CHAPTER 3

Cost of Interruption

Interruptions are an inevitable part of our everyday life as it is hard to get through the entire day without being interrupted. As suggested by Zabelina et al. in [109], people are sensitive to their surroundings and they receive more information through interruptions, which might help them in their everyday tasks and even boost their creativity. This represents a positive aspect of interruptions helping users to effortlessly receive information from different sources. Numerous studies [5, 8, 21, 23] have also demonstrated that interruptions have a detrimental effect on users’ memory, emotional and affective states, and ongoing task execution. These findings indicate a negative facet of interruptions when they arrive at inappropriate situations.

In this chapter we discuss the cost associated with the arrival of interruptions at inopportune moments. The summary of the findings from studies investigating the impact of interruptions is summarized in Table 3.1.

3.1    IMPACT ON MEMORY

In 1927, Zeigarnik performed a classic psychological study [110] (as cited in [6]) with the goal of examining the mechanisms of retrospective remembering with and without interruptions. In this study the participants were given a series of practical tasks, for instance sketching a vase and putting beads on a string. Some tasks were interrupted and others were carried out without any interruption. Tasks could be performed in any order by participants. It was possible to switch to another task without completing the ongoing task, which could be taken up later. On the completion of all tasks, they were asked to do a recall test. The results of this study demonstrate that people can recall the content of interrupted tasks with greater accuracy compared to the case of uninterrupted tasks. This indicates that people have selective memory associated with the interruptions they receive. Such observed behavior is referred to as the Zeigarnik Effect.

Although the Zeigarnik Effect suggests that interruptions are useful for retrospective memory, many other applied studies have argued that interruptions have an adverse impact on memory [5, 21, 27, 36]. In particular, Dix et al. argued that humans can memorize only a limited list of tasks they have been engaged in due to the nature of their cognitive abilities [27]. Moreover, they suggested that if interrupted during a task, humans are likely to lose track of what they were doing. Following these suggestions, Edwards and Gronlund [5] conducted an experiment to investigate the memory representation for the primary task after handling an interruption. Their study was orthogonal to the Zeigarnik Effect experiment as in the latter participants were not asked to resume or recall where they left the primary task on arrival of the interruption. In particular, participants were given a task comprising of ten items and an interruption was delivered after they completed the fifth item. The interruption was in the form of a message indicating that the first phase of the experiment was over. This message was displayed on the screen of participants’ primary task and the same message was orally reiterated by the experimenter to ensure the interruption was received by them. Through their experiment, Edwards and Gronlund showed that people tend to need a certain amount of time before resuming back to the primary task after an interruption. Moreover, they demonstrated that people possess a stronger memory representation of an uninterrupted task as compared to an interrupted task.

Table 3.1: Studies in the area of understanding the cost of interruptions

3.2    IMPACT ON ONGOING TASK PERFORMANCE

In 1981, Kreifeldt and McCarthy [55] argued that interruptions could have an adverse effect on the completion time and errors made while performing a computing task. They demonstrated that people perceive more disruption and become more prone to make errors on getting interrupted while performing a complex task compared to a simple task. Later in 1989, through a series of experiments, Gillie and Broadbent [36] investigated the impact on the primary task by three aspects of an interruption: (i) length; (ii) similarity with the primary task; and (iii) the action required to handle it. Their results show that people feel distracted when interruptions share characteristics with the ongoing task or if they are limited to complex tasks but the length of an interruption does not make it disruptive. However, these findings are not aligned with those of Czerwinski et al. [23] who demonstrated that people perceive less disruption if the interruption is highly relevant to the current task.

In [8], Bailey at al. studied whether the performance of an ongoing task is influenced by interruptions. Their experiment utilized six types of web-based tasks: addition of numbers; counting of items; comprehension of images; comprehension of written text; registration; and selection. Participants were interrupted when they were approximately in the midway to completion of each task. They were presented with a news report or an investment decision as an interruption. Their findings show that: (i) people perform interrupted tasks slower compared to non-interrupted tasks and (ii) the amount of disruption perceived depends on the type of ongoing task. Later, in another study [9], Bailey’s et al. demonstrated that the amount of disruption perceived also depends on the mental load of a user on the arrival of an interruption.

Czerwinski et al. [22] studied the impact of interruptions while performing different types of sub-tasks. Their results show that people perceive varying levels of disruption while performing different sub-tasks. They proposed that deferring interruptions until a new subtask is detected could also reduce the perceived disruption. These findings were extended by Cutrell et al. [21] to investigate the effects of instant messaging on different types of computing tasks. They found that the perceived disruptiveness is higher when users are engaged with tasks that require their attention.

3.3    IMPACT ON USERS’ EMOTIONAL STATE

In [111], Zijlstra et al. for the first time studied the effect of interruptions on users’ psychological state. They investigated whether interruptions produce an adverse effect on users’ emotions and well-being, and raise their activeness level. They conducted a series of experiments by creating a simulated office environment for performing realistic text editing tasks. Their findings suggest that users’ emotion and well-being are negatively impacted by interruptions, but they do not affect the activeness level.

In 2001, Bailey at al. investigated the effects of interruption on users’ annoyance and anxiety levels [9]. Through a series of experiments, they demonstrated that people experience annoyance on arrival of an interruption. The annoyance level experienced by users depends on the type of ongoing task, but not on the type of the interruption task. They also show that the increase in users’ anxiety level is higher when they receive interruptions during a primary task as compared to the arrival of interruption on completion of the primary task.

In another study [2], Adamczyk and Bailey investigated the impact on users’ emotional state by interruptions arriving at particular times during task execution. In their experiment, participants were asked to perform tasks (such as text editing, searching, and watching video) and a periodic news alert was triggered as an interruption. Their findings show that users experience annoyance and frustration on receiving interruptions. Moreover, interruptions arriving at different moments have a varying impact on users’ emotional state.

In a study concerning mobile notifications [56], Kushlev et al. investigated whether interruptions coming from mobile phones cause lack of attention and symptoms linked to Attention Deficit Hyperactivity Disorder (ADHD). They asked participants to maximize interruptions by turning on their phones’ notification alerts and trying to mostly be within the reach of their phone. Later, participants were asked to minimize interruptions by turning off their phones’ notification alerts and trying to stay away from their phones. Their results show that people reported higher levels of hyperactivity and distraction during the first phase of the experiments. This suggests that by simply adjusting existing phone settings people can reduce inattention and hyperactivity levels.

3.4    IMPACT ON USER EXPERIENCE

“The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.”—Mark Weiser [106].

The above quote from Mark Weiser’s seminal paper [106] captures and summarizes his vision for ubiquitous computing. His goal was to design an environment with embedded unobtrusive computing and communication capabilities that can blend with users’ day-to-day life. The two key aspects of his vision were: (i) effective use of the environment in order to fuse technology with it and (ii) making the technology disappear in the environment.

The second aspect of his vision focuses on the user experience and suggests making technology disappear from the user’s consciousness. Another classic paper of Weiser [107] describes the disappearing technologies as calm computing. In this paper, the authors suggested that the technology should enable the seamless provision of information to users without demanding their focus and attention. However, interruptions have the potential to take mobile technology and Mark Weiser’s vision far apart because they not only create potential information overload but also demand user attention.

As suggested by Mark Weiser [106], technology should be “transparent” to users so that they should not notice that they are interacting with computing devices. In one of the first works in this area, Kreifeldt and McCarthy investigated the design of different user interfaces in order to reduce the effects of interruptions [55]. Their results suggest that interaction design plays a key role in affecting users’ ability to successfully resume the interrupted task. More specifically, their findings suggest that complex interfaces make it difficult to handle interruptions.

3.5    INDIVIDUAL DIFFERENCES IN PERCEIVING DISRUPTION FROM INTERRUPTIONS

As defined earlier, interruptions are unanticipated events that fragment the flow of execution of an ongoing task by demanding users to switch their attention to interrupting tasks. This leads to a multitasking environment for users. Humans naturally have skills to handle interruptions and adapt to a multitasking environment; however, they do show individual differences in their ability to perform tasks in such setting [4, 12, 54, 105].

In 1953, Atkinson investigated the role of people’s motivation to recall the completed and interrupted tasks [4]. The study demonstrates that highly motivated users are likely to recall interrupted tasks better as compared to recalling uninterrupted tasks. On the other hand, less motivated people show a tendency to recall completed tasks better than recalling the interrupted task. Overall, his findings suggest that the ability to recall the primary task after handling interruptions varies across people. A few years later, similar findings were reported by Bernard

Weiner [105].

Joslyn and Hunt proposed “The Puzzle Game”—an empirically validated test that can quantify the performance of users for multitasking [54]. Through a series of experiments, the authors showed that the ability to make rapid decisions for task switching is not the same for everyone. They suggested that people who are good in making quick decisions can be identified through testing their psychological characteristics, which can be captured by their puzzle game.

Brause and Wickens in [12] investigated the individual differences in sharing time between tasks in a multitasking environment. Their analysis show that there are differences in time-sharing ability of individuals; these are linked to their potential to process information. Moreover, their findings suggest that people have different strategies for time-sharing, which introduces differences in individuals’ multitasking ability.

Moreover, studies also have demonstrated that people show significant differences in their cognitive style with respect to multitasking [13, 47]. In particular, these tudies have demonstrated that the performance in carrying out an interrupted task is affected by users’ anxiety [47] and arousal [13] levels. However, since the levels of anxiety and arousal vary across people, different levels of disruption are perceived by them through the arrival of interruptions at inopportune moments [13, 47].

3.6    SUMMARY

In this chapter we have given an overview of the a large body of work that shows that, even though notifications are extremely beneficial to the users, they still are a cause of potential disruption as they often require users’ attention at inopportune moments. In fact, previous studies have found that interruptions at inopportune moments can adversely affect task-related memory, lead to high task error rate, impact the emotional and affective state of the user, and hinder their experience of interacting with computing devices. In the next chapters, we will present the solutions proposed by numerous studies to address the issue of delivering the right information at the right time.