Bob Dylan said, “A man is a success if he gets up in the morning and gets to bed at night, and in between he does what he wants to do” [1]. There is no denying that computers have fundamentally changed our lives for the better, making us more creative, connected and productive. However, our devices are increasingly directing our thoughts and actions away from the tasks and goals we hold most important. How many times have you found yourself at your computer working on something important when an email notification grabs your attention? Out of these times, how often have you stopped what you’re doing to read the email, realised after a few minutes that it’s not important and then returned to your work—after staring into space for a second while you try to pick up your train of thought?

Situations such as this are often the fault of push-based digital notifications (i.e. visual, auditory or haptic alerts) letting us know about new emails, text messages, friend requests, online dating matches and breaking news. A subtle ping, ding or ring may seem harmless, but there is a growing body of evidence which suggests that the cost of an interruption is much greater than we realise. Studies have shown that interruptions caused by the receipt of a new notification incur attentional, emotional and task performance costs because they (1) take time to recover from, (2) increase the probability of errors being made, (3) disrupt ongoing tasks, and (4) can lead to feelings of stress and frustration. Of course, these effects are compounded by the sheer number of notifications we receive—a recent study on smartphone usage found that people receive on average 63.5 notifications per day [2]. I will discuss each of these claims and ultimately argue that all digital notifications should be blocked.

Interruptions Take Time to Recover From

An important concept underpinning much of the interruptions research is the Memory for Goals (MFG) theory of forgetting—a theoretical model for understanding how one reengages with a task directly after an interruption [3]. The model assumes that during goal-directed behaviour, individuals rely on mental representations of both their intentions to achieve the goal as well as temporary information related to the task [3]. When a person is interrupted, and their focus shifts to a secondary objective, the memories associated with the primary goal begin to decay rapidly. Additionally, the time it takes to resume the suspended goal is dependent on its level of activation in memory. In other words, the theory predicts an interruption duration effect: the longer the interruption, the more time it will take to resume the primary task, assuming no goal rehearsal takes place.
Researchers have confirmed this effect across multiple studies [4, 5, 6]. Hodgetts and Jones [4] used the Tower of London mental planning task to test whether unexpected interruptions incur a resumption time cost when individuals return to the interrupted task. The task required participants to rearrange a set of disks displayed on three vertical pegs to match a provided goal-state configuration. The researchers found that as interruption duration was varied from 6-s to 18-s, primary task resumption time increased from 5.01-s to 6.26-s [4]. A follow-up study conducted by Monk et al. [5] made similar observations. In a series of controlled experiments, subjects were instructed to complete a computer-based procedural task interleaved with interrupting tasks of varying duration. The findings revealed that as interruption duration increased from 3-s to 15-s, primary task resumption time rose with it, going from 1.25s to 1.75s [5].

These papers demonstrate that people are slower to retrieve a primary goal when it has been suspended for longer, confirming the time-based predictions of the MFG theory. This means that every second spent engaging with a notifying application not only causes people to spend time away from their interrupted goal but also adds to the amount of time it will take them to pick up where they left off. However, the impact of a digital notification is not merely a cost in time efficiency. As we will see in the following section, this time lag can cause people to make mistakes.

Interruptions Cause Errors

Interruptions increase the probability of a person making an error when performing a task because critical parts of the task are missed [6, 7, 8]. Trafton et al. [6] conducted a study to understand whether a model based on the MFG theory could successfully predict errors due to interruptions. Participants were interrupted at select moments while completing a computer-based procedural task. As expected, the findings showed that mean error rate was significantly higher following an interruption (9.3%) than during the non-interruption control condition (0.9%) [6]. Brumby et al. [7] found similar results in a study which explored whether the speed at which a task is resumed affects the probability of a mistake being made. The primary and secondary tasks were similar to those used in the previous study. The researchers found that as the time-cost of making an error increased from 0-s to 20-s, mean error rate rose from 10% to 24% [7]. Critically, error rates were significantly lower when no interruption occurred: 2.78% and 2.55% for the high and low-cost conditions.

Interruptions that occur after a task's primary goal have been achieved, but before the entire task has been completed can be particularly detrimental [8]. For example, if your phone were to ring just after you withdrew money from a cashpoint, but before you collected your bank card, there is a high probability that you would forget to retrieve your card. This kind of error is referred to as a postcompletion error (PCE). A study examining the effects of interruption timing on PCEs showed that error rates were significantly higher when interruptions occurred just before the postcompletion (PC) step [8]. Subjects were instructed to complete a procedural task which was delineated by a short interleaving task. Additional interrupting tasks were introduced at varying points throughout the experiment. Importantly, the PC action had to be executed shortly before moving from the first task to the second. The investigation found that mean PCE rate was highest when the task was interrupted just before the PC step (29.8%) and considerably lower when interruptions occurred at other points in the task (8.9%) [8].

Once again, the findings from each of these papers support the time-based decay predictions of the MFG theory: when attention shifts to the interrupting task, the memories associated with the suspended goal begin to deteriorate rapidly. This decay leads to the retrieval of incorrect goals when the primary task is resumed, which increases the probability of a mistake occurring. So far, we have established that interruptions take time to recover from and lead to errors. As the next section explains, many additional disruptive effects can arise from the receipt of a digital alert.

Interruptions are Disruptive

A great deal of research shows that interruptions are disruptive in most cases [9, 10, 11, 12]. In a diary study which explored how information workers interleave tasks and manage interruptions, Czerwinski et al. [9] concluded that people find it challenging to return to high-mental effort tasks following an interruption. Iqbal et al. [10] explored these observations further in a controlled experiment to understand how mental workload changes throughout a task. Twelve participants were instructed to complete a hierarchical task while their concentration levels were tracked by measuring eye pupil size. The results showed that mental workload dips at subtask boundaries [10]. Taken together, the findings from these papers suggest that the least disruptive moment for an interruption to occur is either at the start or end of a subtask. An important implication can be drawn from this work: a task management application that can recognise when a user is reaching a subtask boundary could target opportune moments for notification delivery. Though, until this functionality becomes widely available, our best bet to avoid the detriments of a mid-task interruption is to remove digital alerts from the equation altogether.

Once a notification breaks concentration in a task, it can cause individuals to enter into a 'chain of distraction' where additional tasks steal time away from an ongoing activity [11]. In one field study, information workers conducted their work while a background application tracked window switches, incoming messages and email notifications [11]. The data showed that participants spent an average of 10 minutes on interruptions caused by notifications and an additional 10 to 15 minutes on other unrelated tasks before returning to the original task [11]. Furthermore, notifications led participants to visit several applications in addition to the alerting application [11]. This observation is not surprising when we consider the infinite array of activities available to computer users. A constant influx of notifications, each presenting an alternate task to be completed can induce a state of digital restlessness.

Kushlev et al. [12] conducted a field study which looked at whether smartphone notifications are causing high levels of inattention and hyperactivity—symptoms often linked with Attention Deficit Hyperactivity Disorder. A total of 221 subjects took part in a two-week-long within-subjects study. In the first week, half of the group were instructed to maximise notifications by keeping their alert sounds on loud and their phones within reach. The other half of the group were asked to minimise notifications by disabling alerts and keeping their phones out of sight. Inattention, hyperactivity, productivity and psychological wellbeing were tracked via self-reported measures. The experiment found that when notifications were minimised subjects reported increased attention levels, improved productivity and better psychological wellbeing [12]. These results are yet another reason to suggest that considerable benefit can be gained from blocking all digital notifications. Kushlev et al. [12] demonstrated that doing so not only impacts how you feel but also how much you can achieve in a day.

Interruptions Cause Negative Emotions

This final section reports on research that has explored the relationship between interruptions and emotions. One particularly well-known study authored by Adamczyk and Bailey [13] considered whether interruption timing affects task performance and emotional state. Sixteen subjects were instructed to complete a series of everyday work tasks: for example, editing a piece of writing. A similar interrupting task was introduced at varying points throughout the experiment. Self-reported measures of annoyance, frustration, time-pressure and mental effort were recorded at the start and end of each trial. The findings showed that when notifications were delivered at random times, participants were more frustrated and annoyed, felt increased time pressure and found the task more cognitively demanding than when uninterrupted [13]. Conversely, no such effects were felt when subjects were interrupted at natural breakpoints in the task [13]. These observations are consistent with the findings of Iqbal et al. [10]: interruptions that occur at subtask boundaries are significantly less disruptive than mid-task interruptions. This research once again highlights that notification delivery mechanisms could be vastly improved if computers could recognise natural breakpoints in a task. Until this is possible, our safest option to avoid the attentional and emotional costs of random notification delivery is to disable all digital alerts.

While Adamczyk and Bailey [13] looked at interruptions in general, several studies have explored whether the use of certain applications causes negative affect [14, 15]. Mark and Cardello [14] conducted a within-subjects study to examine whether cutting off email use (and consequently blocking email notifications) affects stress. The study lasted for eight days and consisted of an initial three day observation period followed by a five-day experimental period in which email was cut off. Throughout both conditions, participant stress levels were directly measured via a wearable which tracked their heart rate variability (HRV). Out of the seven participants from which viable data was recorded, the researchers found that mean HRV was significantly reduced during the observation period (77.03) than during the no email condition (80.39) [14]. Kushlev and Dunn [15] found similar results in a two-week-long field study with 124 participants. Subjects reported lower stress and improved psychological wellbeing when email checking was restricted to three times per day compared to when email checking was unrestricted [15]. These findings provide further evidence to suggest that disabling digital alerts, particularly email notifications, can lead people to feel less frustrated, annoyed and stressed.

In conclusion, this essay has critically explored the impact of digital notifications on productivity, attentional state and emotion. The diverse range of literature I have described suggests that random delivery of digital notifications can result in a slew of detrimental effects. The MFG theory underpins much of this research, indicating that interruptions are a memory-based phenomenon: people forget elements of what they are working on following an interruption, resulting in added task resumption time and an increase in the likelihood of errors being made. These interruptions break concentration in challenging tasks, cause individuals to enter into a chain of distraction and lead to increased levels of inattention and hyperactivity. Moreover, the research suggests that digital alerts, particularly email notifications, are annoying, frustrating and can produce elevated stress levels. In light of these findings, if one’s goal is to maximise productivity, ensure accuracy, improve wellbeing, and maintain focus on the task at hand—all digital notifications should be blocked.

References

  1. Bob Dylan and Jonathan Cott. 2006. Dylan, The Essential Interviews. Wenner Books, New York, NY.
  2. Martin Pielot, Karen Church, and Rodrigo de Oliveira. 2014. An in-situ study of mobile phone notifications. In Proceedings of the 16th international conference on Human-computer interaction with mobile devices & services (MobileHCI '14). Association for Computing Machinery, New York, NY, USA, 233–242. DOI:https://doi.org/10.1145/2628363.2628364
  3. Erik M Altmann and J Gregory Trafton. 2002. Memory for goals: an activation-based model. Cognitive Science (2002), 46. DOI:https://doi.org/10.1016/S0364-0213(01)00058-1
  4. Helen M. Hodgetts and Dylan M. Jones. 2006. Interruption of the Tower of London task: Support for a goal-activation approach. Journal of Experimental Psychology: General 135, 1 (2006), 103–115. DOI:https://doi.org/10.1037/0096-3445.135.1.103
  5. Christopher A. Monk, J. Gregory Trafton, and Deborah A. Boehm-Davis. 2008. The effect of interruption duration and demand on resuming suspended goals. J. Exp. Psychol. Appl. 14, 4 (Dec. 2008), 299–313. DOI:http://dx.doi.org/10.1037/a0014402
  6. J. Gregory Trafton, Erik M. Altmann, and Raj M. Ratwani. 2011. A memory for goals model of sequence errors. Cognitive Syst. Res. 12, 2 (Jun. 2011), 134-143. DOI:http://dx.doi.org/16/j.cogsys.2010.07.010
  7. Duncan P. Brumby, Anna L. Cox, Jonathan Back, and Sandy J. J. Gould. 2013. Recovering from an interruption: Investigating speed-accuracy tradeoffs in task resumption behavior. Journal of Experimental Psychology: Applied 19, 2: 95-107. DOI:http://doi.org/10.1037/a0032696
  8. Simon Y. W. Li, Ann Blandford, Paul Cairns, and Richard M. Young. 2008. The effect of interruptions on postcompletion and other procedural errors: An account based on the activation-based goal memory model. Journal of Experimental Psychology: Applied 14, 4 (2008), 314–328. DOI:https://doi.org/10.1037/a0014397
  9. Mary Czerwinski, Eric Horvitz, and Susan Wilhite. 2004. A diary study of task switching and interruptions. In _Proceedings of the 2004 conference on Human factors in computing systems  _(CHI ’04), Association for Computing Machinery, Vienna, Austria, 175–182. DOI:https://doi.org/10.1145/985692.985715
  10. Shamsi T. Iqbal, Piotr D. Adamczyk, Xianjun Sam Zheng, and Brian P. Bailey. 2005. Towards an index of opportunity: understanding changes in mental workload during task execution. In _Proceedings of the SIGCHI conference on Human factors in computing systems _(CHI ’05), Association for Computing Machinery, Portland, Oregon, USA, 311. DOI:https://doi.org/10.1145/1054972.1055016
  11. Shamsi T. Iqbal and Eric Horvitz. 2007. Disruption and recovery of computing tasks: field study, analysis, and directions. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '07). Association for Computing Machinery, New York, NY, USA, 677–686. DOI:https://doi.org/10.1145/1240624.1240730
  12. Kostadin Kushlev, Jason Proulx, and Elizabeth W. Dunn. 2016. “Silence Your Phones”: Smartphone Notifications Increase Inattention and Hyperactivity Symptoms. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems, Association for Computing Machinery, San Jose California USA, 1011–1020. DOI:https://doi.org/10.1145/2858036.2858359
  13. Piotr D. Adamczyk and Brian P. Bailey. 2004. If not now, when? the effects of interruption at different moments within task execution. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '04). Association for Computing Machinery, New York, NY, USA, 271–278. DOI:https://doi.org/10.1145/985692.985727
  14. Gloria Mark, Stephen Voida, and Armand Cardello. 2012. "A pace not dictated by electrons": an empirical study of work without email. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '12). Association for Computing Machinery, New York, NY, USA, 555–564. DOI:https://doi.org/10.1145/2207676.2207754
  15. Kostadin Kushlev and Elizabeth W. Dunn. 2015. Checking email less frequently reduces stress. Computers in Human Behavior 43, (February 2015), 220–228. DOI:https://doi.org/10.1016/j.chb.2014.11.005