The results of numerous past studies have indicated that individuals have a limited ability to perform multiple tasks simultaneously. Moreover, negative effects on memory have been identified in situations where subjects were engaged in divided attention tasks at the time of memory encoding. The present study examined the effect of device-delivered notifications on the accuracy of memory through the use of a word recognition paradigm. Two groups of students were presented with an identical word list and were then tested with a two-alternative forced-choice recognition test consisting of a combination of previously presented and novel words. The first group received numeric notifications during word presentation which they were instructed to note down while the other group received no notifications. The results showed that false alarm rates were significantly greater and hit rates were significantly lower for the group that received notifications, demonstrating a mirror effect, and thereby providing support for the hypothesis that notification-derived divided attention has a negative impact on memory. Therefore, using one's phone while doing something else (whether that be studying, watching a film, or enjoying a record) really does impact one's ability to focus on what one is doing, regardless of what many people might claim.
Keywords: divided attention, two-alternative forced-choice recognition, multitasking, notifications
Technology, in recent years, has become incredibly pervasive in contemporary society, for a multitude of reasons, not least of which has been the increasing portability of electronic devices. For instance, it was reported by Statistics Canada (2017) that approximately 76% of Canadians owned a smartphone. This rapid spread of technology has led to an increased sense of connectivity between individuals, and has almost undeniably generated a slew of numerous benefits. However, the rapid spread of portable technology has also been found to have significant downsides with a considerable amount of research produced on the potential dangers and downsides of using such technology in various contexts. Perhaps one of the most widely reported risks of using portable technology has been the increased number of distracted driving accidents caused by drivers who used cellular devices while operating a motor vehicle. Specifically, according to the National Safety Council (2013), as many as 27% of all automobile accidents that occurred in the United States in said year were caused by drivers who were absorbed by the screens of their cellular devices. Moreover, other studies have found that operating a cellular phone while driving was as dangerous as driving drunk due to the magnitude of distraction that the use of a cellular phone causes the driver (Strayer, Drews, & Crouch, 2006).
Clearly, despite the many claimed benefits of mobile technology, significant downsides to its use have likewise been found, including its noted formidable ability to distract its users. However, the popularity of mobile technology and its wide set of applications and capabilities has led to its acceptance in a variety of contexts and environments, including most workplaces and places of education, concerningly including universities and elementary schools alike. As an unsurprising consequence, the majority of the student and working population alike has become likely to have some form of technological device in their possession at all times and moreover, such devices are being actively utilized in most classroom and workplace settings. This effect had been examined in the educational context by a variety of researchers who found that when students utilized devices in the classroom setting, they effectively divided their attention between their device and the lesson being taught (Glass & Kang, 2018). Furthermore, even when such devices are not in active use, many device applications create audible and/or tactile notifications that are explicitly designed specifically to capture the attention of their users. Such notifications themselves have been shown to potentially lead to attention-related errors even in cases when they were not actively attended to (i.e., 'ignored') by subjects (Marty-Dugas, Ralph, Moakman, & Smilek, 2018).Therefore, the use of portable devices in their various shapes and forms has been found to lead to a society-wide phenomenon of divided attention. Importantly, a large body of research had been conducted on divided attention in general, both prior to and following the advent of mobile technology, which concluded that human ability to split attention between multiple tasks is severely limited (Broadbent, 1958). Building on this foundation, more contemporary researchers examined the effects of multitasking involving technology in the educational setting. As an illustrative example, a study by Hembrooke and Gay (2003) was carried out to determine whether students who used technology during lectures demonstrated significantly decremented memory for the content of lectures in comparison to students who were not permitted to use such devices. In other words, if the results of the experiment demonstrated that the students who used technology at the time of encoding demonstrated worse memory performance than those who did not, this would support the theory that technology had an adverse effect on memory.
In the experiment conducted in the aforementioned Hembrooke and Gay (2003) study, the researchers assigned students to two conditions, wherein half of the subjects were instructed to use their devices during the lecture and the other half were instructed not to use them. Following the end of the lesson, both groups were evaluated on their memory for the content of the lecture. This process was replicated with the two groups of subjects switching roles, wherein the students who were allowed to use devices in the first part of the study were prohibited from using them in the second part, and vice versa. The results of the study clearly indicated that the students that were allowed to use technology during the lecture performed significantly worse on the recognition test in both the original and the replicated portion of the study. This led the researchers to conclude that the students that were allowed to use devices effectively performed multiple tasks simultaneously during the lecture, and it was this fact that led to the significantly lower scores that this group demonstrated at time of testing.
A follow-up study was led by Glass and Kang (2018) in order to further investigate the role of technology in the classroom and its role in creating a divided attention effect. In this study, subjects were also separated into two groups wherein the first group was permitted the use of devices during classes held on Tuesday and not during those on Thursday, while the second group was allowed to use their devices only during lectures held on Thursday and not during those on Tuesday. The memory of the students was assessed during each class, and also with final and unit exams. Therefore, this study extended the study conducted by Hembrooke and Gay (2003) in that the subjects were tested over 23 lectures rather than just one.
The results of this study demonstrated that while the performance of students on the daily classroom quizzes was not affected by whether students were allowed to use their devices, the exam performance of the students was worse for the material that was taught during the classes in which devices were permitted. The researchers therefore concluded that while short-term memory and comprehension was perhaps unaffected by technology, long-term memory was affected, indicating that the effect of technology-caused divided attention impacted long-term retention of material.
Taken together, the results of these studies strongly support that the usage of technology causes worse memory performance, due to the fact that this causes individuals to split their attention between the target content and using their device. To build on this, the study conducted by Stothart, Mitchum, and Yehnert (2015) investigated whether the auditory and/or tactile notifications generated by such devices could themselves be distracting to subjects. Specifically, the investigators sought to examine the situation in which subjects did not actively utilize their devices throughout the study period but rather merely received notifications, with the goal of determining whether such a condition would cause a similar distractive effect. Participants were split into three groups: the first group received calls during the experiment, the second received text messages, and the third received no notifications from the experimenters. The subjects were presented with numbers that were presented for 1 second and were instructed to press a key each time a number was presented, unless it was a certain nontarget lure. The results were analyzed with two measures of attention performance which had been previously found to be associated with task-unrelated thoughts (Cheyne, Solman, Carriere, & Smilek, 2009). Interestingly, the results showed no significant difference between the two kinds of distractors used in the experiment. In other words, participants showed a similar decrease in memory performance regardless of whether their devices rang or whether they merely received notifications. Therefore, the researchers concluded that even unattended device notifications significantly impair the performance of subjects on attention-demanding tasks.
Given these results which strongly indicate that the use of technological devices during study has an effect on memory performance, and particularly given the finding that even unattended notifications have a significant negative effect on attentional tasks, the current study was designed to extend prior research on the distracting role of technology and its negative impact on learning and memory tasks. Specifically, as the use of devices in general and the role of notifications in particular had been found to cause divided attention in subjects, this experiment was designed to examine the impact that notifications have on memory. The participants of this study were divided into two groups, wherein the experimental group received regular notifications during the study period while the control group received none. Both groups were separately presented with the same 40 mutually unrelated words and both subject groups were instructed to attempt to memorize the words. The experimental group received 3-digit number notifications throughout the presentation of the words and were instructed to record them. Both groups were then presented with identical word recognition tests that consisted of previously presented and novel words and participants were then asked to identify the words that were shown before.
On the basis of prior research, it was predicted that the memory performance of subjects who were subjected to the divided attention task would be significantly worse than that of the control group and that the subjects in the experimental condition would demonstrate greater false alarm rates and lower hit rates on the test than the subjects in the control conditions. This would be in accordance with the findings made by Stothart et al. (2015).
Participants
Participants were 20 undergraduate university students enrolled in a senior laboratory course in cognitive psychology at Wilfrid Laurier University. Participants were randomly assigned to one of two groups: the first group that received notifications during study or to the control group which did not receive notifications during study. Subjects were compensated for their participation in the study through engagement marks that were awarded toward their overall course grades.
Apparatus and Materials
The subjects were presented with a study list that consisted of 40 unrelated words. The words were selected from the MRC Psycholinguistic Database and were unrelated common nouns, 5-8 letters in length, that were rated from 300 to 500 on the concreteness and frequency scales that have a full range between 100 and 700. The total number of words in the word pool was 60. All stimuli were presented on a screen located at the front of the classroom through the use of the Google Docs web-based service and an overhead projector. The notifications were delivered to participants through the use of the web-based Slack client, with each participant signed into a separate account, with 12 Slack accounts being created for this purpose. The notifications were delivered through the classroom desktop computers, with one computer assigned to each participant in the notification condition. The numbers that were used for the notifications were three-digit numbers that were randomly generated using the Random.org service.
Procedure
The experiment was conducted in a computer laboratory classroom. Subjects were randomly assigned to either the notification or control condition; the students assigned to the second condition were instructed to leave the classroom and wait in the hallway. The remaining students were situated in seats that were distributed throughout the room, in front of computers that each had a browser open to a logged-in Slack session with the volume set to a level of 60%. Each participant was also provided with a physical response sheet for recording the numerical notifications that were delivered via Slack. The study list was projected onto a screen positioned at the front of the classroom.
The students were then presented with items from the study list at a rate of 2 seconds per item. Each word was presented in uppercase in a black Calibri font of size 60 centered on a plain white background. The subjects were instructed to try to remember each presented word and also to write down the 3-digit numbers that were delivered to them via Slack notifications. During the presentation, participants received notifications every 10 seconds, for a total of 8 times throughout its full duration. Following the conclusion of the presentation, there was a 1.5-minute retention interval between the study period and the test that was occupied with the distribution of the physical response sheets for the test and communicating instructions to participants. After this, subjects were presented with a 2-alternative forced-choice recognition test that consisted of 20 words that were previously presented and 20 novel words, for a total of 40 words. The test words were shown at a rate of 4 seconds per item. Each test word was presented in the same format as was used during the presentation (viz., in uppercase in a black Calibri font of size 60 centered on a plain white background). The participants were asked to indicate whether they believed that a given word had been shown previously in the study list by circling ‘Yes or ‘No’ on their test response sheets at the time of each presentation. The results were recorded by the participant physically and processed electronically by the researchers. These results were processed to indicate the individual accurate recall rate for the old words as well as the false alarm rate for the new words.
Following this, response sheets were collected by the researchers and the first group of students was sent out into the hallway while the second group was invited into the classroom. The latter group was presented with the same study list shown at an identical rate but did not receive notifications. Following the conclusion of the presentation, the same retention period was used, and participants were subsequently tested using an identical recognition test procedure as for the first group.
The main effect of recognition type was reliable, F(1,17) = 35.500, MSe = .035, p < .001. The hit rate was greater in the control (.68) than in the notification condition (.59), indicating that word recognition was significantly higher for the words that were presented in the study list rather than for the novel words in the test list. At the same time, the false alarm rate was found to be greater in the notification condition (.37) than in the control condition (.14). The main effect of group type was not found to be reliable, F(1,17) = 2.641, p = .123. This suggests that neither group answered with a ‘Yes’ response significantly more than the other. Finally, a significant interaction was found between the effects of recognition type and group type on word recognition, F(1,17) = 6.060, MSe = .035, p =.025. This indicated that there was a significant interaction between the effects of recognition type and group type on word recognition, such that correct word recognition was greater for the control than for the notification group. Moreover, the false alarm rate was found to be higher for the experimental group than for the control group, thereby demonstrating a mirror effect. Put another way, the group that was distracted by notifications during presentation had significantly more false alarms and made significantly fewer hits than the participants in the control condition.
In order to further interpret this interaction, Corrected Recognition Scores were calculated which were equivalent to the hit rate less the false alarm rate. The mean and standard deviation proportions of the Corrected Recognition Scores for each respective group are shown in Table S2 (available upon special request). An independent t-test was performed on the Corrected Recognition Scores and it was found that participants significantly differed in their recognition performance, t(17) = -2.462, p = .025. Specifically, participants in the notification condition demonstrated significantly worse performance at recognizing previously studied items, thus demonstrating that the divided attention effect that was attributable to the delivery of notifications in the learning context resulted in a diminished memory for the learned content.
The results of the experiment revealed a reliable main effect of recognition type, with a higher hit rate in the control than in the notification condition and with a higher false alarm rate in the notification than in the control condition. Additionally, there was no reliable main effect of group type, suggesting that neither group recognized old words significantly better than the other. A significant interaction effect between recognition type and group type was also found, which implied that the correct word recognition rate (hit rate) was higher in the control condition than in the notification condition, and also that the false alarm rate was greater in the notification than in the control condition, thusly demonstrating a mirror effect. Furthermore, an analysis of Corrected Recognition Scores showed that the subject groups differed significantly in their recognition performance, with participants in the notification condition demonstrating significantly worse performance at identifying the words that were previously presented. Therefore, this finding supported the hypothesis that divided attention caused by notifications during information encoding would lead to reduced memory performance for said information. This result was in line with that of the Glass and Kang (2018) study that indicated that testing performance was worse for the students who were allowed to use electronic devices during study. Similarly, the results of Hembrooke and Gay (2003) strongly suggested that the usage of electronics during lessons significantly reduced the memory of students for lecture information. Clearly, divided attention at time of encoding can be caused by the use of electronic devices in general and is particularly likely to be caused by the delivery of notifications, and this has a negative effect on the retention of target encoding material.
To recap, the present study was split into two groups – one that received notifications that they were instructed to attend to and record, and another that received no notifications whatsoever. Future research could further explore the distracting role of notifications through splitting the experimental group into three further sub-groups: wherein the first group would be instructed to note the content of notifications, the second asked to simply view the notifications (without recording them), while the third would merely receive an audio or tactile signal without actively interacting with the device whatsoever. The current body of evidence suggests that the first experimental sub-group would be the one that would experience the most significant reduction in memory performance, but it would be of interest to determine as to whether the other sub-groups would also display a significant reduction in memory retention, as well as the comparative strength of this effect in each group.
Broadbent’s theory of selective attention proposes that information passes through a limited processing channel after the sensory processing stage but prior to being encoded in the short-term memory store (Broadbent, 1958). Furthermore, it claims that when this channel is overloaded, such as in the case wherein subjects receive notifications at the same time as they are attempting to attend to a different cognitive task, some information is filtered out and therefore remains unprocessed, while other information is retained and is therefore subsequently further analyzed by higher-level brain processes. This theory is generally practiced in a particular paradigm wherein subjects perform some primary task while also being instructed to monitor a secondary task for information, a paradigm that was reflected in the present study. As has been found in numerous prior studies, the fact that the two tasks are being performed simultaneously leads to a reduction in primary task performance. Therefore, electronic devices and particularly the notifications that they tend to frequently display can be seen as effectively replicating this experimental paradigm in our everyday lives – making everything that we, as the proud owners of these devices, are trying to focus on into split-attention tasks. The notifications that we receive therefore become a secondary task that we feel forced to attend to while we try to work on the things that are actually meaningful to us. It is therefore a rather sobering, if not a particularly surprising conclusion that our devices and the notifications that they helpfully provide facilitate a significant detrimental effect on our primary task and memory performance. Small wonder then that we seem to be unable to focus on anything, to get anything productive done, to even truly and fully enjoy art, that we find our unremembered days blending into one another as our lives slip away from us. If that is the price to be paid for constant connection and being always in touch, perhaps the time is long overdue for us to take a step back and ask ourselves whether this sacrifice is worth it, before it is too late, before the next notification comes in and derails our brief moment of self-reflection. The Industrial Revolution and its consequences, as someone once memorably said....
REFERENCES
i Broadbent, D. (1958). The Selective Nature of Learning. Perception and Communication, 244-267. doi:10.1016/b978-1-4832-0079-8.50012-8
ii Cheyne, J. A., Solman, G. J., Carriere, J. S., & Smilek, D. (2009). Anatomy of an error: A bidirectional state model of task engagement/disengagement and attention-related errors. Cognition, 111(1), 98-113. doi:10.1016/j.cognition.2008.12.009
iii Glass, A. L., & Kang, M. (2018). Dividing attention in the classroom reduces exam performance. Educational Psychology, 1-14. doi:10.1080/01443410.2018.1489046
iv Hembrooke, H., & Gay, G. (2003). The laptop and the lecture: The effects of multitasking in learning environments. Journal of Computing in Higher Education, 15(1), 46-64. doi:10.1007/bf02940852
v Marty-Dugas, J., Ralph, B. C., Oakman, J. M., & Smilek, D. (2018). The relation between smartphone use and everyday inattention. Psychology of Consciousness: Theory, Research, and Practice, 5(1), 46-62. doi:10.1037/cns0000131
vi National Safety Council. (2015, May 18). Cell phones are involved in an estimated 27 percent of all car crashes, says National Safety Council. PR Newswire. Retrieved from https://www.prnewswire.com/news-releases/cell-phones-are-involved-in-an-estimated-27-percent-of-all-car-crashes-says-national-safety-council-300084920.html
vii Statistics Canada. (2017). Life in the fast lane: How are Canadians managing?, 2016. Ottawa, ON: Statistics Canada.
viii Stothart, C., Mitchum, A., & Yehnert, C. (2015). The attentional cost of receiving a cell phone notification. Journal of Experimental Psychology: Human Perception and Performance, 41(4), 893-897. doi:10.1037/xhp0000100
ix Strayer, D. L., Drews, F. A., & Crouch, D. J. (2006). A Comparison of the Cell Phone Driver and the Drunk Driver. Human Factors: The Journal of the Human Factors and Ergonomics Society, 48(2), 381-391. doi:10.1518/001872006777724471