Applied Cognitive Psychology, Appl. Cognit. Psychol.

27: 593–599 (2013)

Published online 18 March 2013 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/acp.2919

Does Multitasking Impair Studying? Depends on Timing

HAROLD PASHLER*, SEAN H. K. KANG and RENITA Y. IP

Department of Psychology, University of California, San Diego, USA

Summary: It is often said that contemporary students frequently study while ‘multitasking’. However, this rather diffuse term
encompasses situations that vary as to the whether the learner controls the pace at which educational materials are provided.
On the basis of prior cognitive research, we hypothesize that this may well be a critical determinant of interference. Three studies
required students to read or listen to several short historical narratives and also to engage in five to eight very short conversations
(akin to an instant messaging conversation). In Experiment 1, subjects read the narratives; here, multitasking marginally
increased total time spent reading the narratives, especially when it occurred at random times. However, final memory for the
narratives was not significantly affected. Similar results were obtained when the narratives were presented in audio format and
the learner could pause them while conversing (Experiment 2). By contrast, when audio narratives did not pause, interruptions
reduced comprehension performance (Experiment 3). Copyright © 2013 John Wiley & Sons, Ltd.

Supporting information may be found in the online version of this article.

It is a commonplace observation that the proliferation of from this research is the fact that the time sequence over
digital technology that has occurred in recent years has led which processing demands unfold is normally critical in
to a much greater amount of ‘multitasking’ in daily life. Of determining the observable consequences of multitasking.
course, even before widespread digital technology, it was For example, in the Psychological Refractory Period design,
undoubtedly common for people to manage multiple tasks a person is required to perform two speeded choice tasks
that overlapped in time (e.g., riding a horse and having a with stimuli presented very close together in time. In this
conversation). However, it seems very plausible that situation, there is often little measurable interference when
contemporary digital technology tempts or even requires the stimulus onset asynchrony (temporal separation from
people to switch from one source of visual or verbal informa- the presentation of one stimulus to the presentation of the
tion to another more frequently than would have been the case second) is, say, 1200 milliseconds; by contrast, when the
in the past. same pair of tasks is performed with greater temporal
Popular discussion of the increasing frequency of multi- overlap (say, with an asynchrony of 100 milliseconds), the
tasking features two prominent—and rather contradictory— time taken to perform the second task may be almost
strands of thought. One is a very broad and undifferentiated doubled (e.g., Pashler, Harris, & Nuechterlein, 2008, Figure 6).
condemnation of multitasking in all forms, especially as en- The point, then, is that whether two tasks show interference
gaged in by students who combine studying with other tasks depends a great deal on the temporal structure of the
(e.g., Golovan, 2011; Straus, 2010). According to advocates situation. This finding seems to have had little impact on
of this negative view, multitasking is a contemporary curse, popular discussions of the effects of multitasking on educational
imperiling not only safe driving but also the educational and other real-world contexts (e.g., Straus, 2010). One of the
achievement of modern youth. A second common idea, purposes of the present paper is to present empirical evidence
starkly opposed to the first, is that a new generation of arguing that the effects of multitasking on real learning tasks
‘super-tasking’ young people is emerging whose brains have are highly dependent on differences in timing (albeit over
developed differently to the point that they have become substantially greater time periods than those discussed in the
superior multitaskers (e.g., Wallis, 2006). Both views seem refractory paradigm).
to rest chiefly on casual observation and conjecture.

Multitasking and retention of information

Dual-task research Many cognitive researchers have performed studies in which
The costs of multitasking1 (often termed ‘dual-task interfer- people are presented with stimuli (such as word lists) while
ence’; footnote 1) have been the subject of experimental engaged in some secondary task and given the goal of trying
research going back to the 1950s (for reviews, see Pashler & to commit these stimuli to memory. The general finding is
Johnston, 1998). One of the most robust findings to emerge that even if the secondary task does not involve sensory
inputs in the same modality as the stimuli to be remembered,
*Correspondence to: Harold Pashler, Department of Psychology 0109, multitasking still substantially impairs later memory for the
University of California, San Diego, La Jolla, CA 92093, USA. stimuli (e.g., Baddeley, Lewis, Eldridge, & Thomson,
E-mail: [email protected]
1984; Craik, Govoni, Naveh-Benjamin, & Anderson, 1996;
1
The term ‘multitasking’ is used here in a broad sense, to encompass any Mulligan, 1998; Naveh-Benjamin, Craik, Guez, & Dori,
situation in which a person seeks to perform multiple tasks simultaneously 1998). The problem appears to be a failure of storage, rather
or in a way that involves switching back and forth between one task and an- than a mere change in mental context, because if people
other. Empirical research on multitasking appears under a wide range of
headers, including dual-task performance, dual-task interference, divided multitask at the time of retrieval, this further exacerbates the
attention, task switching, and task interruption. interference, rather than alleviating it (Baddeley et al., 1984).

Copyright © 2013 John Wiley & Sons, Ltd.

594 H. Pashler et al.

In light of this well established literature, it might seem On the basis of what has been described thus far, it might
obvious that any sort of multitasking occurring while people appear that any careful examination of multitasking effects
study educational materials is bound to be harmful. Despite on studying is bound to disclose that multitasking either
great interest in this topic, there has been rather little harms learning or has no effect. In fact, though, it does not
controlled research in this area. A recent study by Fox, seem inconceivable that multitasking could potentially
Rosen, and Crawford (2009) is pertinent, however. These enhance learning in certain situations. As Bjork and
researchers had 69 subjects read texts (described as ‘SAT colleagues have pointed out, manipulations that add
passages’ or ‘GRE passages’, whose length was not complications to the tasks performed during study may often
mentioned). Subjects read the passage either in isolation or reduce performance during learning but enhance the success
while engaging in a conversation with an experimental of learning as assessed by performance on a later test
confederate using the internet chat utility AOL Instant (Schmidt & Bjork, 1992; Battig, 1979). Thus, one could
Messenger. They were instructed ‘to take as much time as imagine that frequent interruptions, requiring the learner to
they needed to read and understand the passage’ (p. 52). A periodically retrieve and reactivate the meaning of the last-read
test on the contents of the passage was given (sometimes portion of the passage, might enhance learning as assessed
while the instant messaging task continued). The results with a later test.
showed that concurrent instant messaging produced an To sort out these possibilities, we performed three experiments,
increase in reading times from 3.33 minutes to 5.53 minutes, examining the consequences of multitasking superimposed upon
a highly significant difference. (It appears that the reading studying of narrative texts—with attention paid to the temporal
time estimates were measured from the start of reading to structure of the situation.
the completion, although this was not completely clear from
the method section.) Interestingly, however, there appeared EXPERIMENT 1
to be no significant effect of concurrent task upon memory
for the contents of the passages (although no information In the first experiment, subjects read three different passages
was provided in that paper about effect sizes and variability while experiencing a different multitasking condition for
or even mean levels of performance). each. The secondary task was chosen to provide at least a
Why might there be so little interference here? One rough experimental equivalent to composing a text message.
important point of contrast between this study and the Two of the conditions required subjects to switch to this
memory experiments described previously is the following. secondary task several times while reading (they differed in
The subjects in Fox et al. (2009) were allowed to take as terms of whether the interruptions occurred randomly or only
much extra time as they needed for the reading task when at the end of paragraphs). The third condition was the read-
it was performed with multitasking. By contrast, the subjects only control in which the subject did not have to engage in
in the laboratory experiments by Baddeley et al. (1984) and the secondary task. At the conclusion of each passage,
the other studies mentioned previously were given a fixed subjects were tested on their comprehension of the material
period to study a set of words, and when a concurrent task they had just read.
was imposed, subjects were not allowed to compensate by
devoting extra time to the word lists. However, it is impor- Method
tant to note that even when adequate time is potentially avail-
able to complete an interrupted task, this completion may or Subjects
may not take place (as Dodhia & Dismukes (2009) point out, From the University of California, San Diego, 109 under-
the completion itself represents a prospective memory task, graduate students participated in return for course credit.
and prospective memory is notoriously fallible.)
Design
The design was a within-subjects comparison between three
different conditions: a control condition (Read-only) involving
The present study
reading without any interruptions, a multitasking condition
From a practical standpoint, both situations (multitasking with interruptions at paragraph breaks (Multitasking-
with and without the potential for compensatory increase in Paragraph Break, denoted MT-PB), and multitasking with
total time devoted to the materials to be remembered) would interruptions at random times (MT-Random). Each subject
seem to be of practical importance to real-world educational reads three passages, one in each of the conditions. The assign-
concerns. The student reading a textbook or listening to an ment of texts to conditions was counterbalanced across subjects.
audiotape or videotape while experiencing frequent interrup- The order in which the passages were presented (and hence the
tions can often take as much extra time as he or she needs. order of conditions) was individually randomized for each
(Of course, time is a limited resource, so it would be impor- subject. The critical dependent variables were as follows: (1) time
tant to know if the sum of the time needed for both tasks is spent reading the texts and (2) performance on comprehension
likely to be greater than it would have been if the two tasks questions relating to each of the three texts.
had been performed serially.) In other situations, however,
such as a student watching a lecture or educational film or Materials
a class discussion (while potentially engaging in some Study passages were approximately 1500 words long and
additional task such as text messaging), the situation would consisted of nine to ten paragraphs, written in an academic
not offer any option to pause the educational presentation. style and chosen to be interesting but unfamiliar to the

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
 Does multitasking impair studying? 595

typical US college student. The passages had a mean Flesch After subjects finished reading each passage, they took a
reading ease score of 45.30 and a Flesch–Kincaid reading comprehension test on the material. Questions were
level of 12.57. The three passages dealt with the Canadian presented one at a time, with no option to return to previous
Arctic expeditions of John Franklin, the history of cheese questions and no feedback about the correctness of
production, and the history of manufactured fibers responses. Upon the completion of the last comprehension
(Supporting Information). The secondary task in the two test, subjects completed the questionnaire described
multitasking conditions required subjects to answer opinion previously (APPENDIX B).
questions (presented one at a time), with responses ranging
from one to three sentences in length. These opinion
questions (APPENDIX A) are related to current events and Results and discussion
topics of general controversy. The comprehension tests The mean reading times are displayed in Figure 1. As can be
given on the study passages consisted of 25 items each, of seen in Figure 1, there was a trend toward an increase in
which 18 were in true/false format and seven were four- reading time for the conditions in which reading was
alternative multiple-choice questions. At the end of the interrupted—and particularly when the interruptions
experiment, subjects were given a questionnaire with ques- occurred while the subject was reading individual paragraphs.
tions on the following: (i) subjects’ personal practices in regard A repeated measures analysis of variance (ANOVA) revealed
to multitasking, instant messaging, and texting; (ii) which a marginally significant effect of study condition, F(2,
articles they found interesting; and (iii) which condition they 216) = 2.75, p = .066,
p2 = .025. Pairwise comparisons among
thought they performed the best in (APPENDIX B). the three conditions indicated that the only significant
difference was between the MT-Random and Control condi-
tions, t(108) = 2.31, p = .023 (uncorrected), d = 0.22.
Procedure The mean comprehension scores are shown in Figure 2.
Subjects were seated at a computer terminal and instructed Differences between conditions were minimal, and a re-
that they would be reading three passages, one under each peated measures ANOVA showed no significant effect of
of three different task conditions. They were also told that study condition, F(2, 216) = 1.96, p = .144. The effect size
sometimes they would be asked to converse in writing about for the difference between the Control and the MT-PB
their opinions on questions posed to them while they were condition was d = 0.12, and the effect size for the difference
reading the passages. Opinion questions were presented in between the Control and the Random was d = 0.07. These are
a separate and smaller pop-up window; when the opinion both in the range commonly understood to be extremely
question was presented, the text of the passage was not small effects.
visible. Subjects were required to provide at least 160 characters When asked on a questionnaire at the end of the experiment
of response (when they submitted a response briefer than that which condition they thought they performed best and worst
they were prompted to elaborate). Subjects were instructed in, the majority of subjects thought that their comprehension
that their most important goal would be to learn as much infor- was worst in the MT-Random condition (57%; 17% answered
mation as possible from the passages—on which they would MT-PB; 26% answered Read-only) and best in the Read-only
later be quizzed. condition (48%; 34% answered MT-PB; 18% answered
In the Read-only (control) condition, the passage was MT-Random). Also, when asked whether they thought that
presented on the computer screen one paragraph at a time ‘instant messaging’ or ‘texting’ while studying affects
(paragraphs contained approximately 150–200 words). The memory for the material, 89% answered in the affirmative.
subject clicked on a button labeled ‘Next’ with a mouse to Interestingly, even though the majority of subjects seemed
advance to the next paragraph. Subjects read at their own to have the metacognitive belief that multitasking hurts
pace but were not allowed to return to previous paragraphs. reading comprehension, 89% of subjects indicated that they
In the MT-PB condition, opinion questions were multitasked ‘sometimes’ or ‘often’ while studying.
displayed immediately after the subject clicked on the Next
button to advance to a new paragraph. These interruptions
occurred on a randomly chosen five out of the eight 500
paragraph breaks.
Reading Time (s)

450
In the MT-Random condition, the interruptions took place
within a randomly chosen five out of nine paragraphs. At a 400
randomly chosen time point between 5 and 15 seconds after
350
they had started reading a paragraph, the opinion question
was presented, and the text that was being read disappeared 300
from the screen during that time. In each of the multitasking
250
conditions, after the subject had typed a response to the Control MT-PB MT-Random
opinion question into the box provided on the computer Condition
screen, she/he was returned to the passage reading task (in
the MT-PB condition, she/he would go on to view the next Figure 1. Mean reading time as a function of study condition in
Experiment 1. The times for the multitasking conditions reflect the
paragraph, whereas in the MT-Random condition, the same total time spent both before and after the opinion question but exclude
text display that had been visible prior to the interruption the time during which the subject viewed and responded to the
would reappear on the screen). opinion question. Error bars represent 95% confidence intervals

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
596 H. Pashler et al.
16
remember as much information as possible about the
Comprehension Score
15
passages because they would later be quizzed on them.
14
Subjects listened to the audio presentation of the material
(Max. = 25)

13
through headphones and were unable to voluntarily pause or
12
11
fast-forward through the content. In the Listen-only
10
condition, subjects listened to an audio presentation of the
9
material, which was approximately 9 minutes long, straight
8
through without any interruptions. In the MT-PB condition,
Control MT-PB MT-Random subjects listened to the audio presentation with five different
Condition interruptions in between paragraphs during which the audio
Figure 2. Mean comprehension score as a function of study condi-
playback stopped and they responded to opinion questions.
tion in Experiment 1. Error bars represent 95% confidence intervals In the MT-Random condition, the interruptions took place
within five of the paragraphs, at times that would have
appeared random to the subject. The timing of the ‘random’
EXPERIMENT 2 interruptions was actually predetermined so as to avoid
having interruptions in the middle of words, which would
The second experiment posed the same question as the first have impaired auditory word recognition. (For each passage,
using the same educational materials, the same conversa- three different random timings for the five interruptions were
tional task, and the same tests. The main difference, created in advance—with the constraint that the interruptions
however, was that the educational materials were audio would not occur mid-word—and the particular set of
taped and presented in spoken form. Aside from the interruption timings used for each subject was randomly
difference in modality, there was also a rather subtle selected by the computer.) In keeping with the presentation
difference: When a person is interrupted in reading, they of the material in Experiment 1, the audio presentation
could potentially start reading at point prior to where they paused while the conversational task was performed. During
left off, thus providing themselves with a brief review of these pauses, the audio stopped playing, and a box appeared
the final portion of the reading materials; with an audiotape on the computer screen containing an opinion question and a
that pauses and then resumes at the exact point of interruption text box to provide a response. Once the subject finished
(as in Experiment 2), this opportunity is not present. typing a response and pressed the submit button, audio
Conceivably, this could affect the results. playback continued where it had left off.
After subjects finished listening to each passage, they took
Method a comprehension test on the material, followed by a
questionnaire at the end of the experiment, presented in
Subjects \exactly the same way as in Experiment 1.
From the University of California, San Diego, 109 under-
graduate students participated in return for course credit.
Results and discussion
Design The mean comprehension scores are shown in Figure 3.
The design was a within-subjects comparison between three Differences between conditions were minimal, and a
different conditions: a control condition involving listening repeated measures ANOVA showed no significant effect of
without any interruptions (Listen-only), MT-PB, and MT-Random. study condition, F(2, 216) = .678, p = .509. The effect size
Each subject listened to three passages, one in each of the for the difference between the Control and the MT-PB
conditions. The assignment of passages to conditions was condition was d = 0.02, and the effect size for the difference
counterbalanced across subjects. The order in which the between the Control and the Random was d = 0.11 (again,
passages were presented (and hence the order of conditions) extremely small effects).
was individually randomized for each subject. The critical The questionnaire at the end of the experiment contained
dependent variable was performance on comprehension the same questions as in the previous experiment except that
questions relating to each of the passages.

Materials
Experiment 2 used the same educational materials, the same
conversational task, and the same tests as in the first
experiment.

Procedure
Subjects were seated at a computer terminal and instructed
that they would be listening to three passages, one under
each condition. They were told that sometimes, they would
have to answer unrelated opinion questions while they were Figure 3. Mean comprehension score as a function of study condi-
listening to the passages but that their main goal was to tion in Experiment 2. Error bars represent 95% confidence intervals

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
 Does multitasking impair studying? 597

it included a question asking whether students used podcasts In the Listen-only condition, subjects listened to the entire
or audiobooks to study. Of subjects, 54% reported using nine-minute passage all the way through. There was no
podcasts or audiobooks to study ‘sometimes’ or ‘often’, way for the subject to pause or advance the passage. In the
whereas 46% use these methods ‘rarely’. The majority of MT-Random condition, the subject was presented with five
subjects thought that their comprehension was worst in the opinion questions, presented randomly throughout the
MT-Random condition (52%; 16% answered MT-PB; 32% presentation of the passage in a separate, smaller pop-up
answered Listen-only) and best in the Listen-only condition window. At the presentation of this pop-up window, an alert
(45%; 41% answered MT-PB; 14% answered MT-Random). noise sounded—a very brief beep similar to one produced by
Although 87% of respondents believed that instant/text an instant messaging program or a text alert on a cell phone
messaging while studying affected their recall memory, During the time that the opinion question was presented, the
67% of subjects said they still did so while studying. audio reading of the passage continued to play. Again, there
was no way to pause or advance the speed of the audio
passage. To ensure that subjects were performing the
EXPERIMENT 3
secondary task, we informed subjects that they had 40 seconds
to answer the question, and in the same window, the time
The first two studies reported previously disclosed that
remaining was counted down. After 40 seconds, the pop-up
whether educational passages were read or spoken, a
window went away.
secondary conversational task presented so that the learner
After subjects finished listening to each passage, they took
had the opportunity to pause the presentation of the educa-
a comprehension test, followed by a questionnaire at the end
tional narratives while doing the conversational task produced
of the experiment, just as in Experiments 1 and 2.
(a) little cost to later memory for the narratives and (b) a slight
increase in the total time spent reading the material. With
regard to (b), it should be kept in mind that the slight increase Results and discussion
is over and above the extra time taken for the secondary task. The mean comprehension scores are shown in Figure 4.
As noted in the Introduction, however, many real-world There was a significant decrease in performance on the
multitasking situations do not allow the learner to control the comprehension test for subjects in the MT-Random
presentation of educational content and to pause presentations condition, relative to the Listen-only condition (the means
at will (listening to a lecture in class is one obvious example of in the two conditions were 45 and 54, respectively). A
this). Experiment 3 used the same audio presentations as repeated measures ANOVA showed a significant effect of
Experiment 2, except that here, the audio presentations did the study condition, F(1, 81) = 26.38, p < .001,
p2 = .25.
not stop while the person carried out the conversational task The difference corresponded to a Cohen’s d effect size of
(as in the previous experiments, the conversational task did 0.57, normally classified as a medium-sized effect.
not involve audio input or output).
GENERAL DISCUSSION
Method
Subjects The results can be summarized as follows. Multitasking
From the University of California, San Diego, 82 undergrad- produced a marked and substantial reduction in information
uate students participated in return for course credit. acquired from educational materials when the materials were
presented in spoken form and played without waiting for the
learner. On the other hand, when the learner read the
Design materials at his or her own pace, there was no sizable or
The design was a within-subjects comparison between two significant reduction in information acquired. This was true
different conditions: Listen-only and MT-Random. Each even when the interruptions occurred at moments chosen
subject listened to two passages being read, one in each of by the experimenter, rather than the learner. Finally, listening
the conditions. The assignment of passages to conditions to the materials and pausing to do the concurrent task were
was counterbalanced. The ordering of the passages (and also relatively harmless.
hence the ordering of the conditions) was randomized
for each subject. The critical dependent variable was
performance on comprehension questions relating to each 15
Comprehension Score

of the passages. 14
13
(Max. = 25)

Materials 12
Experiment 3 used the same educational materials, the same 11
conversational task, and the same tests as in the first 10
two experiments. 9
8
Listen-only MT-Random
Procedure
Experiment 3 used the same procedure as Experiment 2, Condition
the only difference being the circumstances of the Figure 4. Mean comprehension score as a function of study condi-
multitask condition. tion in Experiment 3. Error bars represent 95% confidence intervals

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
598 H. Pashler et al.

The results have practical and theoretical implications. may overcompensate for interruption, in which case
The practical implications are that there is no reason to proximity could (for some or all subjects) be associated with
believe that people studying factual material and permitting improvement rather than cost.
themselves to be interrupted on many occasions are In any case, from a practical standpoint, whether or not
consequently likely to learn much less. Undifferentiated one chooses to draw the conclusion that multitasking in this
warnings against the perils of multitasking while studying situation is ‘harmless enough’ should depend upon the
(e.g., Golovan, 2011; Straus, 2010)—at least when the practical situation one finds oneself in. If the goal is
studying involves reading fact-dense text material—seem absolutely to maximize performance, then, it would probably
overblown. On the other hand, we do not see any ‘desirable be wise to avoid all forms of multitasking.
difficulty’ effect (Schmidt & Bjork, 1992) whereby the Second and perhaps more importantly, the learning task
interruptions would enhance learning. Our results also here required the learner to comprehend and store many facts
indicate that when the presentation of educational materials in long-term memory, and it seems like a reasonable model
is paced by events in the world—as it is in a lecture situation, for many tasks that go by the name ‘studying’. However, it
for example—multitasking can have a large detrimental did not require the learner to synthesize new contents of their
effect on learning—generally in keeping with the literature own, as for example in constructing a math proof, writing a
on ‘divided attention effects on memory encoding’ (e.g., computer program, or writing an essay. It seems plausible
Baddeley et al., 1984). Although these effects of temporal to the current authors that frequent interruption during this
overlap are perhaps not entirely surprising from a common- sort of synthetic tasks might well be more injurious to
sense standpoint, it is striking that this distinction between performance (and perhaps also to the acquisition of greater
what might be termed interruption multitasking and simulta- skill). Thus, it would be a mistake to conclude from the
neous multitasking has not, as far as we can tell, figured in present results that multitasking while doing schoolwork is
popular discussions of the perils or benefits of multitasking. quite harmless so long as it is possible to pause the educa-
As mentioned in the Introduction (footnote 1), formal tional task—as a superficial reading of the present results
studies of multitasking (in the broad sense in which that term might imply. Testing the effects on a range of synthetic
is used here) have appeared under a variety of labels, academic tasks would seem to be an important task for future
including divided attention, task switching, interruption, research on multitasking and education, but it will be
and dual-task performance. It would be tempting to say that challenging given the difficulties of obtaining reliable and
the results described here suggest that interruption produces valid measures of the quality of the output people produce
fairly minor effects on memory storage compared with true in such tasks.
dual-task performance. However, research on bottleneck
models of divided attention suggests that a great deal of what
appears superficially to be parallel performance of two tasks ACKNOWLEDGEMENTS
at the same time is actually accomplished by a covert
switching at the level of central processing operations, This work was supported by the Institute of Education
necessitated by the existence of a central processing bottleneck Sciences (US Department of Education, Grant R305B070537
(Pashler, 1998, Chapter 7). It is not entirely clear whether to H. Pashler), the National Science Foundation (Grant
encoding of information into long-term memory is subject to BCS-0720375, H. Pashler, PI; and Grant SBE-0542013,
this same bottleneck in whole or in part (Pashler, 1998, Chapter G.W. Cottrell, PI), and a collaborative activity award from
8). If it is, then it would seem accurate to say that even the J.S. McDonnell Foundation. We also thank David Yee
seemingly concurrent performance of a task and storage of and Daniel Price for programming the experiments.
information in memory actually involves a covert process of
interruption and resumption.

REFERENCES
Limitations
Several very important potential limitations should be noted. Baddeley, A., Lewis, V., Eldridge, M., & Thomson, N. (1984). Attention
and retrieval from long-term memory. Journal of Experimental Psychology.
First, while the effects of multitasking in the pause
General, 113(4), 518–540.
conditions are small (as quantified, for example, with Battig, W. F. (1979). The flexibility of human memory. In L. S. Cermak, &
Cohen’s d or when viewed as a proportional reduction in F. I. M Craik (Eds.), Levels of Processing in Human Memory. Erlbaum:
learning), a more sensitive test might have disclosed some Hillsdale, NJ, 23–44.
effects. To the current authors, it seems implausible to sup- Craik, F. I. M., Govoni, R., Naveh-Benjamin, M., & Anderson, N. D.
(1996). The effects of divided attention on encoding and retrieval
pose that the true costs are exactly zero. For example, it is
processes in human memory. Journal of Experimental Psychology.
possible that content close to the interruption point suffers General, 125(2), 159–80.
from the interruption, but this effect was undetectable in Dodhia, R. M., & Dismukes, R. K. (2009). Interruptions create prospective
the current study because it was aggregated together with memory tasks. Applied Cognitive Psychology, 23, 73–89.
content further from the interruption point. Follow-up Fox, A. B., Rosen, J., & Crawford, M. (2009). Distractions, distractions: Does
research on this point could potentially use two instant messaging affect college students’ performance on a concurrent
reading comprehension task? Cyber-Psychology and Behavior, 12, 51–53.
counterbalanced sets of interruption points to determine if Golovan, L. (2011, January). Is multitasking harmful or helpful to
proximity to the interruption point is associated with greater teenagers? Your Teen Magazine. Retrieved from http://yourteenmag.
costs. Of course, in principle, it is also possible that people com/2011/01/is-multitasking-helpful-or-harmful-to-teenagers/

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
 Does multitasking impair studying? 599

Mulligan, N. W. (1998). The role of attention during encoding in implicit APPENDIX 2

and explicit memory. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 24(1), 27–47. Questionnaire
Naveh-Benjamin, M., Craik, F. I. M., Guez, J., & Dori, H. (1998). Effects of
divided attention on encoding and retrieval processes in human memory:
Further support for an asymmetry. Journal of Experimental Psychology: How often do you multitask while studying?
Learning, Memory, and Cognition, 24(5), 1091–1104.
Pashler, H. (1998). The Psychology of Attention. MIT Press: Cambridge,
(i) Rarely (ii) Sometimes (iii) Often
MA. Do you use any kind of instant messaging program or SMS
Pashler, H., & Johnston, J. C. (1998). Attentional limitations in dual-task
performance. In H. Pashler (Ed.), Attention. 155–189). Hove, England:
(texting)?
Psychology Press. (i) Rarely (ii) Sometimes (iii) Often
Pashler, H., Harris, C., & Nuechterlein, K. (2008). Does the central
bottleneck encompass voluntary selection of hedonically-based choices? Do you open your instant message program or text while
Experimental Psychology, 55, 313–321. studying?
Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice:
Common principles in three paradigms suggest new concepts for training. (i) Yes (ii) No
Psychological Science, 3, 207–217.
Straus, V. (2010, May 24). Data show kids shouldn’t multitask. Washington Do you think that instant messaging/texting while studying
Post. Retrieved from http://voices.washingtonpost.com/answer-sheet/ affects your recall memory?
guest-bloggers/data-shows-kids-shouldnt-multi.html
Wallis, C. (2006, March 27). genM: The multitasking generation. Time (i) Yes (ii) No
Magazine. Retrieved from http://www.time.com/time/magazine/article/
0,9171,1174696,00.html Was the article about manufactured fibers interesting to
you?
(i) Yes (ii) No
APPENDIX 1
Was the article about the Franklin Expedition interesting
Opinion Questions to you?
(i) Yes (ii) No
During your midterm you notice two students sitting nearby
cheating on their test. What would you do? Was the article about the history of cheese interesting to you?
What is your opinion of people spanking their children? (i) Yes (ii) No
Is capital punishment ever justified, even when there is solid
evidence linking the accused to the crime? Why or why not? In which task do you think you did best?
Should the family of the murder victim have the right to (i) No Interruptions
decide the murderer’s fate? Why or why not? (ii) Random Interruptions
What are your beliefs on animal research? When is animal (iii) Paragraph Interruptions
research justified? In which task do you think you did worst?
Should marijuana be legalized? Why or why not?
(i) No Interruptions (ii) Random Interruptions
What are your views on former vice-presidential candidate
(iii) Paragraph Interruptions
Sarah Palin?
Under what conditions would you help a homeless person? We are trying to understand how multitasking during
Do you believe that the new UC fee hikes are necessary studying affects students’ performance. Please leave any
given the lack of state funding for higher education? Why comments describing how well you think you performed
or why not? during the different conditions and why.

Copyright © 2013 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 27: 593–599 (2013)
Copyright of Applied Cognitive Psychology is the property of John Wiley & Sons, Inc. and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.