Looking at Media Multitasking in Today’s Youth
A recent publication by the Kaiser Family Foundation (Foehr, 2006) sought to determine prevalence, predictors and pairings of media multitasking in today’s youth. Media multitasking as described throughout the article and this paper can be defined as using different media simultaneously. Foehr explains that it is imperative to examine media multitasking due to the increase in the trend, which an earlier publication (Roberts, D.F., Foehr, U.G., & Rideout, J.D., 2005) revealed to have risen from 16% in 1999 to 26% of media time that is spent multitasking.
In this publication (2006), Foehr discusses results of a study analyzing diary recordings of 694 3rd-12th grade students, as well as survey responses from those in grades 7-12. The sample was originally obtained using a stratified national probability sample; however due to the involved nature of the diary task, participants for this portion were self-selected. Despite the non-representative nature of the sample, the data reported is useful in obtaining information regarding student self-reported media use for a large sample size.
Of the sample of 3rd-12th grade students, 19% reported using media without engaging in simultaneous other media use, thus indicating the need for evaluating the nature of the majority of youth’s media activity. Television viewing, which was the most commonly rated activity at 17%, and includes TV, DVDs and videos, and video gaming were both more likely to be tasks performed alone (55% of the time). Furthermore, when TV viewing was paired with another activity, it was more likely to be paired with non-media related activities (28%), such as eating, than it was to be paired with media activities (13%). Contrastingly, computer activities were noted as being most likely to be paired with another activity, and that second activity is more likely to be media based. Computer activities were paired with secondary computer activities (e.g. Instant Messaging paired with websites or email) between 21% and 38% of primary computer time, and listening to music occurred between 7% and 19% of primary computer time. When the primary computer activity was doing homework, 65% of that time youth are involved in another task, 50% of which is media activity such as Instant Messaging (IMing), listening to music or watching TV.
The results above indicate that watching television remains one of the primary activities that occupy much time of American youth, when they are not multitasking; yet, when they are multitasking, TV viewing is also engaged in as a secondary activity. In addition, results indicate that computers are very conducive to multitasking with other computer activities, as well as TV viewing, and make up a large portion of multitasking activity. Such results may be of concern for a number of reasons. First of all, the ubiquitous availability of media today raises a question regarding the efficacy of media multitasking. Furthermore, since cognitive psychology research has a history of research indicating that people are not very good at multitasking due primarily to capacity limitations (Lien, Ruthruff & Johnston, 2006); there is a need to determine whether media multitasking is perhaps challenging this belief and a new conceptualization of information processing in the brain is necessary. This paper will address this concern by studying the effects of multitasking in the following medium: television viewing, laptop use, and Instant Messaging and will consider whether media multitasking can be learned for greater efficiency. Given that many of the people multitasking are younger (Carrier, Cheever, Rosen, Benitez, & Chang, 2009) and likely to be in academic settings, the final question for us to ask is whether media multitasking will prove either advantageous or detrimental in learning tasks.
Multitasking as a Skill Set
First, to address multitasking in cognitive psychology one should consider a recent article wherein the authors (Lee & Taatgen, 2004) set forth a model of information processing that discusses multitasking as a skill that can be acquired. Indeed, the efficiency promoted in the model more closely reflects human performance on the Kanfer-Ackerman Air Traffic Control (KA-ATC) task than previous models. The model proposes that production composition allows one to combine production rules into a single rule so that the two-step process of starting a rule for retrieval, followed by another rule for acting on the accessed information, is condensed into a single task-specific rule (i.e. proceduralization). Multitasking occurs when sequencing these condensed rules to take advantage of the time that is saved by proceduralization. While this model fits well into previous models of cognitive architecture (e.g. ACT-R or Soar), and more closely mimics human performance on the KA-ATC task, it should be noted that later improvements in task performance remain unaccounted for.
This ability to acquire multitasking as a skill, as facilitated by the proceduralization of certain tasks, is the central issue of this paper. For, while some tasks are conducive to multitasking, there are few lab conditions that note an ability to bypass the central bottleneck that is a major source of dual-task interference (Lien, Ruthruff, & Johnston, 2006). Thus we see some interesting findings in the analysis of an online survey of the multitasking behavior (Carrier et al, 2009). The survey compared responses of 1319 participants that across three generations: Baby Boomers- born between 1946 and 1964, Generation X- born between 1965 and 1979, and the Net Generation- born between 1980 and the present. The questionnaire asked about 12 tasks typically done at home: surfing the Web, offline computing, emailing, IMing/online chatting (synchronous communication), using the telephone, text messaging, playing video games, listening to music, watching television, eating, reading books and magazines for pleasure, and talking face-to-face with someone. Participants responded as to how many hours per day they engaged in each of the 12 tasks, whether they combined the task with any of the others, if combined-task performance was “difficult” or “easy,” and which tasks they might do together during “typical free time.”
Analysis of variance in responses indicated that the later generations reported multitasking more often than earlier generations, each to a significant degree (p < .001). There were two tasks that did not show this pattern, namely TV viewing and reading. Both of these exhibited a pattern of similar time spent by the Baby Boomers and Net generation, with increased reading in Generation X and a corresponding decrease in TV viewing time. Despite these differences, when looking at the types of tasks that people paired together, Carrier et al. found that there was remarkable similarity in these across generations. Furthermore, although the Baby Boomers rated many more task combinations as “difficult” when compared to Net Generation members (p < .001), the tasks that were rated difficult to combine by multitasking, were similarly rated by all three generations. These findings are particularly notable since they exhibit a pattern of difficulty combining particular tasks across generations, indicating a distinct cognitive limitation, outside of differences in exposure, preference and practice at multitasking. The findings also display the prominence of multitasking among youth, potentially supporting the role of multitasking as a skill to be acquired, within the limitations of task mentioned earlier.
Multitasking and Media
Given this unresolved conflict regarding the potential for learning to multitask efficiently, let us look at more real-world applications of this task set, as it applies to particular media: television, laptops and instant messaging.
More studies have been conducted on television and interference than many other media since it has been around for so long, and has also taken such a central role in the lives of many (Foehr, 2006). Furthermore, since TV is a common secondary activity to homework when it is done on the computer, it is useful to look at what research says about multitasking with television viewing as one of the tasks. In a summative article, Armstrong and Chung (2000) address the deleterious effects of dual-task television viewing on reading memory and whether there are differentiations regarding where in the process impairments occur; during comprehension, encoding or retrieval. They discuss numerous studies where TV as background noise resulted in performance decrements on complex cognitive tasks such as reading comprehension, memory, and visuo-spatial analysis and phonological short-term memory (Armstrong, Boiarsky, & Mares, 1991; Armstrong, 1993; Armstrong & Sopory, 1997 as cited in Armstrong & Chung, 2000).
Of particular interest to Armstrong and Chung was the means of interference that impaired performance on these dual-tasks. They hypothesized (H1) that deleterious effects to reading memory would be visible only when background television was present during the initial reading and that if it was present during reading, it wouldn’t cause any further memory impairment if it was also present during recall. This hypothesis is based in the assumption that ease of retrieval is facilitated by elaborative rehearsal during encoding. If encoding is impaired via capacity interference when attention is invested in the background TV stimuli, then one would expect to see those decrements from faulty encoding, and if background TV is present during recall, it should not impair performance further. Also, since elaborative rehearsal is a necessary component for organizing data to encode into long-term memory, decrements in memory performance were hypothesized (H2) to be stronger in recall memory than in recognition memory, since the latter is less reliant on cognitive organization and more a judgment of familiarity. Finally, since previous research has noted memory improvements when learning contexts have been reinstated (underwater versus dry land; Godden & Baddley, 1975, as cited in Armstrong & Chung, 2000), the authors posited (H3) that people exposed to the same television program at the time of reading and at the time of recall would exhibit better memory than if they were exposed to difference programs during reading and recall.
To test these hypotheses, Armstrong and Chung obtained a sample of 90 participants, 57% of which were female, with the median age being 20 years old. They randomly assigned participants to read one of two science articles in one of four conditions: TV-during reading only, in which they were to read while there was a background drama program on TV and then tested in silence; TV-during remembering only, in which participants read in silence and were tested with background television; TV-during-both-same program, in which participants had the same background TV program during reading and testing; and the TV-during-both-different program, where participants were exposed to two different television programs at reading and testing. Reading time was set at 12 minutes, and tests, which assessed both recall (short answer) and recognition (multiple choice), were presented after a 5 minute delay to ensure that they were accessing memory stores.
A series of ANCOVA analyses were run to evaluate the hypotheses and found that H1 was supported and significant decrements found if TV was present during reading (p < .01) which were not apparent in the TV during remembering only condition; nor did significantly different decrements result from background TV being present during both reading and testing. Results also supported H2 in that recall was significantly impeded (p < .01) by TV- during reading conditions, while recognition showed no similar effect. Finally, no significant context facilitation effects were found (H3). These results are integral to the discussion of multitasking skills such as reading and television viewing, since the capacity limits that manifest due to background television conditions significantly hamper the encoding process and result in poorer recall of information. And, although it is unlikely that the intention of students who multitask TV viewing and homework to ignore the background television program, as participants were instructed in this experiment, the TV programs in this study were chosen based on their salience to the target population and thus still provide a good measure of effects on students who are attempting a reading task with a preferred TV program running in the background.
Similar results were found when looking at merely presenting more complex materials via television news programs. While some theoretical models discuss the concept of perceptual grouping which could extend parallel processing abilities to functionally enable limited multitasking, Bergen, Grimes, and Potter (2005) discovered that semantically discordant presentation of information does not allow for this perceptual grouping. They thought that perhaps news programs, which have taken to presenting both the traditional news anchor and also including a running ticker along the bottom of the screen, were accessing a multitasking ability in people to attend to these different stimuli simultaneously. They sought to determine whether viewers’ comprehension of story facts, or demands on either their visual or auditory attentional processes were better when presented with a news anchor that took up the entire screen, or when the screen had the news anchor and ancillary visual information (such as weather or sports scores in a ticker running along the bottom of the screen).
Bergen, Grimes, and Potter (2005) conducted four experiments, the last of which will be described here, given its particular relevance to the considerations of this paper. This fourth experiment was administered to 174 participants (college students participating for credit), 88 of which were randomly assigned to the visually complex condition (with ancillary visual information,) and 86 to the visually simple condition. Participants were asked to watch four news stories that had a total running time of 11:40, and were counterbalanced across participants. Participants were given no other instruction regarding what to attend to, nor were there any incentives mentioned for attending to news information, both of which contribute to the strength of this experiment’s generalizability to real-world scenarios. After watching the stories, participants took a multiple choice recognition test (10 questions per story) based on the content of overall themes in viewed stories. A two-tailed t-test (α=.05) found significantly better recognition of story facts (p < .01) for participants in the visually simple condition than for those in the visually complex condition. Thus we see that even when presented in the visual modality, under conditions that are much closer to real-world experiences, increased complexity placed greater attentional demands, thus limiting the ability to encode memory and multitask effectively. Indeed, none of the four experiments that Bergen, Grimes, and Potter performed found any improvements based on increased presentation complexity, and all exhibited significant decrements in memory in the complex conditions. It is therefore unlikely that such newscasts are tapping an increased multitasking ability or perceptual grouping; rather, the authors suggest that it is likely that television programs targeting an audience that likes to multitask is appealing to that audience through this complex format despite its relative inefficiency at conveying meaning.
Laptops in classrooms
Another medium through which youth have indicated a preference and prevalence for multitasking is via computers (Foehr, 2006). Given the ubiquitous nature of media availability due to the affordability and convenience of laptops today, it is necessary to address how they influence performance in the classroom. Moreover, as laptops have moved into classrooms, they have posed many questions for instructors. Most importantly, as we will address here, is whether the multitasking that is connected to laptop use, effects learning, and if so what direction those effects appear.
Hembrooke and Gay (2003) performed an experiment designed specifically to differentiate learning effects of laptop use in the classroom. Forty-four students (50% male) in an upper level Communications course were issued laptops that they were encouraged to use throughout the course. Over time students had begun using the laptops for unrelated activities such as web browsing, IMing, and chatting, in addition to supplementing instruction (i.e. media multitasking), which was anecdotally considered a reason for concern. For experimental conditions students were randomly selected into two groups, one that sat through a lecture as usual (i.e. able to access their laptops in regular fashion), and one group that was told to close their laptops at the start of lecture; each condition was presented lecture separately. Lecture presentation did not differ structurally in any way from normal. After lecture the students had a surprise quiz on the lecture content that consisted of ten recognition questions (multiple choice) and ten recall questions (short answer). The experiment was repeated two months later, with participants who had been in the closed laptop condition switching to the open laptop condition, and vice versa for the other participants.
To interpret results, scores for student performance were coded as total correct, proportion of recall correct, and proportion of recognition correct. After conducting ANOVA tests, Hembrooke and Gay found significantly lower scores on total and recall questions for students who had their laptops open (experiment1: p < .04; & p < .03 respectively; experiment2: p < .004, p< .02 respectively), when compared to those who had them closed. Although media multitasking behavior itself was not measured in this study, the fact that we see markedly poorer overall performance when utilizing a laptop during class indicates some interference effect of laptop use in the classroom. In addition, the differences between performance of open and closed laptop conditions on recognition questions only approached significance (experiment1: p< .07, experiment2: p< .11), thereby supporting similar results observed in TV multitasking behavior (Armstrong & Chung, 2000), where recognition was not significantly impaired by interference from multitasking, indicating a difference in processing.
Indeed, this differential processing seems to be a more common theme across media multitasking than an acquisition of skill processing that effectively bypasses the capacity limitations of the central processing bottleneck. Some studies have directly addressed this processing difference that may result from increased media multitasking.
One such study by Levine, Waite and Bowman (2007) wanted to determine three factors relating to IM use in college students: 1) the amount of time devoted to IMing relative to other media use, 2) the nature of this IMing, and 3) if IMing was related to distractibility for academic tasks. To explore these, researchers administered a 55 item questionnaire to 115 female and 46 male college students, aged 17 to 20 years old (mean age = 18.37). Students reported on the amount of time they spent using particular media (Internet, IMing, TV, video/computer games, listening to music, reading books for pleasure, reading magazines, or reading newspapers). Participants also reported about their IM use in general and their most recent IM use. Distractibility was measured using five-point Likert scales that were compared against standard measures of distractibility and impulsiveness and found to have acceptable construct validity.
Correlations were found between students who were quick to respond to IMs and reports of being more distractible during their most recent IM sessions (p = .009). Additionally, at the zero-order, distractibility on academic tasks was negatively correlated (p < .05) with reading books for pleasure and reading magazines; while it was positively correlated (p <.05) with frequency of IMing and listening to music. The authors discussed three ways that IMing might have been interfering with successful academic reading. The first proposal was that the time spent on IMing supplanted time spent on academic tasks such as reading. This seems likely since data showed that the second highest use of the Internet, after research, was for IMing, which 90% of the sample reported using for an average of 75 minutes. Direct interference was proposed as another explanation, since the majority of respondents were multitasking at the time and 63% reported responding immediately upon receiving an IM, and interacting with three or four people simultaneously via IM. The final proposition is most salient, as it suggests that IMing can facilitate in producing a cognitive style that is characterized by short and shifting attention. All three of these possibilities can be used to explain the data that correlate increased distractibility with increased IMing and listening to music.
Cognitive Style of Media Multitaskers
To address Levine’s final proposition of a cognitive style related to media multitasking, Ophir, Nass and Wagner (2009) explored the possibility of chronic media multitaskers having different cognitive control abilities than light media multitaskers. The notion stems from an underlying question of whether the breadth-bias exhibited by multitaskers for media consumption also reflected in their cognitive control, which would be expected to manifest in increased attentiveness to irrelevant environmental stimuli and irrelevant representations in memory. First researchers developed a Media Multitasking Index and surveyed 262 university students in order to determine Heavy Media Multitaskers (HMMs; one standard deviation above the mean) and Light Media Multitaskers (LMMs; one standard deviation below the mean). Then, a series of experiments was done on different samples of these HMMs and LMMs to explore potential differences in processing abilities of these two groups, three of which will be described here.
One experiment tapped cognitive ability by studying the ability to filter out irrelevant information or the lack in filtering irrelevant information from working memory, the latter of which would be in line with a breadth orientation. Participants (N=32) in this experiment were shown a series of arrays that consisted of red and blue rectangles at different orientations, and they were given instructions to only pay attention to the red rectangles and to ignore the blue ones. Trials consisted of participants being shown an array of rectangles for 100 ms, then being presented with a second array where they had to determine if one of the red rectangles had changed orientation (clockwise or counterclockwise rotation of 45°) by pressing a “yes” button or a “no” button. The number of targets (red rectangles) and distractors (blue rectangles) were varied across trials. It was presumed that people who can filter out distracters well would not show any impeded performance on trials with more distracters; whereas people who do not filter effectively would show decrements in performance on trials with more distractors. Indeed, results found a significant negative correlation between HMMs performance and the number of distractors (p < .01), as opposed to LMMs whose performance was not impeded by an increase in the number of distractors (p > .68).
A second experiment looked at working memory and the ability to maintain and update representations in HMMs versus LMMs. Participants (N=30) were first administered a series of two-back tasks, where they had to determine if the letter presented on the screen was the same as the letter presented two slides before by pressing a “yes” key or a “no” key on a keyboard. Following the two-back task series, participants completed a series of three-back tasks where they made the same judgment, but compared the letter to one presented three slides before. Results for this experiment showed a similar decrease in accuracy (p < .05) for both HMMs and LMMs when comparing two-back tasks to three-back tasks; however, HMMs showed significantly (p < .03) more false alarms (i.e. misidentification of targets that were not correct), and these misidentifications were generally letters that had been presented, but were outside the range (two- or three-back). This pattern of misidentification increased in the three-back task (p < .03), especially for letters that had appeared more times throughout trials. These results indicate that HMMs exhibited interference from familiar items, and this interference increased as cognitive load did, while LMMs did not show this pattern of interference.
Finally, Ophir, Nass and Wagner looked at task switching ability in HMMs versus LMMs, proposing that HMMs should exhibit better task switching abilities due to their identification as heavy multitaskers. Cost of task switching was measured via a letter-number task where participants (N=32) were instructed to classify the presented stimulus pair (number, letter) either as vowel or consonant (letter) or as even or odd (number), depending on cues that were provided before the stimulus pair. Switch costs were measured by how long people took to respond when the trial was the same classification task as the previous trial versus when it was different. Surprisingly, HMMs responses were 167ms slower than those of LMMs (p < .01); HMMs were 426ms slower at responding when there was a classification switch (p < .01), and 259ms slower on non-switch trials (p < .03).
The findings of this final experiment are particularly unexpected since task switching seems to be a large part of multitasking; nevertheless, the data indicate that HMMs have a higher task switching cost. This result, when combined with the others of Ophir, Nass, and Wagner (2009) signify specific processing differences in HMMs and LMMs. Such results, when viewed in light of the previous debate regarding media multitasking seems to imply that while media multitasking is indeed a skill set that can be learned, it has some negative effects on specific attention tasks, and are particularly vulnerable to interference of irrelevant stimuli. Furthermore, given that the types of tasks that are commonly combined in multitasking situations (Carrier et al, 2009) are similar across generations, despite differences in practice and prevalence of multitasking indicates that the cognitive limitations of information processing are not easily bypassed. Indeed, much of the research looking at the effects of multitasking with technology, indicate a cost in performance, particularly on academic tasks such as reading (Armstrong & Chung, 2000; Bergen, Grimes, & Potter, 2005; Hembrooke & Gay, 2003; Levine Waite and Bowman, 2007).
Future research is would be useful in looking more directly at the effects of media multitasking on academic outcomes, as many of the studies discussed in the article explored one of the above without manipulating or directly measuring the other. This difficulty arises when considering the ethical obstacle of obtaining measures of participants’ private media use as well as academic performance indicators. Nevertheless, real-world measures will help to provide strong recommendations for the use of technology such as laptops in academic settings, and instructional implications for technology use. Additionally, generalizability to younger populations is sorely impractical given that most of the studies have been conducted with college students. Implications for youth who may be developing media multitasking skills are difficult to make given the narrow sample of many of the studies, as well as developmental differences that teens and preteens will show in comparison to college students.
Armstrong, G.B. & Chung, L. (2000). Background television and reading memory in context: Assessing TV interference and facilitative context effects on encoding versus retrieval processes. Communication Research, 27, 327-352.
The experiment carried out was intended to assess where in processing interference from background TV is active. The authors proposed three areas where interference may occur, encoding, retrieval, and comprehension. Results indicate that elaborative rehearsal is impeded by background television, thereby disrupting the encoding process.
Bergen, L., Grimes, T., & Potter, D. (2005). How attention partitions itself during simultaneous message presentations. Human Communication Research, 31 (3), 311-336.
This article describes four experiments used to answer research questions regarding how people process more complex input presented on news programs (complex visual material with banners and tickers running along the bottom of the screen, as well as auditory stimuli from the newscasters). The data collected supports previous research findings that indicate that people do show attentional limitations that cause depreciation in recognition when presented with more complex visual material. Furthermore, they hypothesize that the media’s target demographic may be more influential in decisions that support utilizing these more complex presentation styles.
Carrier, L.M., Cheever, N.A., Rosen, L.D., Benitez, S., &Chang, J. (2009). Multitasking across generations: Multitasking choices and difficulty ratings in three generations of Americans. Computers in Human Behavior, 25(2), 483-489.
This article describes generational trends in multitasking among three generations: Baby boomers- born between 1946 and 1964, Generation X- born between 1965 and 1979, and the Net Generation- born between 1980 and the present. An online questionnaire was presented to 1319 participants asking about 12 tasks typically done at home: surfing the Web, Offline computing, Emailing, Instant Messaging/online chatting, Using the telephone, Tex messaging, Playing video games, Listening to music, Watching television, Eating, reading books and magazines for pleasure, and Talking face-to-face with someone. The authors draw conclusions regarding the relative ease and increased self-report of multitasking with media that members of the later generations exhibit over earlier ones; however, their results also indicate that there are commonalities across generations regarding which type of tasks to combine, perhaps indicating the cognitive limitations suggested by previous research.
Foehr, U.G. (2006). Media multitasking among American youth: Prevalence, predictors, and pairings. Menlo Park, CA: Kaiser Family Foundation.
This report was compiled based on responses of 694 3rd-12th grade students’ diary responses and survey responses from 7th-12th grade students regarding their engagement in 12 activities which include both media tasks (email, IM, etc.) as well as non-media tasks (reading, eating, etc.). Data was analyzed to determine patterns of multitasking behaviors.
Hembrooke, H., & Gay, G. (2003). The laptop and the lecture: The effects of multitasking in learning environments. Journal of Computing in Higher Education 15, 46-64.
The authors observed increased media multitasking occurring during their lectures and sought to examine whether this was affecting academic performance. A simple two group design was conducted for the span of one lecture, and academic performance assessed by a pop quiz looking at both recall and recognition. Overall higher performance was noted in participants who participated in lecture with laptops closed. No data was collected regarding particular laptop use during the lecture.
Lee, F. J., & Taatgen, N. A. (2002). Multitasking as skill acquisition. In W. D. Gray & C. D. Schunn (Eds