from question-
naires that we administered after each part. In particular, we
observed the following dependent variables:
a) b) c)
Figure 3: Screenshots of prototype implementing the three UI
designs: a) baseline UI; b) postpone UI; c) multiplex UI.
Time on task: We logged how long people were working
on the primary task (TOT), i.e. how long they were work-
ing with the four apps (question, agenda, map, contacts)
to answer the question. Further, we distinguish between
the time on task before the interruption (TOT1) and the
time on task after the interruption when participants con-
tinued to work on the task (TOT2).
Time of notification: We recorded the time the notifica-
tion of an incoming call was active (TN), i.e. the time
from first notification of the call until the conversation
started. We also kept track of the time the notification was
visible to the user (TNV), i.e. the time span a notification
was postponed would count for TN but not for TNV.
TNV ends when the postpone button is pressed.
Time on call (TOC): We measured how long the people
took for the word creation task on the phone, i.e. the dura-
tion of the phone call. All time measures had an accuracy
of milliseconds.
We administered a NASA-TLX [11] after each part to
assess the participants workload. In addition (as in [1]),
we asked people how annoying the phone call interrup-
tions were, and we asked how respectful the phone appli-
cation was according to the interruptions during the tasks.
All these measures were on the same 20-point scale.
The independent variables were the three conditions of
baseline UI, postpone UI and multiplex UI, whereas the
dependent variables were the measures explained previous-
ly. In total we collected 720 data points for each time-
related measure (12 participants x 20 tasks x 3 conditions),
and we averaged each user’s measures over the 20 trials per
condition. We collected 36 data points (12 participants x 3
conditions) for the TLX-related measure.
Results of Study I
We were mainly interested in whether the different UIs had
an impact on the workload and task performance time. Ef-
fects of the conditions were analyzed using one-way within-
subject ANAOVA (with Mauchly’s sphericity test satisfied)
and post-hoc analysis (with Bonferroni correction).
Impact on Qualitative Measures
The most important results can be drawn from usersfeed-
back on the three UIs. Figure 4 shows the results of the an-
swers people gave regarding the paper-based questionnaires
for the three UIs we tested.
Mental demand: The UI condition impacted the mental
demand of tasks (F(2,22)=4.22, p<.05). We can see that
the multiplex UI (M=11.33) was mentally significantly
less demanding than the baseline UI (M=14.17, p<.05).
The postpone UI was in between (M=13.17).
Effort: The UI condition also impacted our participants’
reported effort (F(2,22)=5.93, p<.01). We can see that
our participants needed significantly less effort for finish-
ing the tasks with the multiplex UI (M=11.58) compared
to the baseline UI (M=13.83, p<.05).
Frustration: We also found a significant impact of the UI
condition on the measured frustration (F(2,22)=16.48,
p<.001). With the multiplex UI (M=7.08) participants
were significantly less frustrated than with the postpone
(M=10.42, p<.05) and baseline UI (M=13.25, p<.001).
Respect: We also found a significant impact of the UI
condition on the perceived respectfulness (F(2,22)=35.00,
p<.001). When interruptions appeared, our participants
found the baseline UI (M=4.92) significantly less respect-
ful than the postpone UI (M=9.83, p<.05) and the multi-
plex UI (M=17.33, p<.001). The postpone UI was seen as
significantly less respectful than the multiplex UI (p<.01).
Annoyance: The UI condition also significantly impacted
how annoying an interruption was perceived as being
(F(2,22)=16.94, p<.001). The multiplex UI (M=9.25)
was significantly less annoying than the baseline
(M=17.50, p<.001), and significantly less annoying than
the postpone UI (M=15.08, p<.05).
These measures show that the multiplex UI allows the users
to solve their primary tasks with less mental demand and
less effort. Further, for the multiplex UI the users reported
*: p< .05
**: p< .01
***: p< .001
*: p< .05
**: p< .01
***: p< .001
less frustration and less annoyance, and perceived the inter-
ruption to be more respectful to their primary tasks.
Impact on Time Measures
On average people took 42.45s (SD=15.75s) to complete
one trial consisting of the primary and secondary task. The
time on task for non-interrupted trials (M=16.66s) was sig-
nificantly lower than for interrupted trials (M=27.15s,
t(11)=12.88, p<.001), which replicates earlier findings that
the interruption indeed introduces an overhead [16].
We analyzed the impact of the call interruptions on the par-
ticipant’s performance in terms of speed and errors. Figure
5 shows the data for this analysis. We did not find differ-
ences in the overall performance (resumption lag, speed and
errors) between the three conditions. Participants were a
little slower regarding performance time when using the
postpone UI. This can be explained by the repetitive open-
ing of the notification when postponing it.
Looking into when people allow for the interruption, we
found that the UI condition had a significant effect on the
time (TOT1) participants spent with the primary task before
the call was accepted (F(2,22)=15.80, p<.001). For the
baseline UI, people on average spent significantly less time
(M=7.63s) with the task before the interruption than with
the postpone UI (M=11.42s, p<.01) and the multiplex UI
(M=12.47s, p<.01). Consequently, the UI condition also has
an significant effect on (TOT2) the time participants spent
on the task after the call (F(2,22)=6.29, p<.01). However,
only the difference between the baseline UI (M=18.64s)
and the multiplex UI (M=14.25s, p<.05) is significant here.
The UI condition has a significant effect on call notification
time TN (F(2,22)=17.29, p<.001). The notification time of
calls in the baseline condition (M=1.70s) was significantly
shorter than for the postpone UI (M=5.00s, p<.01) and the
multiplexed UI (M=6.37s, p<.01). For the time the call noti-
fication was shown to the user we can also find a significant
effect of UI condition (F(2,22)=15.43, p<.001). For the
baseline UI, the TNV equals the TN since it cannot run in
parallel to the primary task, but it is significantly smaller
than for the postpone UI (M=23.82s, p<.01) and the multi-
plex UI (M=51.63s, p<.01). More interestingly, the notifi-
cation with multiplex UI was shown longer than for the
postpone UI (p<.01).
These findings suggest that people actively used the multi-
plex UI to display the call notification in parallel while
working on the task: they used it to defer the interruption.
With the multiplex UI people used the postpone option less
frequently per call (0.26 times mean) than for the postpone
UI (0.50 times mean). Though insignificant, this tendency
suggests that the multiplex UI decreases the value of the
postpone option, since the multiplex UI already makes it
possible to continue working on the primary task.
Application Switching Behavior
The UI condition has a significant effect on the usage of the
four apps required for answering the task’s question
(F(2,22)=6.725, p<.01). With the baseline UI people
launched required apps 6.74 times (SD=0.40), with the
postpone UI they launched the apps more often (7.09 times,
SD=0.68), and with the multiplex UI most often 7.44
times (SD=0.58). The latter is significantly higher than with
the postpone condition (p<.01).
Resuming from an interrupted task is a reconstruction pro-
cess [20], which requires re-opening the apps in our study.
While participants switched more often between apps with
the multiplex UI, we did not find any significant differences
in the usage time of the apps and the total time on the pri-
mary task (TOT); i.e., app switching frequency was highest
for the multiplexed UI. This suggests that solving tasks re-
quiring more than one app is easier with the multiplexed UI
than with the two other UI designs. This is because the mul-
tiplex UI does not interfere with the app switching itself,
while the baseline and postpone UIs disturb app transitions.
As a consequence, the multiplex UI allows for better task
reconstruction when re-opening apps is required.
Discussion Study I
Our participants told us that for the postpone UI they would
like to have an option for getting back to a call notification
in the postpone state instead of waiting until it comes back
automatically. This would also allow users to immediately
turn their attention towards the call after reaching an inter-
mediate state or finishing a subtask in the primary app.
Considering the small effect that for the postpone UI the
time on task was longest and the time on call was shortest
(see TOT and TOC in Figure 5), it seems like people used
the waiting time in the primary task (waiting for the notifi-
cation to return from being postponed) to prepare for the
call task. In contrast, with the multiplex UI one retains con-
trol over the call notification and can immediately accept
the call if the primary task (or subtask) is completed.
In the lab study, we did not find any significant effect of the
UI design on the time participants needed to finish their
tasks. Although this study provides us with insights into the
differences between the design solutions that we proposed,
this study is limited in that the interruptive call is simulated
and the tasks are artificial. Since we did not want to over-
strain our participants (which might have led to fatigue;
sessions already took about 60 minutes) their tasks were
rather short. Further, we enticed them to always accept each
incoming call. To understand how people would use the
multiplex UI in a natural context, we conducted a second
study in the wild. For the reasons explained above we de-
cided to choose and implement the multiplex UI for further
We extended the prototype of our lab study to a market-
ready app for end-users called CallHeads. We enhanced the
multiplex design of the call notification and implemented
the call notification as a circular widget that would show
the caller’s contact picture and name below: see Figure 4.
Further, by dragging this widget to the screen’s edges, the
user can accept (green, right), decline (red, left) or postpone
(yellow, bottom) the call (see Figure 4b). The colored edges
also each show an icon and appear as soon as the user
touches the widget. The postpone duration for CallHeads
was 5 seconds by default and users were able to change it.
Design of Study II
Our second study is designed as a natural experiment [22]:
neither we as the researchers nor the study participants had
control over the incoming calls and people’s tasks. The call
interruptions and the tasks the users were currently carrying
out on their phones were subject to their natural contexts.
We released CallHeads on the Google Play app store so
that people could install it to their devices and use the mul-
tiplex UI. To study the usage of CallHeads we released a
second app. Instead of implementing the study within
CallHeads (which would have required permission to ac-
cess the Internet), we decided to release a second distinct
app for running the study. This allowed users to use the
original app without taking part in our study. We believe
that this is an important step for making it very transparent
to the user that by installing the second app they will take
part in a research study (since this was the sole and clearly-
stated purpose of the second app). In addition, within the
study app we asked participants for consent to take part in
the study following the two-buttons approach [18]. We did
not collect any qualitative measures that we surveyed from
participants of Study I. Study II was fully anonymous and
we did not collect any data that would disclose anyone’s
identity or content of conversations.
Results of Study II
We analyzed the data set to characterize the behavior of
people when they were called. For investigating differences
between groups, we conducted paired t-tests on subsets of
the sample containing both conditions per user.
Characteristics of Collected Data
We released CallHeads publicly on July 4, 2013. More than
32,000 users downloaded it within 10 weeks and about
10,500 users had it actively installed at the time of writing.
652 of those users agreed to submit data for our study. We
withdrew the first two days of data for every user to remove
possible self-tests with the app. To purge unnatural user
behavior, we further only considered users providing data
for more than two days. As such, our final cleaned data set
comprises 525 users with data for 31.03 days (SD = 13.86d)
per user on average. During this time we observed 88,516
incoming calls, 160.05 per user on average (SD = 227.51)
and 3.3 calls on average per user and day. 28,906 calls
came in while the user had his device unlocked, with 54.36
per user (SD = 84.82). These 32.66% of calls constitute
interruptions of concurrent app usage that we are interested
in, and we limit our further analysis to them. This high
number of call interruptions substantiates this as a practical
problem for current smartphone usage.
Cases of Call Handling
In our uncontrolled study we can observe three kinds of call
endings: accepted calls (the user answers the call), declined
calls (the user declines the call), and unanswered calls (the
caller hangs up or is answered by voicemail). Table 1 pro-
vides an overview of the occurrences of the following cases
of interruptive calls:
Accepted calls: Out of all interruptive calls more than half
(16,119; 55.77%) were accepted, i.e. 32.64 (SD = 49.89)
per user on average.
Declined calls: Out of all interruptive calls 2,311 (7.99%)
were declined, on average 7.36 (SD = 13.56) per user. In
this case the callee actively refuses the call by dragging
the widget to the red area (Figure 4). A call can be de-
clined for various reasons [19]. One simple explanation is
that the callee does not want to start talking to the caller,
or does not want to be interrupted from his current app.
Unanswered calls: Out of all interruptive calls 10,476
(36.24%) went unanswered, i.e. 4.22 (SD = 19.89) per us-
er on average. Note that in these cases of unanswered
calls, the phone was not in standby and the user was likely
to be using his phone. When a certain time limit is
reached, the caller might hang up or the call might possi-
bly be intercepted by the callee’s voicemail. This time
limit can be reached if the callee repeatedly makes use of
a) b)
Figure 6. Screenshots of the phone call app deployed on the
Google Play Market with a) call notification and b) user inter-
acting with the notification widget to accept a call.
per user
Incoming calls total
… non-interruptive
… interruptive
Interruptive calls accepted
…after being postponed
Interruptive calls declined
…after being postponed
Interr. calls unanswered
…after being postponed
Postpone events
Widget move events
Table 1. Descriptive stats on number of calls and events.
the postpone function or keeps the notification’s widget
active on screen, maybe after moving it to a corner of the
screen, while continuing using the concurrent app.
We can see that most calls were answered and only a few
were declined. Note that for the 36% unanswered calls the
user had his device in active mode, i.e. they were not unan-
swered due to unavoidable unavailability, but due to inten-
tional or enforced unavailability [19], as the user just let it
ring. Actively using the phone while a call alert is being
shown but leaving the call unanswered is a new behavior
introduced with our design; we call this passive decline.
Timing of Call Handling
It is worth mentioning that only our new design allows the
user to continue using his app while the call notification is
pending before making a decision on how to handle the
call. We analyzed the timing of the decisions to understand
how participants made use of this new opportunity. This
relates to our lab study’s time of notification, but in the wild
we also saw people not accepting some calls since they
showed natural behavior that we did not control. Again, we
can distinguish three cases:
Time until accept: Before a call was accepted, its notifica-
tion time was 7.08 seconds on average (SD = 6.07s). The
reason for the relatively long waiting time is that in this
case the callee can also postpone the call or move the call
icon out of the focused area. 106 calls (0.65% of accept-
ed) were postponed at least once before being accepted.
Time until decline: The notification time before declining
a call was 11.06 seconds on average (SD = 4.41s). This
time is 1.65 times higher than for accepted calls. This is a
significant difference (t-test on subset of paired cases;
t(307) = 2.14, p <. 05). 114 calls (4.93% of declined) were
postponed at least once before being declined.
Time until ending unanswered: On average, the notifica-
tion time for an unanswered call was 23.35 seconds (SD =
18.66s) before the phone stopped ringing. This is signifi-
cantly higher than the notification time before accepting a
call (t-test on subset of paired cases; t(437) = 13.80, p <
.001). 539 calls (5.15% of unanswered) were postponed at
least once before being left unanswered.
We can see that for calls resulting in declines, notifications
last longer. For calls where the callee instead makes no de-
cision and waits for the caller to hang up, or the voicemail
answers, the notification time is even longer. Possibly the
notification time is longer when declining since declining a
call might be a more cumbersome decision. Further tempo-
rizing this decision results in unanswered calls, where the
user also might want to pretend unavailability.
Usage of Postpone
In total we recorded 770 postpone events; 197 were se-
quences of postpone events, i.e. cases where the callee
postponed the same call more than once. Nearly half of all
users (47.05%) postponed a call at least once; on average a
user postponed 4.46 times (SD = 6.08). And on average
2.66% of calls were postponed at least once.
Interestingly, 499 cases (64.81%) of postpone actions led to
unanswered calls. As already mentioned, this can either
result from the callee having voicemail, or the caller being
unwilling to wait any longer and hanging up. Looking at
these nearly two-thirds of postpone cases, we found that for
23.64% of calls the caller was willing to wait for more than
40s; average waiting time was 30.22s (SD = 13.31s).
We expected the postpone option to be used more often
than only in 2.66% of calls. This underpins that postpone is
not essential for mitigating interruptions when using a mul-
tiplex UI, but it can be helpful in certain contexts.
App Usage with Notification Being Multiplexed
We also investigated whether the user’s call handling is
influenced by the app which is being interrupted. Therefore
we looked for apps that we observed to frequently be inter-
rupted when a phone call comes in. We found that the like-
lihood of using the postpone option is high when media
applications are being used. The probability that a call inter-
rupting the app “MX Player” will be postponed is 0.24 (17
interr. calls), and for YouTube 0.23 (139 interr. calls). In
contrast, interruptions of apps that belong to the communi-
cations category have a lower probability that the call will
be postponed. For instance, for the contact book, the MMS
application and WhatsApp the probabilities of calls being
postponed are 0.03 (1,779 interr. calls), 0.04 (978 interr.
calls), and 0.06 (1,409 interr. calls) respectively.
Discussion of Study II
The study reveals that one-third of incoming phone calls
interrupt concurrently-used apps. This emphasizes the need
to improve UIs for handling phone calls, since these inter-
ruptions introduce a significant overhead [16].
Analyzing the use of the postpone function, we found that
users leverage it to passively decline calls even though they
are using their phones. So far, without this function the
phone could not be used, and app usage could not be con-
tinued, until the call was either accepted, declined or left
unanswered. Since the caller does not know when the call is
postponed, the caller has the feeling that the callee is not
available (e.g. away from the phone). This is only possible
through our new design.
Further Refinements of the Design
Resulting from 32,000 installations of CallHeads, we also
received valuable feedback through both the app store’s
comment function as well as by email. The most requested
feature that people would like to use is to be able to drag-
and-drop the widget to an additional area for declining and
sending prewritten text messages to the caller. This is inter-
esting, since we found that it takes longer to decline a call
than to accept a call. Providing users with an option to in-
form the caller about the reason why they were declined
might improve this decision-making. In addition, some of
those calls that ended unanswered might have been declined
with an explanation provided to the caller instead of pre-
tending unavailability [19]. Further, it might also be valua-
ble to provide a function to accept a call in speaker-phone
mode, so that after accepting the call the user can stay in the
app and start talking directly and hands-free to the caller.
This would further allow for multitasking between the app
usage and the conversation, but also has other implications
(e.g. people nearby can listen to the conversation).
Our in-the-wild study is limited by the inherent properties
of the method of running large-scale studies through the
app store. Most importantly, we cannot know about the
user’s context when calls came in. For a better understand-
ing of the reasons why a call ended unanswered even
though the user was on his phone, we plan to enhance the
quantitative study presented in this paper with qualitative
methods of experience sampling in the large. Further, we do
not know anything about the relation between caller and
callee; this might impact how the callee handles the call.
With Study I we found that the multiplex UI is best suited
for handling call interruptions, and with Study II we ana-
lyzed how people use the multiplex UI in natural contexts.
We also need to consider the caller and discuss effects we
could not reveal in our studies.
Keeping the Caller Waiting on the Line
When the receiver postpones a call the caller will be kept
waiting on the line, and as we found this might result in an
unanswered call. Since related work [4,25] found that it has
a positive effect when the caller is aware of the callee’s
status, one idea might be to signal the caller as to what is
happening. One possibility could be the design of special
call-progress tones, or we could use speech synthesis to
signal the callee’s current app usage context to the user, e.g.
“the person you are calling is currently playing Angry
Birds”. Then the caller might be able to make a more in-
formed decision about how long he wants to wait for the
callee to pick up the call. For future work we plan to signal
the callee’s context back to the caller to investigate how the
caller reacts if he knows that the other person is using his
phone, but does not want to start a conversation; issues of
privacy and social aspects also need to be considered.
Social Implications
The current design of phone call UIs also has some social
implications: current implementations on the different oper-
ating systems force the user to either accept or decline an
incoming call if he wants to continue using his device. Oth-
erwise he would have to wait until the voicemail answers or
the caller hangs up. If the user wants to keep using his cur-
rent app, he will have to decline the call if he wants to
evade the interruption. As a result, declining the call might
have an impact on the peers’ social relation. Our new de-
sign allows the user to pretend unavailability while continu-
ing to use apps on his smartphone. Future work will study
the effect of this opportunity on people’s phone communi-
cation behavior leveraging the CallHeads deployment.
Switching to Call after App Episodes
The multiplex UI design we propose allows people to con-
tinue using their apps while being notified of the phone call
in parallel. Both studies provide evidence that people make
use of this new functionality to defer the interruption for a
short time to finish micro-interactions with their apps. In
particular, this form of pre-alerting for incoming calls al-
lows people to finish their current episode of interactions
with the current app, before they allow the interruption [8].
We saw in Study I that people kept the notification open for
a certain time that they would need to reach an intermediate
state in their tasks before they decided to switch to the call.
Change Blindness and Call Blindness
When adding dynamic visual content to the display one has
to be concerned about the effect of change blindness [7]:
the popup of the notification might result in the user miss-
ing changes within the primary app. This effect is greatest
when the popup opens full-screen, as with current phone
call apps. However, on the other hand, if the notification is
made smaller or placed less prominently, it might not be
noticed by a user engaging with other applications, which
might result in the user not recognizing the call. In fact, we
got requests from users of the CallHeads app to be able to
change the appearance of the notification (e.g. increase the
size of the font and the widget itself). However, in the
CallHeads app the call was also signaled by the ringtone
and vibration (if configured) as long as the user did not re-
act to the notification (by touching the widget). However,
every one of our users receiving interruptive calls interacted
with the widget at least once, so our users did not miss it.
Nonetheless, finding the optimal size for the notification,
i.e. the ratio of multiplexing between the primary app and
the notification, is a subject for future work.
Other Modalities
Our design considerations target visual attention. In addi-
tion, incoming calls are also announced by auditory and
haptic signals. A holistic design needs to consider these
modalities to notify the user of incoming phone calls. One
possibility could be changing the ringtone to unobtrusive
sounds. Hence, the user could be notified about an incom-
ing phone call in an ambient way. Also, one could apply
different vibration patterns to create haptic notifications in
accord with the visual notification. The integration of dif-
ferent modalities therefore needs to be addressed in future
work, and should be aligned with the visual notification.
The Non-interruptive Case
While the proposed multiplex UI is dedicated to the inter-
ruptive case, where a user is engaged with another task on
the mobile device, we have to raise the question of how to
proceed for the 67% of non-interruptive calls (i.e., the
phone is in standby mode when a call comes in). CallHeads
was built in such a way that it does not show up in this case,
and instead the default phone call appears. Another option
would be to apply the multiplex UI that we proposed for the
interruptive case. The problem, however, is that there is no
primary task on the phone when the phone is not being
used, though one could consider the device’s lock-screen as
the in-use app. From our users we got strong feedback that
they would also like to use the multiplex UI and the option
to postpone a call in the non-interruptive case. This could
support users engaged in non-phone tasks when calls come
in. This would instantly mute the phone (similar to silenc-
ing) and the notification would come back after a few se-
conds. Postponing a call might be beneficial when one has
to leave the room before being able to answer a call.
This paper presents a multiplex UI for handling incoming
calls on smartphones. This design solution tackles the prob-
lem that calls can interrupt concurrent application usage.
We revisited the current design of phone call UIs, extended
the options for handling incoming phone calls and present-
ed considerations for possibilities to postpone calls and
multiplex the call notification with the concurrent app. We
studied these two proposals for the design of phone call
apps in a small-scale controlled lab study. We found that
the multiplex UI improves call handling with concurrent
app use, in particular because it is less frustrating and an-
noying. We also released an implementation of the multi-
plex UI to more than 32,000 users through a commercial
app store. Some of these users (525) contributed to a study
to understand how the app was used in the wild. Results
showed that one-third of incoming calls interrupt concur-
rent app usage, and that people use the postpone option to
continue using their apps, often leaving their call unan-
swered. This was not possible with previous phone call UIs.
We thank Jessica Cauchard and Markus chtefeld for
providing us their task designs [6].
1. Adamczyk, P.D., Bailey, B.P. If not now, when?: the
effects of interruption at different moments within task
execution. In Proc. of CHI04
2. Alt, F., Shirazi, A. S., Schmidt, A., Atterer, R. Bridging
waiting times on web pages. In Proc. of MobileHCI ’12.
3. Altmann, E.M., Trafton, J.G. Memory for goals: an acti-
vation-based model. Cog. Sci., 26, (2002), 39-83.
4. Avrahami, D., Gergle, D., Hudson, S. E., Kiesler, S.
Improving the match between callers and receivers: a
study on the effect of contextual information on cell
phone interruptions. Behav. Inf. Tech. 26, 3 (2007).
5. Bogunovich, P., Salvucci, D.D. The effects of time con-
straints on user behavior for deferrable interruptions. In
Proc. of CHI 11.
6. Cauchard, J., Löchtefeld, M., Fraser, M., Krüger, A.,
Subramanian, S. m+pSpaces: virtual workspaces in the
spatially-aware mobile environment. In MobileHCI ‘12
7. Davies, T., Beeharee, A. The case of the missed icon:
change blindness on mobile devices. In Proc. of CHI 12
8. Fischer, J. E., Greenhalgh, C., Benford, S. Investigating
episodes of mobile phone activity as indicators of op-
portune moments to deliver notifications. In Proc. of
MobileHCI ’11.
9. González, V.M., Mark, G. Constant, constant, multi-
tasking craziness: managing multiple working spheres.
In Proc. of CHI ’04.
10. Grandhi, S. A., Schuler, R., Jones, Q. G. Telling calls:
facilitating mobile phone conversation grounding and
management. In Proc. of CHI 2011.
11. Hart, S., Staveland, L. Development of NASA-TLX
(Task Load Index): results of empirical and theoretical
research. In Human Mental Workload. 1988.
12. Ho, J., Intille, S. S. Using context-aware computing to
reduce the perceived burden of interruptions from mo-
bile devices. In Proc. of CHI ’05.
13. Iqbal, S.T., Horvitz, E., Ju, Y. C., Mathews, E. Hang on
a sec!: effects of proactive mediation of phone conversa-
tions while driving. In Proc. of CHI ’11.
14. Iqbal, S.T., Bailey, B.P. Oasis: a framework for linking
notification delivery to the perceptual structure of goal-
directed tasks. ACM ToCHI, 17, (2010), 15.
15. Knittel, J., Shirazi, A. S., Henze, N., Schmidt, A. Utiliz-
ing contextual information for mobile communication.
In Proc. of CHI EA ’13.
16. Leiva, L., Böhmer, M., Gehring, S., Krüger, A. Back to
the app: the costs of mobile application interruptions. In
Proc. of MobileHCI ’12.
17. Miyata, Y., Norman, D.A., Psychological issues in sup-
port of multiple activities. In User Centered System De-
sign: New Perspectives on HCI. (1986), 265-284.
18. Pielot, M., Henze, N., Boll, S. Experiments in app stores
- how to ask users for their consent? In CHI’11 Work-
shop on Ethics, Logs and Videotape.
19. Salovaara, A., Lindqvist, A., Hasu, T., Häkkilä, J. The
phone rings but the user doesn’t answer: unavailability
in mobile communication. In Proc. MobileHCI 2011.
20. Salvucci, D.D., Bogunovich, P. Multitasking and mono-
tasking: the effects of mental workload on deferred task
interruptions. In Proc. CHI 2010.
21. Schmidt, A., Takaluoma, A., Mäntyjärvi, J. Context-
aware telephony over WAP. Pers. Ubi. Computing,
4(4):225229, 2000.
22. Shadish, W. R., Cook, T. D., Campbell, D. T. Experi-
mental and Quasi-Experimental Designs for Generalized
Causal Inference. Cengage Learning, 2ed, 2001.
23. Stamm, K., Ahamed, S. I., Madiraju, P., Zulkernain, S.
Mobile intelligent interruptions management (MIIM): a
context aware unavailability system. In Proc. SAC ’10.
24. Strayer, D. L., Johnston, W. A. 2001. Driven to distrac-
tion: dual-task studies of simulated driving and convers-
ing on a cellular telephone. Psych. Sci. 12, 462-466.
25. Teevan, J., Hehmeyer, A. Understanding how the pro-
jection of availability state impacts the reception incom-
ing communication. In Proc. of CSCW ’13.
26. ter Hofte, H. Xensible interruptions from your mobile
phone. In Proc. of MobileHCI ’07.
27. Trafton, J.G., Monk, C.M., Task interruptions. In Re-
views of Hum. Factors and Ergo., 3 (2007), 111-126.
28. Want, R. When cell phones become computers. IEEE
Pervasive Computing, 8(2):25, 2009.
... Privacy awareness involves notifying individuals about the collection and use of their data, whereas privacy control allows individuals to configure certain aspects of their privacy preferences. Studies have also shown that effective privacy notices provide individuals with privacy awareness, allowing them to make informed privacy decisions and have better control over their data [25,77,126]. However, managing IoT privacy has a greater degree of potential options than managing web or mobile application privacy. ...
... Following the identification of the five major privacy factors for IoT devices' privacy notice, these factors must be effectively delivered to the user. Several studies have found that effective privacy notice allows users to understand the privacy subject better and be more informed about their data, allowing them to make better privacy decisions [25,77,126]. During our survey, we distinguished that the majority of the literature enhancements to privacy notice visualisations follow a threestep design, (i) presentation, (ii) framing, and (iii) interaction, as shown in Figure 11 (i). ...
... Furthermore, Christin et al. [34] incorporate images with different colours in the privacy notifications to raise users' privacy awareness. The authors of [25] discussed displaying a small notification at the top of the screen to reduce the users' annoyance when they receive a call. ...
Full-text available
The balance between protecting user privacy while providing cost-effective devices that are functional and usable is a key challenge in the burgeoning Internet of Things (IoT). While in traditional desktop and mobile contexts, the primary user interface is a screen, in IoT devices, screens are rare or very small, invalidating many existing approaches to protecting user privacy. Privacy visualisations are a common approach for assisting users in understanding the privacy implications of web and mobile services. To gain a thorough understanding of IoT privacy, we examine existing web, mobile, and IoT visualisation approaches. Following that, we define five major privacy factors in the IoT context: (i) type, (ii) usage, (iii) storage, (iv) retention period, and (v) access. We then describe notification methods used in various contexts as reported in the literature. We aim to highlight key approaches that developers and researchers can use for creating effective IoT privacy notices that improve user privacy management (awareness and control). Using a toolkit, a use case scenario, and two examples from the literature, we demonstrate how privacy visualisation approaches can be supported in practice.
... Existing experiment results showed that instant messaging negatively affects the performance of primary attention and response tasks (e.g., sustained attention to response [79] and Go/No-Go tasks [35]). Incoming phone calls are often alerted with full-screen notifications, which are known to be more disruptive than regular pop-up messages, causing a significant negative impact on the performance of ongoing phone tasks [5,43]. Notification modalities (i.e., sound, vibration, and silent) greatly influence a user's decision-making process for notification handling (e.g., react, focus, read, act) [85]. ...
... This will allow users to ignore irrelevant notifications and receive necessary notifications, thereby effectively handling fear of missing out while attending classes. Likewise, a user's phone usage context can be adapted to control the salience of incoming notifications, similar to how incoming calls can be displayed as a pop-up window instead of a full-screen notification when using a navigation app, for example [5]. Offering a location-based reminder for blocking phone usage [30] showed positive effects on lowering smartphone usage in class. ...
Full-text available
Smartphone notifications inform users of events to prompt timely engagement. However, there are growing concerns about messaging behaviors in schools because frequent in-class checking and responding may hinder students' learning. This study analyzed a large-scale smartphone usage dataset collected from 81 first-year college students for 14 weeks. We quantified notification arrival and checking patterns in classrooms and analyzed how notification handling behaviors were correlated with students' smartphone usage and academic performance. Our results showed that receiving messages and frequent checking occur during class. Frequent checking behaviors were positively related to longer app usage, while willful neglect of incoming messages was positively related to overall academic performance. Our work demonstrates that problematic behaviors exist related to instant messaging in learning contexts and recommends further study to devise effective intervention mechanisms.
... It has also been mentioned together with the resumption lag in Böhmer's research. The research by Böhmer et al. [6] on the interruptions from phone calls looks at resumption lag, which is a process that occurs after being interrupted. This process takes time and opens the possibility for errors. ...
Conference Paper
Full-text available
The technology of mobile devices has changed our daily lives. Since smartphone have become a multi-functional device, many people spend unnecessary time on them, and could be interrupted by inappropriate notifications such as unimportant messages from social media. Notifications from smartphone could draw people's attention and distract them from their priorities and current tasks. This research investigated that if the users were notified by their surroundings instead of smartphone, would it create less distraction and keep their focus on the present task. The experiment was a simulation of a lamp and door notification. Notifications related to work will be embedded in the lamp such as an email from a colleague. A notification that is useful when going outside such as weather information, traffic information, and schedule reminder will be embedded in the door. The experiment was conducted by sending notifications to the participant while he or she was working on a primary task and the working performance was measured. The results show that the lamp notification had fewer interruption effects than the smartphone. For the door notification, it was simulated in order to gain opinions and insights on ambient notifications from participants. Many participants agreed that the ambient notifications are useful and being informed by them could lessen the usage of their smartphone. The results and insights from this research could be used to guide the design process of ambient notifications.
... The key to user acceptance of smartwatches is result demonstrability [28,57]-indicating tangible end-user benefits of such devices. Typical beneficial characteristics of smartwatches are increased user availability via increased device proximity [5,51], quicker and more convenient responses to prompts such as notifications [61] and reduced social tension which can be caused by frequent smartphone use [32]. ...
Full-text available
Despite large investments in smartwatch development, the market growth remains smaller than forecasted. The purpose of smartwatch use remains unclear, indicated by the lack of large-scale adoption. Thus, we aim to better understand the early adoption and everyday smartwatch use. We investigate a diverse usage data of smartwatches logged over a period of up to 14 months from 79 individuals between December 2015 and March 2017, one of the largest wearable datasets collected. First, we identify both explorative and accepted behaviours that users exhibit and further investigate how the individual usage traits and features differ between the two categories. Our analysis offers an insightful perspective on how smartwatch use evolves organically. Our results improve our shared understanding of smartwatch use and users adapting their use of smartwatch over time to match the capabilities of the technology by validating numerous findings from previous literature.
... Recently, we have seen an increase in novel attempts to address these problems by utilising contextual data [7,9,13,15,17,20] or notification content [14,21] in order to deliver notifications at more opportune moments, the use of ambient information presentation or augmentation, or by making it easier to deal with interruptions [2]. However, significant Figure 1: This alarm clock uses ambient light to softly wake up its user. ...
Human attention has become a critical resource for the effective design of smart services in which control may move back and forth between humans and computers. To avoid errors in critical conditions when the mental load is high, computer systems need to manage ongoing interruptions. In particular, the effect of interruptions can be mitigated with previews of computer-generated notifications. While previews have been used to increase engagement, research on their potential to mitigate the effect of interruptions is scarce. Using an experiment based on a game environment with varying task loads, we investigated the effect of previews on mitigating interruptions at several levels of mental load. We found interruptions that displayed previews added less to participants’ mental load but did not improve their overall performance. These results were consistent in all levels of task load. We summarize the article by discussing how previews can be designed to minimize the negative effects of interruptions.
Mobile ecological momentary assessments (mEMAs) require substantial user efforts to complete, resulting in low user compliance. One major source of incompliance is triggering mEMAs at inopportune moments. In this work, we propose a framework for implementing adaptive mEMAs using reinforcement learning (RL) to address the timing and context challenge, aiming to improve long term response compliance. To effectively model user state, we also propose a two-level user model with both momentary and routine state features. A novel k-routine mining algorithm is developed to extract routine state from passive sensing data. Using real mobile sensing data collected from 220 participants for over two weeks, we show that our proposed RL strategies consistently outperform the baseline methods including a random strategy and a supervised strategy in user compliance.
Driving safely on a university campus is critical. Cell phone use while driving is considered unsafe and can increase the risk of traffic crashes. The purpose of this study is to measure the rate of cell phone use while driving on campus in a major university in the United States. The overall rate of cell phone use was estimated at 2.5%. This rate is not considered high compared to previously measured rates in the United States. The cell phone use rate among faculty and staff drivers (3.5%) was higher in comparison with the student drivers (2.1%). The results indicate that there is a need for more interventions to reduce this dangerous behavior on campus, and several solutions were proposed.
IoT (Internet of Things) real paper uses sensors and interfaces inevitably grasp the virtual digital world. For VIoT video surveillance, ChD (Change Detection) is an essential component. Time calculating the difference between the numbers of various ChD recorded in the video monitor progress span similar scene pattern. The histogram threshold is dependent on the data set in this document, and checking performed using highlight significant for video surveillance and VIoT calculation order ChD weighed. In the 21st century, heftiness is the most common and true happiness difficult, as the World Health Organization (WHO) says is a risk of personal life. In this case, the priority is set in a family environment, and controlling overweight children's heftiness reduces good tendencies. In other words, this is a good thing; a little bit of effort has been a healthy way of life, focused on helping parents in advance. This paper warned of two brilliant gadget models to promote stability and prevent overweight adolescent's reliable behavior. The separated data may be clients, engineers, and creators who enhance management in these gadgets, useful activities, and progress. In both cases, the data can be used to negotiate a free Internet safely and protect the age thing. In this work, the only convenient AI applications that rely on the phone's sensor information also appear. It also illustrates the use of similar sensors and similar calculation of abnormal side-channel attacks.
Conference Paper
Full-text available
High-speed Internet connectivity makes browsing a convenient task. However, there are many situations in which surfing the web is still slow due to limited bandwidth, slow servers, or complex queries. As a result, loading web pages can take several seconds, making (mobile) browsing cumbersome. We present an approach which makes use of the time spent on waiting for the next page, by bridging the wait with extra cached or preloaded content. We show how the content (e.g., news, Twitter) can be adapted to the user's interests and to the context of use, hence making mobile surfing more comfortable. We compare two approaches: in time-multiplex mode, the entire screen displays bridging content until the loading is finished. In space-multiplex mode, content is displayed alongside the requested content while it loads. We use an HTTP proxy to intercept requests and add JavaScript code, which allows the bridging content from websites of our choice to be inserted. The approach was evaluated with 15 participants, assessing suitable content and usability.
Conference Paper
Full-text available
Mobile phones enable us to be reachable by phone calls anywhere and anytime. However, it is not always appropriate to answer a phone call. Even a ringing or vibrating phone can be inappropriate in some situations. The information required to assess if a call is appropriate is split between the caller and the callee. Only the caller knows the importance of the call and only the callee knows her context. Sharing parts of this context with the potential caller would enable the caller to make a better decision. Based on previous work we conducted a survey to learn about the contextual information that users believe to be important for this decision. We derive context information that users will to share and consider relevant and helpful. Further, we present a mobile application that augments users' address book with contextual information that we aim to study in the large.
Conference Paper
Full-text available
Distribution channels, such as the Android Market, provide researchers the opportunity to conduct experiments with a large number of participants. However, sometimes it may be necessary to ask for the users" consent beforehand. One question that
Many communication systems infer and project information about a user's availability, making it possible for others to decide whether and how to contact that user. Presumably when the system infers people are busy, they are less open to interruption. But analysis of 103,962 phone calls made using a popular enterprise communications tool reveals that people are actually significantly more likely to answer the phone when the system projects that they are busy than at other times. A follow-up survey of 569 users of the system suggests that this seemingly counter-intuitive fact may arise because people care a lot about the recipient's availability when initiating phone communications and are unlikely to attempt to call someone who appears to be busy unless the communication is important. Recipients thus perceive incoming calls as more important when they are busy than at other times, making them more likely to answer.
We review both the theoretical and applied research on task interruptions. We provide a brief task analysis of interruptions and resumptions, discuss how and why they are disruptive, and show how multiple theories attempt to explain the interruption and resumption process. We also review a great deal of the empirical work and show how it fits into previous theoretical accounts. Finally, we discuss what factors make interruptions more or less disruptive as well as theory-based recommendations for reducing the disruptiveness of interruptions.
Insights into human visual attention have benefited many areas of computing, but perhaps most significantly visualisation and UI design [3]. With the proliferation of mobile devices capable of supporting significantly complex applications on small screens, demands on mobile UI design and the user's visual system are becoming greater. In this paper, we report results from an empirical study of human visual attention, specifically the Change Blindness phenomenon, on handheld mobile devices and its impact on mobile UI design. It is arguable that due to the small size of the screen - unlike a typical computer monitor - a greater visual coverage of the mobile device is possible, and that these phenomena may occur less frequently during the use of the device, or even that they may not occur at all. Our study shows otherwise. We tested for Change Blindness (CB) and Inattentional Blindness (IB) in a single-modal, mobile context and attempted to establish factors in the application interface design that induce and/or reduce their occurrences. The results show that both CB and IB can and do occur while using mobile devices. The results also suggest that the number of separate attendable items on-screen is directly proportional to rates of CB. Newly inserted objects were correctly identified more often than changes applied to existing on-screen objects. These results suggest that it is important for mobile UI designers to take these aspects of visual attention into account when designing mobile applications that attempt to deliver information through visual changes or notifications.
Conference Paper
Smartphone users might be interrupted while interacting with an application, either by intended or unintended circumstances. In this paper, we report on a large-scale observational study that investigated mobile application interruptions in two scenarios: (1) intended back and forth switching between applications and (2) unintended interruptions caused by incoming phone calls. Our findings reveal that these interruptions rarely happen (at most 10% of the daily application usage), but when they do, they may introduce a significant overhead (can delay completion of a task by up to 4 times). We conclude with a discussion of the results, their limitations, and a series of implications for the design of mobile phones.
A problem with the location-free nature of cell phones is that callers have difficulty predicting receivers' states, leading to inappropriate calls. One promising solution involves helping callers decide when to interrupt by providing them contextual information about receivers. We tested the effectiveness of different kinds of contextual information by measuring the degree of agreement between receivers' desires and callers' decisions. In a simulation, five groups of participants played the role of ‘Callers’, choosing between making calls or leaving messages, and a sixth group played the role of ‘Receivers’, choosing between receiving calls or receiving messages. Callers were provided different contextual information about Receivers' locations, their cell phones' ringer state, the presence of others, or no information at all. Callers provided with contextual information made significantly more accurate decisions than those without it. Our results suggest that different contextual information generates different kinds of improvements: more appropriate interruptions or better avoidance of inappropriate interruptions. We discuss the results and implications for practice in the light of other important considerations, such as privacy and technological simplicity.