• Home   /  
  • Archive by category "1"

Multitasking Research Essay

Over the past decade, academic research has increasingly examined issues of multitasking and distraction as people try to squeeze more activities into their busy lives. Prior to the Internet age, some cognition science research focused on how behavior might be better understood, improved and made more efficient in business, hospital or other high-pressure settings. But as digital technology has become ubiquitous in many people’s daily routines — and as multitasking has become a “lifestyle” of sorts for many younger people — researchers have tried to assess how humans are coping in this highly connected environment and how “chronic multitasking” may diminish our capacity to function effectively.

In 2009, a Stanford University study published in the Proceedings of the National Academy of Sciences, “Cognitive Control in Media Multitaskers,” provided some of the most definitive evidence yet of the perils of multitasking in a digital age. It was subsequently cited hundreds of times and raised many unanswered questions and myriad research directions to pursue. But one of the study’s co-authors, Clifford Nass, notes that scholarship has remained firm in the overall assessment: “The research is almost unanimous, which is very rare in social science, and it says that people who chronically multitask show an enormous range of deficits. They’re basically terrible at all sorts of cognitive tasks, including multitasking.”

Scholars from many different disciplines are designing experimental and observational studies of all kinds to assess how we may be changing our mental habits. As the Pew Internet & American Life Project has found in conversations with experts on the subject, the very idea of “multitasking” continues to be debated and refined. The topic has also produced important book-length meditations informed by research, such as Sherry Turkle’s Alone Together, Nicholas Carr’s The Shallows and William Powers’s Hamlet’s Blackberry.

Of particular interest to researchers have been the habits of, and outcomes for, young persons — the so-called “Net Generation” or “digital natives.” (New research from students themselves suggests a higher rate of “supertaskers” — those who claim to thrive while multitasking — among younger cohorts than has been previously reported.) Research in the past few years has focused on how social networking technologies such as Facebook might affect offline performance and learning. Survey research from institutions such as the Kaiser Family Foundation and Pew Research can also complement the academic studies on the way teens and Millennials are living highly connected lives.

Below are more than a dozen representative studies in these areas:

_______

“Cognitive Control in Media Multitaskers”
Ophira, Eyal; Nass, Clifford; Wagner, Anthony D. PNAS: Proceedings of the National Academy of Sciences, August 24, 2009. doi: 10.1073/pnas.0903620106.

Findings: The study used experiments to compare heavy media multitaskers to light media multitaskers in terms of their cognitive control and ability to process information…. When intentionally distracting elements were added to experiments, heavy media multitaskers were on average 77 milliseconds slower than their light media multitasker counterparts at identifying changes in patterns. In a longer-term memory test that invited participants to recall specific elements from earlier experiments, the high media multitaskers more often falsely identified the elements that had been used most frequently as intentional distracters. The researchers conclude that the experiments “suggest that heavy media multitaskers are distracted by the multiple streams of media they are consuming, or, alternatively, that those who infrequently multitask are more effective at volitionally allocating their attention in the face of distractions.” The findings raise profound, still-unanswered questions about human cognition in the future: “If the growth of multitasking across individuals leads to or encourages the emergence of a qualitatively different, breadth-biased profile of cognitive control, then the norm of multiple input streams will have significant consequences for learning, persuasion, and other media effects. If, however, these differences in cognitive control abilities and strategies stem from stable individual differences, many individuals will be increasingly unable to cope with the changing media environment.”

 

“Multitasking across Generations: Multitasking Choices and Difficulty Ratings in Three Generations of Americans”
Carriera, L. Mark; Cheever, Nancy A.; Rosena, Larry D.; Beniteza, Sandra; Changa, Jennifer. Computers in Human Behavior, Vol. 25, Issue 2, March 2009, 483-489. http://dx.doi.org/10.1016/j.chb.2008.10.012.

Abstract: “This study investigated whether changes in the technological/social environment in the United States over time have resulted in concomitant changes in the multitasking skills of younger generations. One thousand, three hundred and nineteen Americans from three generations were queried to determine their at-home multitasking behaviors. An anonymous online questionnaire asked respondents to indicate which everyday and technology-based tasks they choose to combine for multitasking and to indicate how difficult it is to multitask when combining the tasks. Combining tasks occurred frequently, especially while listening to music or eating. Members of the ‘Net Generation’ reported more multitasking than members of ‘Generation X,’ who reported more multitasking than members of the ‘Baby Boomer’ generation. The choices of which tasks to combine for multitasking were highly correlated across generations, as were difficulty ratings of specific multitasking combinations. The results are consistent with a greater amount of general multitasking resources in younger generations, but similar mental limitations in the types of tasks that can be multitasked.”

 

“Supertaskers: Profiles in Extraordinary Multitasking Ability”
Watson, Jason M.; Strayer, David L. Psychonomic Bulletin & Review, August 2010, Vol. 17, Issue 4, 479-485.

Abstract: “Theory suggests that driving should be impaired for any motorist who is concurrently talking on a cell phone. But is everybody impaired by this dual-task combination? We tested 200 participants in a high-fidelity driving simulator in both single- and dual-task conditions. The dual task involved driving while performing a demanding auditory version of the operation span (OSPAN) task. Whereas the vast majority of participants showed significant performance decrements in dual-task conditions (compared with single-task conditions for either driving or OSPAN tasks), 2.5% of the sample showed absolutely no performance decrements with respect to performing single and dual tasks. In single-task conditions, these ‘supertaskers’ scored in the top quartile on all dependent measures associated with driving and OSPAN tasks, and Monte Carlo simulations indicated that the frequency of supertaskers was significantly greater than chance. These individual differences help to sharpen our theoretical understanding of attention and cognitive control in naturalistic settings.”

 

“Facebook and Texting Made Me Do it: Media-induced Task-switching while Studying”
Rosen, Larry D.; Carrier, L. Mark; Cheever, Nancy A. Computers in Human Behavior, 2013, Volume, 948-958. doi: http://dx.doi.org/10.1016/j.chb.2012.12.001.

Abstract: “Electronic communication is emotionally gratifying, but how do such technological distractions impact academic learning? The current study observed 263 middle school, high school and university students studying for 15 minutes in their homes. Observers noted technologies present and computer windows open in the learning environment prior to studying plus a minute-by-minute assessment of on-task behavior, off-task technology use and open computer windows during studying. A questionnaire assessed study strategies, task-switching preference, technology attitudes, media usage, monthly texting and phone calling, social networking use and grade point average (GPA). Participants averaged less than six minutes on task prior to switching most often due to technological distractions including social media, texting and preference for task-switching. Having a positive attitude toward technology did not affect being on-task during studying. However, those who preferred to task-switch had more distracting technologies available and were more likely to be off-task than others. Also, those who accessed Facebook had lower GPAs than those who avoided it. Finally, students with relatively high use of study strategies were more likely to stay on-task than other students. The educational implications include allowing students short ‘technology breaks’ to reduce distractions and teaching students metacognitive strategies regarding when interruptions negatively impact learning.”

 

“Millennials Will Benefit and Suffer Due to Their Hyperconnected Lives”
Anderson, Janna; Rainie, Lee. Pew Internet & American Life Project report, February 2013.

Abstract: “In a survey about the future of the Internet, technology experts and stakeholders were fairly evenly split as to whether the younger generation’s always-on connection to people and information will turn out to be a net positive or a net negative by 2020. They said many of the young people growing up hyperconnected to each other and the mobile Web and counting on the Internet as their external brain will be nimble, quick-acting multitaskers who will do well in key respects. At the same time, these experts predicted that the impact of networked living on today’s young will drive them to thirst for instant gratification, settle for quick choices, and lack patience. A number of the survey respondents argued that it is vital to reform education and emphasize digital literacy. A notable number expressed concerns that trends are leading to a future in which most people are shallow consumers of information, and some mentioned George Orwell’s 1984 or expressed their fears of control by powerful interests in an age of entertaining distractions.”

 

“No A 4 U: The Relationship between Multitasking and Academic Performance”
Juncoa, Reynol; Cotten, Shelia R. Computers & Education, 2012, Vol. 59, Issue 2, September 2012, 505-514. doi: http://dx.doi.org/10.1016/j.compedu.2011.12.023.

Findings: The researchers examine how the use of Facebook — and engagement in other forms of digital activity — while trying to complete schoolwork was associated with college students’ grade point averages. Students gave the researchers permission to see their grades. The participant group was 64% female, and 88% were of traditional college age, 18 to 22 years old. The study’s findings include: During coursework, “students spent the most time using Facebook, searching for non-school-related information online, and emailing. While doing schoolwork outside of class, students reported spending an average of 60 minutes per day on Facebook, 43 minutes per day searching, and 22 minutes per day on email. Lastly, students reported sending an average of 71 texts per day while doing schoolwork.” The data suggest that “using Facebook and texting while doing schoolwork were negatively predictive of overall GPA.” However, “emailing, talking on the phone, and using IM were not related to overall GPA.”

 

“Media Multitasking is Associated with Symptoms of Depression and Social Anxiety”
Becker, Mark W.;  Alzahabi, Reem; Hopwood, Christopher J. Cyberpsychology, Behavior, and Social Networking, February 2013, Vol. 16, Issue 2, 132-135. doi:10.1089/cyber.2012.0291.

Abstract: “We investigated whether multitasking with media was a unique predictor of depression and social anxiety symptoms. Participants (N=318) completed measures of their media use, personality characteristics, depression, and social anxiety. Regression analyses revealed that increased media multitasking was associated with higher depression and social anxiety symptoms, even after controlling for overall media use and the personality traits of neuroticism and extraversion. The unique association between media multitasking and these measures of psychosocial dysfunction suggests that the growing trend of multitasking with media may represent a unique risk factor for mental health problems related to mood and anxiety. Further, the results strongly suggest that future research investigating the impact of media use on mental health needs to consider the role that multitasking with media plays in the relationship.”

 

“The Impact of Engagement with Social Networking Sites (SNSs) on Cognitive Skills”
Alloway, Tracy Packiam; Alloway, Ross Geoffrey. Computers in Human Behavior, Vol. 28, Issue 5, September 2012, 1748-1754. doi: http://dx.doi.org/10.1016/j.chb.2012.04.015.

Abstract: “The aim of the present study was to investigate the effect of social networking sites (SNSs) engagement on cognitive and social skills. We investigated the use of Facebook, Twitter and YouTube in a group of young adults and tested their working memory, attentional skills, and reported levels of social connectedness. Results showed that certain activities in Facebook (such as checking friends’ status updates) and YouTube (telling a friend to watch a video) predicted working memory test performance. The findings also indicated that Active and Passive SNS users had qualitatively different profiles of attentional control. The Active SNS users were more accurate and had fewer misses of the target stimuli in the first block of trials. They also did not discriminate their attentional resources exclusively to the target stimuli and were less likely to ignore distractor stimuli. Their engagement with SNS appeared to be exploratory and they assigned similar weight to incoming streams of information. With respect to social connectedness, participants’ self-reports were significantly related to Facebook use, but not Twitter or YouTube use, possibly as the result of greater opportunity to share personal content in the former SNS.”

 

“Media Use, Face-to-Face Communication, Media Multitasking and Social Well-Being Among 8- to 12-Year-Old Girls”
Pea, Roy; Nass, Clifford; Meheula, Lyn; Rance, Marcus; Kumar, Aman; Bamford, Holden; Nass, Matthew; Simha, Aneesh; Stillerman, Benjamin; Yang, Steven; Zhou, Michael. Developmental Psychology, March 2012, Vol. 48, Issue 2, 327-336. doi: 10.1037/a0027030.

Findings: The researchers examined how digital media consumption and multitasking may impact social and cognitive development of ’tween girls. Media use included “video, video games, music listening … e-mailing/posting on social media sites, texting/instant messaging, and talking on phones/video chatting.” Researchers used data collected from nearly 3,5000 respondents to an online survey sponsored by Discovery Girls magazine in the summer of 2010. Major findings include: Watching videos, communicating online and media multitasking “were consistently associated with a range of negative socioemotional outcomes…. Face-to-face communication and online communication are not interchangeable.” Despite increased media use by ’tween girls, “no more than 10.1% of respondents ranked online friends more positively than in-person friends for even one item. Even heavy media users tended to derive … positive feelings principally from in-person friends.” Most media use had a neutral or slightly negative correlation with social well-being. In particular, watching videos was strongly associated with more negative feelings. However, “face-to-face communication was positively associated with feelings of social success [and] was consistently associated with a range of positive socioemotional outcomes.”

 

“Too Much Face and Not Enough Books: The Relationship between Multiple Indices of Facebook Use and Academic Performance”
Junco, Reynol. Computers in Human Behavior, 2011, Vol. 28, Issue 1, 187-198.

Abstract: “Because of the social media platform’s widespread adoption by college students, there is a great deal of interest in how Facebook use is related to academic performance. A small number of prior studies have examined the relationship between Facebook use and college grade point average (GPA); however, these studies have been limited by their measures, sampling designs and failure to include prior academic ability as a control variable. For instance, previous studies used non-continuous measures of time spent on Facebook and self-reported GPA. This paper fills a gap in the literature by using a large sample (N = 1,839) of college students to examine the relationship among multiple measures of frequency of Facebook use, participation in Facebook activities, and time spent preparing for class and actual overall GPA. Hierarchical (blocked) linear regression analyses revealed that time spent on Facebook was strongly and significantly negatively related to overall GPA, while only weakly related to time spent preparing for class. Furthermore, using Facebook for collecting and sharing information was positively predictive of the outcome variables while using Facebook for socializing was negatively predictive.”

 

“Facebook and Academic Performance”
Kirschner, Paul A.;Karpinski, Aryn C. Computers in Human Behavior, November 2010, Vol. 26, Issue 6, 1237-1245. doi: http://dx.doi.org/10.1016/j.chb.2010.03.024.

Abstract: “There is much talk of a change in modern youth — often referred to as digital natives or Homo Zappiens — with respect to their ability to simultaneously process multiple channels of information. In other words, kids today can multitask. Unfortunately for proponents of this position, there is much empirical documentation concerning the negative effects of attempting to simultaneously process different streams of information showing that such behavior leads to both increased study time to achieve learning parity and an increase in mistakes while processing information than those who are sequentially or serially processing that same information. This article presents the preliminary results of a descriptive and exploratory survey study involving Facebook use, often carried out simultaneously with other study activities, and its relation to academic performance as measured by self-reported Grade Point Average (GPA) and hours spent studying per week. Results show that Facebook users reported having lower GPAs and spend fewer hours per week studying than nonusers.”

 

“Perceived Academic Effects of Instant Messaging Use”
Junco, Reynol; Cotton, Sheila R. Computers & Education, Vol. 56, Issue 2, February 2011, 370-378. doi: http://dx.doi.org/10.1016/j.compedu.2010.08.020.

Abstract: “College students use information and communication technologies at much higher levels and in different ways than prior generations. They are also more likely to multitask while using information and communication technologies. However, few studies have examined the impacts of multitasking on educational outcomes among students. This study fills a gap in this area by utilizing a large-sample web-based survey of college student technology usage to examine how instant messaging and multitasking affect perceived educational outcomes. Since multitasking can impede the learning process through a form of information overload, we explore possible predictors of academic impairment due to multitasking. Results of this study suggest that college students use instant messaging at high levels, they multitask while using instant messaging, and over half report that instant messaging has had a detrimental effect on their schoolwork. Higher levels of instant messaging and specific types of multitasking activities are associated with students reporting not getting schoolwork done due to instant messaging. We discuss implications of these findings for researchers studying the social impacts of technology and those in higher education administration.”

 

“Cognitive Pitfall! Videogame Players Are Not Immune to Dual-Task Costs”
Donohue, S.E.; James, B.; Eslick, A.N.; Mitroff, S.R. Attention, Perception & Psychophysics, July 2012, Vol. 74, Issue 5, 803-809. doi: 10.3758/s13414-012-0323-y.

Findings: The researchers look at how gamers and non-gamers perform simultaneous tasks and whether serious gamers were better at multitasking than non-gamers. The researchers devised three simulations that measured driving speed and safety, multi-object tracking and image search skills. Each simulation had two versions: a single-track version involving only the simulation task; and a dual-track version in which participants were asked trivia questions while completing the simulation. The study’s findings include: “All of the participants … performed worse during the dual-task condition, and there were no differences in how they were affected.” None of the subjects, including both gamers and non-gamers, met the requirements to be classified as supertaskers. The authors suggest that “there are indeed limits to [gaming’s] benefits” and that gamers’ heightened powers of perception may be restricted to one task at a time. The researchers suggest that a gamer’s “heightened visual attention may come at the expense of the attentional resources available to other modalities” such as sound, and that these shortcomings may only emerge when faced with unfamiliar tasks. “This result demonstrates just how detrimental a concurrent distracting task can be,” the authors conclude. “Combined with other, previous evidence … this highlights how important it is for society to understand the limits of attentional processing.”

 

“Gender Differences in Multitasking Reflect Spatial Ability”
Mäntylä, Timo. Psychological Science,April 2013, Vol. 24, No. 4, 514-520. doi: 10.1177/0956797612459660.

Abstract: “Demands involving the scheduling and interleaving of multiple activities have become increasingly prevalent, especially for women in both their paid and unpaid work hours. Despite the ubiquity of everyday requirements to multitask, individual and gender-related differences in multitasking have gained minimal attention in past research. In two experiments, participants completed a multitasking session with four gender-fair monitoring tasks and separate tasks measuring executive functioning (working memory updating) and spatial ability (mental rotation). In both experiments, males outperformed females in monitoring accuracy. Individual differences in executive functioning and spatial ability were independent predictors of monitoring accuracy, but only spatial ability mediated gender differences in multitasking. Menstrual changes accentuated these effects, such that gender differences in multitasking (and spatial ability) were eliminated between males and females who were in the menstrual phase of the menstrual cycle but not between males and females who were in the luteal phase. These findings suggest that multitasking involves spatiotemporal task coordination and that gender differences in multiple-task performance reflect differences in spatial ability.”

 

Tags: technology, youth, Facebook, research roundup

Last updated: July 11, 2013

 

We welcome feedback. Please contact us here.

Education, Social MediaFacebook, research roundup, technology, youth

Abstract

Multitasking is prevalent during computer-mediated work. Users tend to switch between multiple ongoing computer-based tasks either due to a personal decision to break from the current task (self-interruption) or due to an external interruption, such as an electronic notification. To examine how different types of multitasking, along with subjective task difficulty, influence performance, we conducted a controlled experiment using a custom-developed multitasking environment. A total of 636 subjects were randomly assigned into one of the three conditions: discretionary, where they were allowed to decide when and how often to switch tasks; mandatory, where they were forced to switch tasks at specific times; and sequential, where they had to perform tasks in sequence, without switching. The experimental environment featured a primary problem-solving task and five secondary tasks. The results show that when the primary task was considered difficult, subjects forced to multitask had significantly lower performance compared with not only the subjects who did not multitask but also the subjects who were able to multitask at their discretion. Conversely, when the primary task was considered easy, subjects forced to multitask had significantly higher performance than both the subjects who did not multitask and the subjects who multitasked at their discretion.

RESEARCH HIGHLIGHTS

  • The study compares the effects of different types of multitasking and subjective task difficulty with an experiment.

  • It uses a custom-developed multitasking environment with three conditions.

  • Compares performance scores of mandatory, discretionary and no multitasking.

  • Those forced to multitask performed the worst when the task was deemed difficult.

  • However, when the task was deemed easy, those forced to multitask performed the best.

1. INTRODUCTION

Multitasking is a prevalent behavior when using personal computers or mobile platforms. Both at home and in the workplace, people are frequently switching tasks to check their email, their social networking site, or another website. Studies report that computer users have multiple applications open, and switch between them frequently (Crook and Barrowcliff, 2001; Czerwinski et al., 2004). In fact, managing one's email alone involves a lot of multitasking (Bellotti et al., 2005). Multitasking is defined as the performance of several tasks at once (Rubinstein et al., 2001). However, depending on how tasks and times are defined, several interpretations are possible (Benbunan-Fich et al., 2011). Some researchers reserve the term multitasking for simultaneously conducted activities (Meyer and Kieras, 1997), while others define it in terms of task switching (Czerwinski et al., 2004). More encompassing definitions accommodate all possible cases. For example, Salvucci and Taatgen (2011) defined a multitasking continuum based on the average time spent on one task before switching to another. At one extreme, there are tasks that involve highly frequent and sometimes imperceptible switching, such as talking while driving. At the other extreme, there are tasks that involve longer spans between switches, such as writing a paper and reading email.

Prior research has identified two different drivers of multitasking: external interruptions and internal decisions to stop ongoing tasks (Gonzalez and Mark, 2004; Mark et al., 2005; Miyata and Norman, 1986). An external interruption occurs when an event in the environment forces a user to switch tasks, while an internal interruption comes from one's self, i.e. self-initiated, when a user decides to switch tasks at his/her discretion (Miyata and Norman, 1986). Self-initiated interruptions occur just as often as external interruptions (Gonzalez and Mark, 2004).

While multitasking has been examined in the human–computer interaction (HCI) literature, there is still ample opportunity to extend research on this topic (McCrickard et al., 2003c). There are at least two areas where additional research might be fruitful. One is the study of voluntary task switching. The work of Payne et al. (2007) established that people switch away from tasks that are no longer rewarding. Related research studies by Janssen et al. (2011) and Duggan et al. (2013) have incorporated explicit payoff structures (rewards) to investigate in more depth the determinants of voluntary task interleaving. The second area that could benefit from additional research is the study of how multitasking affects performance. The existing literature in this regard is somewhat fragmented. Some studies have examined how users' performance is impacted when receiving external interruptions (Bailey and Konstan, 2006; McFarlane, 2002; Speier et al., 2003). Other studies have focused on the relation between discretionary multitasking and the resulting performance. The findings suggest that although some amount of multitasking may not be detrimental for performance (Davidson, 2011; Palladino, 2007), intensive multitasking, characterized by high frequency switching and a large number of ongoing tasks, tends to degrade performance (Adler and Benbunan-Fich, 2012; Bailey and Konstan, 2006; Hembrooke and Gay, 2003).

The present study seeks to systematically compare different types of multitasking and its effects on performance. Using a laboratory experiment, three types of multitasking resulting from mandatory interruptions (Bailey and Iqbal, 2008), discretionary self-interruptions (Payne et al., 2007) and sequential execution where all tasks are performed in succession (Salvucci and Bogunovich, 2010) are compared. These alternative multitasking scenarios are examined in conjunction with subjective task difficulty to investigate the effects on performance. An analysis of the differences in performance among these scenarios will enable HCI researchers to have a better understanding of the positive or negative impacts of multitasking.

2. RELATED WORK

2.1 Mandatory multitasking

In HCI, multitasking has been examined from the perspective of external interruptions and the role of notification systems (McCrickard et al., 2003a,b; McFarlane, 2002; McFarlane and Latorella, 2002; Oulasvirta and Saariluoma, 2004, 2006; Trafton et al., 2003). Interruptions tend to have negative effects on performance (Oulasvirta and Saariluoma, 2004, 2006). In particular, being interrupted with a secondary task can impact performance on the primary task because of the extra time and effort needed to recall the primary task when it is resumed.

Interruptions are a frequent occurrence in computer-mediated work (Iqbal and Horvitz, 2007). Bailey and Konstan (2006) found that when interrupted, users needed more time to finish their primary task, made more errors in both tasks and had more annoyance and anxiety than those who were not interrupted. A substantial body of research has examined the disruptive effects of interruptions and has documented that increased complexity in the interrupting task leads to slower resumption times (Hodgetts and Jones, 2006) and lower primary task accuracy (Gillie and Broadbent, 1989). Cades and colleagues found that interruption complexity, defined by the number of mental operators required to complete a task, reduces the opportunity for rehearsal in the primary task, leading to an increase in the disruptiveness of the interruption (Cades et al., 2007, 2010).

Of the four different types of interruptions (immediate, negotiated, mediated and scheduled) identified by McFarlane (2002), immediate is the most detrimental for performance. The other types are not as detrimental because the user has some level of control, as he/she can decide whether or not to respond immediately (negotiated), a middle agent determines whether the interruption will occur (mediated) or the interruptions occur at predetermined intervals (scheduled).

Receiving interruptions during a task and being forced to respond at that moment is disruptive and causes users to lose their thought process and control during the performance of a task (Altmann and Trafton, 2002). Bogunovich and Salvucci (2011) discuss the concept of cognitive load interruptibility and argue that forced interruptions are less disruptive when the cognitive load is low. A key determinant of cognitive load is the level of difficulty of a task, which can be assessed through an objective measure of task complexity or through a subjective perception of complexity (Maynard and Hakel, 1997). From an objective perspective, task difficulty can be determined by task designers based on an estimation of the amount of mental resources required to complete a task. In contrast, subjective task difficulty refers to the perception that some tasks seem harder due to an intuitive sense of difficulty (Cades et al., 2008).

The impact of interruptions are contingent upon the level of difficulty of the task being performed (Gillie and Broadbent, 1989). For example, Speier et al. (2003) found that interruptions helped improve performance on simple tasks but hurt performance on more complicated tasks. When users are interrupted during complex tasks, their cognitive ability is impaired and task performance suffers. During a complex task, a distraction can interrupt the user's concentration and therefore can cause negative results (Altmann and Trafton, 2002). However, during simple tasks, where users do not have to invest a substantial amount of cognitive resources on the task at hand, interruptions can actually help them focus their attention (Speier et al., 2003), thereby improving their performance.

2.2 Discretionary multitasking

Multitasking also occurs when users decide at their own volition to interrupt the current task to pursue another one. Jin and Dabbish (2009) identified seven categories of internal interruptions. These categories explain why a user would switch to another task: adjustment, break, routine, wait, inquiry, trigger and recollection. A user may need to take a breakwhen frustrated or tired, or multitask due to a trigger or recollection when recalling a related or completed new task. People also multitask due to routine, such as checking one's email out of habit, or they may multitask due to necessary adjustments of the working environment. Other causes of multitasking include a wait, which involves filling downtime during a task, or an inquiry to receive necessary information that will help complete the task.

Discretionary multitasking has also been examined in the psychology literature. Payne et al. (2007) conducted a set of experiments designed to investigate different types of multitasking. In their second experiment, participants were performing two similar computer-based tasks and were allowed to switch between these tasks at will. The results of this experiment indicated that people switched either because tasks were no longer rewarding or because they finished a sub-goal and decided to take a break from the current task by attending to another. In fact, when given a choice, people prefer to switch at low cognitive load points (Bogunovich and Salvucci, 2011) because workload decreases upon the completion of a sub-task and the disruptive effects of interruptions are minimized at natural breaking points (Bailey and Iqbal, 2008).

In terms of task difficulty, Czerwinski et al. (2004) found that complex tasks were more difficult for subjects to resume. However, given a set of tasks, the level of difficulty affects which tasks subjects decide to pursue, the order in which these are executed and the extent to which they are interleaved (Yeung, 2010). When faced with multiple tasks, people can strategically control their allocation of attention to maximize their payoffs and meet specific performance goals (Duggan et al., 2013; Janssen and Brumby, 2010; Janssen et al., 2011).

2.3 Sequential task completion

While conceptually different, both discretionary and mandatory multitasking are theoretically important, particularly when compared with sequential task performance, which is free from interruptions. The most important difference is that the user controls the pace and timing of self-interruptions in discretionary multitasking, but does not control them in the mandatory interruption scenario. The sequential scenario, where tasks are performed consecutively and without interruptions, serves as a control condition to systematically compare different types of multitasking. In sequential execution (also called serial or mono-tasking), only one task is executed at a time from beginning to end. Although multiple tasks are completed in a time frame, there is no task interference and no switching. Thus, this mode is widely used to establish a baseline condition for performance.

2.4 Performance effects and task complexity

The relation between multitasking and performance can be explained from the perspective cognitive skills/abilities or with other factors, such as personality traits or psychological states. The level of arousal is one of the factors that has been used to explain the effects of multitasking on task performance (Oswald et al., 2007). Complex tasks produce higher levels of mental workloads and lead to higher arousal than easier tasks. Therefore, the level of difficulty of a task imposes mental workload demands on the performer that interacts differently with task interruptions. At low levels of workload, performance is compromised due to inattention and lack of stimulation, while at high levels, performance also suffers due to the cognitive inability to deal with overload. Optimum performance is in the middle, where there is the right combination of workload and attention. This inverted-U relationship between workload and performance is known as the Yerkes–Dodson law (Yerkes and Dodson, 1908). According to this law, easy tasks produce low levels of arousal, and performance can improve when the user faces additional stimuli (Teigen, 1994). Therefore, receiving interruptions during an easy task may help performance. In contrast, because difficult tasks already require substantial cognitive resources for their performance, extra interruptions further increase the overload, and performance is impaired (Altmann and Trafton, 2002).

3. HYPOTHESES

Prior research has examined performance differences considering objective task difficulty (Payne et al., 2007; Speier et al., 2003), usually in the context of a single multitasking scenario. In discretionary switching, Payne et al. (2007) found that time allocation is sensitive to the level of difficulty of each task as participants seek to optimize performance. In an interruption scenario, Speier et al.'s (2003) comparative study of simple and complex tasks found that interruptions during a task helped performance with simple tasks, but hurt the execution of more complicated ones.

Complex tasks require more cognitive effort than easier tasks and task performance is impacted. However, the same task can be difficult for one person and easier for another. Maynard and Hakel (1997) indicate that task performance depends not only on objective task complexity but on subjective perceptions of task difficulty as well. As a result, we propose that depending on the subjective difficulty of the task at hand, performance will be affected differently as a result of discretionary or mandatory interruptions. Furthermore, performance will be impacted for both the primary task and the interrupting tasks.

Deciding to take a break or being forced to take a break can affect a user's performance in different ways. The negative effects of mandatory interruptions are due to the cognitive costs associated with switching between ongoing tasks at times that are beyond the control of the user (Altmann and Trafton, 2002). When people multitask at their discretion, they can decide when and how often to switch among ongoing tasks. During a complex task, receiving unplanned interruptions can impact performance more than planning an interruption at a suitable breaking point. Therefore, we propose:

H1a. During a task they consider harder, those who are forced to multitask will perform worse on all tasks than those who do not multitask.

H1b. During a task they consider harder, those who are forced to multitask will perform worse on all tasks than those who multitask at their discretion.

In contrast, during an easier task, less cognitive resources are used. Based on the Yerkes–Dodson law, easy tasks are associated with low arousal levels. Any increase in the level of arousal can improve performance. For example, an unexpected interruption will raise the levels of arousal, and performance will improve.

Performance of those who are forced to multitask will also be different from performance of those who choose to multitask at their discretion. When multitasking at one's discretion a user may choose to switch tasks after a sub-goal has been completed (Payne et al., 2007), depending upon their priorities (Janssen and Brumby, 2010; Janssen et al., 2012). Because the user can decide when to switch and it is not unexpected, these self-interruptions are known and do not increase arousal. Given that receiving unexpected interruptions can provide greater stimulation, those forced to multitask can improve their performance more than those who multitask at their discretion. Based on these arguments, we formulate the following hypotheses:

H2a. During a task they consider easier, those who are forced to multitask will perform better in all tasks than those who do not multitask.

H2b. During a task they consider easier, those who are forced to multitask will perform better in all tasks than those who multitask at their discretion.

4. MATERIAL AND METHODS

4.1 Participants

Six hundred and thirty-six subjects (334 male and 302 female) were recruited from a large urban college in the Northeast USA. About half (307) received $10 monetary compensation and the other half (329) received course credit. Subjects performed the computer-based experiment in a laboratory setting. Participants were equally distributed in the three conditions (212 subjects in each).

4.2 Design

We developed an experimental multitasking environment in Microsoft Visual C++. In this environment, we conducted a controlled experiment where participants had to perform a primary task and a set of secondary tasks. Participants were randomly assigned into one of three multitasking conditions: discretionary,mandatory and sequential.

  • Discretionary: In the discretionary condition, all tasks were presented at once, in different tabs and subjects were able to choose when to complete each task. Subjects in this condition were allowed to switch tasks at any point. The interface kept track of when subjects were switching and how often.

  • Mandatory: In the mandatory condition, the secondary tasks appeared while subjects were in the middle of completing the primary task. In this condition, subjects were interrupted at different intervals of time with pop-up windows that forced them to complete other tasks. The interrupting task had to be completed before the user was able to resume the primary task. In this instance, one of the visual exercises covers the screen and subjects have to answer as many answers as they can before time for the time for this task expires. Once the time limit was reached, subjects were brought back to the primary task screen.

  • Sequential: In the sequential condition, the secondary tasks were displayed as pop-up windows only after the primary task was completed (i.e. the total allotted time on task had elapsed).

4.3 Tasks

The experimental environment presented six game-like tasks for participants in all three conditions. The primary task was a Sudoku puzzle.1 The goal of a Sudoku puzzle is to fill in all the boxes in a 9×9 grid, so that each column, row and 3×3 box have the numbers 1–9 without any of those numbers being repeated.

There were five secondary tasks: one textual task, two visual tasks and two number series tasks. The textual task consisted of unscrambling a series of letters to find up to 20 words. The visual tasks required subjects to select the shape that best fit the pattern. Subjects were shown four shapes and had to choose the shape that did not belong. There were ten visual multiple-choice problems and there were two of these visual tasks (i.e. two sets of ten visual exercises).2 The Number Series tasks involved subjects guessing the missing number in the series of numbers presented. Subjects had two number series exercises to complete, each with ten questions.3

Sudoku was chosen as the primary task as it requires more time and concentration to complete than the secondary tasks. In addition, when subjects are performing other tasks and return to the primary task, they need to remember their thought process. While multitasking may not be as disruptive when dealing with multiple tasks on different modalities, such as one auditory task and a separate visual task, having multiple tasks in the same modality is more disruptive (Wickens, 2002). Although the chosen tasks were unique in that they used different skills (visual, textual or numeric), they all required significant cognitive resources for their successful performance.

The goal was to implement tasks that required different skills and durations in order to emulate an actual computer usage session. Generally, users work on a primary task, which requires more time and concentration. They might be interrupted by an instant message alert, which will not require as much time or thought to respond to. Or, perhaps they receive an email message that requires a little more time than the IM alert, but less than their primary task. Our experiment tried to mimic this by providing different types of tasks with different durations.

The time to complete each task was limited and determined based on prior pilot testing. The time for each task was intentionally shorter than the time subjects needed to complete the task in order to avoid subjects being idle. For the primary task (Sudoku), the maximum time limit was set to 18 min. For the secondary tasks, the time for the word task was set to 1.5 min, while the time for the two visual and two numeric series tasks was set to 48 s. Since time allotted for each task was the same for every subject, we were able to compare the performance results across all the three different conditions, ruling out the potential influence of time on task.

4.4 Procedure

Upon arrival to the lab, each subject was randomly assigned to one of the three conditions and given specific instructions according to their condition. After signing a consent form, subjects started to use the multitasking environment. They were presented with a pre-test questionnaire, which included demographic questions (i.e. age and gender) as well as questions about usage of and comfort with a computer, and prior experience with Sudoku. After this questionnaire session, participants had a practice round of Sudoku to familiarize themselves with the Sudoku as well as the interface for this task. Next, the system presented a reminder of the game instructions and the tasks were presented according to the condition (one at a time in sequential, all at once in discretionary or through interruptions in the mandatory condition). Once the time for all the tasks expired, subjects were brought to the post-test questionnaire. The results of the tasks and questionnaires were automatically written to a unique log file generated for each participant.

4.5 Measures

Sudoku performance was calculated as the number of correct answers as a percentage of the total answers required. Specifically, in the Sudoku task there were 49 empty spaces that needed to be filled out with the appropriate numbers. The score was the number of correct values entered divided by the total number of squares that had to be filled during the session (49).

Secondarytask performance was computed by averaging the performance scores of all five secondary tasks. For the word task, there were 20 acceptable words that could be generated from unscrambling the letters. The percent correct is the number of correct responses out of 20. The same method was applied to calculate the visual and number series tasks' scores.

Overall performance was calculated as the average of all six tasks (Sudoku and secondary tasks).

Subjective task difficulty: While the Sudoku puzzle chosen was from an online selection of puzzles in the easy category, not all subjects may find it easy. Therefore, we measured subjective task difficulty in the post-test questionnaire by asking the subjects to rate the level of difficulty of the primary task (from1=easy to5=hard).

5. RESULTS

Table 1 presents the basic statistics of the demographic and skills characteristics (age, computer skills and Sudoku experience) of the sample.

Table 1.

Descriptive statistics (n=636).

Pre-test variablesMean SD Min Max 
Agea22.44 4.69 18 54 
Computer skillsb3.70 0.80 
Sudoku experiencec1.55 1.43 
Pre-test variablesMean SD Min Max 
Agea22.44 4.69 18 54 
Computer skillsb3.70 0.80 
Sudoku experiencec1.55 1.43 

View Large

To ensure that randomization worked and to rule out alternative explanations, the demographic characteristics of participants were first checked for possible pre-existing differences among conditions. None of the continuous pre-test questionnaire variables showed a systematic variation. Separate ANOVAs were performed using age (MeanDiscretionary=21.98; MeanSequential=22.53; MeanMandatory=22.80; F(2,632)=1.71 ns), computer skills (MeanDiscretionary=3.71; MeanSequential=3.70; MeanMandatory=3.70; F(2,633)=0.02 ns) and Sudoku experience (MeanDiscretionary=1.54; MeanSequential=1.57; MeanMandatory=1.53; F(2,633)=0.04 ns) as dependent variables. A separate χ2 analysis was conducted for gender. The results showed that male and female participants were equally distributed across conditions (χ2=1.68; P=0.43 ns). The demographic variables (age, gender, computer skills and Sudoku experience) were similarly distributed across conditions. In the subsequent statistical analyses, the previous experience with Sudoku will be used as a control given its potential effect to explain differences in Sudoku performance.

To examine whether the experimental condition had any influence on the subjective task difficulty, we compared subjective task difficulty across conditions and found no variation. Thus, the subjects' subjective task difficulty is independent of the multitasking condition to which they were assigned. In particular, Sudoku level of difficulty (mean=3.22; 1.38 SD and median=3), was not significantly different across conditions (F(2,633)=0.54 ns).

5.1 Test of hypotheses

In order to formally test our hypotheses, we examined whether there was an interaction effect between subjective task difficulty and experimental condition. To perform the analyses, we divided the sample into three categories: those who found the primary task (Sudoku) to be difficult (i.e. rated its difficulty with 1 or 2), those who found it easier (i.e. gave a rating of 4 or 5) and those who were neutral (i.e. chose the medium level 3). Three models were run, one for each dependent variable (overall performance, Sudoku performance and secondary task performance).

The results of each model are reported in Table 2. The name of the corresponding dependent variable is listed at the top of the table. The top portion shows the means of the nine conditions (3 modes, 3 levels of difficulty). The bottom portion of the table shows the F for the entire model and the corresponding percentage of variance explained (R2). The main effects for the explanatory variables are listed below, indicating in each case the F-statistic, its significance and the eta square (η2) to indicate the strength of the association or effect size.

Table 2.

Comparison of performance by multitasking condition and task difficulty.

Multitasking condition Subjective Sudoku difficulty Overall performance
Sudoku performance
Secondary task performance
Mandatory Easy 44.44 77.65 37.80 
Mandatory Medium 38.23 69.54 31.96 
Mandatory Hard 34.59 46.15 32.28 
Sequential Easy 42.90 77.09 36.06 
Sequential Medium 43.52 74.63 37.29 
Sequential Hard 36.47 49.43 33.88 
Discretionary Easy 43.69 75.04 37.42 
Discretionary Medium 42.01 80.48 34.32 
Discretionary Hard 36.44 49.39 33.85 
Model F

One thought on “Multitasking Research Essay

Leave a comment

L'indirizzo email non verrà pubblicato. I campi obbligatori sono contrassegnati *