Major memory for microblogs
Online social networking is vastly popular and permits its members to post their thoughts as microblogs, an opportunity that people exploit, on Facebook alone, over 30 million times an hour. Such trivial ephemera, one might think, should vanish quickly from memory; conversely, they may comprise the sort of information that our memories are tuned to recognize, if that which we readily generate, we also readily store. In the first two experiments, participants’ memory for Facebook posts was found to be strikingly stronger than their memory for human faces or sentences from books—a magnitude comparable to the difference in memory strength between amnesics and healthy controls. The second experiment suggested that this difference is not due to Facebook posts spontaneously generating social elaboration, because memory for posts is enhanced as much by adding social elaboration as is memory for book sentences. Our final experiment, using headlines, sentences, and reader comments from articles, suggested that the remarkable memory for microblogs is also not due to their completeness or simply their topic, but may be a more general phenomenon of their being the largely spontaneous and natural emanations of the human mind.
KeywordsMemoryMicroblogFacebookLevels of processing
The online world permits its citizens to post their every thought as blogs, Facebook status updates, forum comments, and the like. Not every such posting is extraordinarily interesting; indeed, Facebook posts have been described as “navel-gazing . . . spam” (Griggs, 2009). However mundane, or even inane, such sharing may be, it is also enormously popular. On Facebook alone, people collectively post over thirty million times an hour (Facebook.com, 2011). This popularity suggests that something about such “microblogging” resonates with human nature. We explored this question in a series of experiments examining the memorability of microblog posts. If microblogs are as vacuous as some propose, they should vanish quickly from one’s mind. On the other hand, as such outputs represent the natural, essentially unfiltered, emanations of human minds, perhaps they also have a special place there as input. The success of Facebook, after all, depends not just on people’s willingness to post their thoughts, but also on their willingness to read those posts.
Particular sorts of information may have a privileged place in memory. Nairne and Pandeirada (2010) suggested that our memory systems developed to solve adaptive problems and that we should see enhanced memory in tasks that tap into the types of issues confronted in our ancestral past. Nairne and colleagues have shown, for example, better memory for information that would have been related to survival needs (Nairne, Pandeirada, & Thompson, 2008; Nairne, Thompson, & Pandeirada, 2007; see also Kang, McDermott, & Cohen, 2008), and social information could likewise be privileged because of our ancestral reliance on group living. Indeed, recently Klein, Cosmides, Gangi, Jackson, and Tooby (2009) described memory as “an adaptive tool for sociality” (p. 284). The research on social memory thus far has focused on autobiographical memory (e.g., birthdays; Merz, Wolf, & Hennig, 2010) or face–name pairs (e.g., Takahashi et al., 2004). The popularity of microblogging affords a way to investigate memory for the daily details of social information.
Baumeister and Leary (1995) proposed that social belongingness is so essential in humans that it should be considered a basic, innate motive. Humans are driven to share their experiences with others—an inclination that emerges early in development, with preverbal infants’ use of pointing and eye gaze in social interactions—and this feature seems to separate us from even our ape cousins. This need to express and share our experiences has been variously documented (see Emery, 2000). Such views support the popularity of Facebook posts and other microblogs and suggest that, if receiving these communications is as much a part of our nature as making them, these reports could be particularly memorable.
In Experiment 1, we examined the strength of memory for Facebook status updates (i.e., messages posted to friends). We compared the memorability of such microblogs to that for sentences from published books (Exp. 1a) and to the memorability of human faces (Exp. 1b).
Thirty-two University of California, San Diego (UCSD) undergraduates (age: M = 21.00 years, SD = 2.14; 27 female, five male) were randomly assigned to one of two conditions, with half in the Facebook condition and half in the book condition.
Five undergraduate research assistants, who were blind to the hypothesis, gathered 200 posts written by others (from, collectively, over 3,000 friends) from their own Facebook feeds. Each post was the most current for its author, and none was connected with games, links, or photos (see the Appendix for examples).
A total of 200 sentences were also selected from books on amazon.com, using the “Last 30 days” option under the “New Releases” section, including both fiction and nonfiction. For each of these books for which the “Look Inside” feature was available, we used the “Surprise Me!” option, and within that random page, a single sentence was randomly selected. Sentences with quotations, single-word sentences, and sentences that contained more than 25 words were not selected (see the Appendix for examples).
Of the 200 stimuli in each condition, 100 were randomly selected as targets for each participant to memorize during the study phase. The other 100 were reserved as lures for the testing phase, to assess whether the participants could identify stimuli they had seen before: (a different combination of targets and lures per participant). Instructions and stimuli were displayed via E-Prime (Psychology Software Tools Inc., Sharpsburg, PA; www.pstnet.com) on a 22-in. monitor in 18-point Courier font.
Participants in the Facebook condition responded generally with the highest levels of confidence and were highly accurate. The average d' (an unbiased measure of memory strength) in the Facebook condition (M = 2.48, SD = 0.69) was significantly greater than the average d' in the book sentence condition (M = 1.72, SD = 0.54), t(30) = 3.47, p = .002. The percent correct measure (calculated as the number of hits plus correct rejections, divided by 200 trials), while potentially biased (Macmillan & Creelman, 2005), is intuitively appealing, and it reveals exactly the same story: significantly greater memory for Facebook posts (M = 85 %, SD = 7.2) than for sentences from books (M = 76 %, SD = 6.2), t(30) = 3.72, p = .001.
Before investigating deeper conceptual explanations for the advantage that Facebook posts seem to enjoy, we ruled out various more superficial possibilities. First, we tested whether the length of the post or sentence was a critical factor. The Facebook posts (M = 11.49, SD = 6.54) had significantly, albeit slightly, more words than the book sentences (M = 10.27, SD = 4.00), t(199) = 6.14, p < .001. To adjust for this, we did a median split on the numbers of words for the Facebook posts and book sentences (median = 10.00 for both), and then compared the d' scores for the shorter Facebook posts and the longer sentences. The average d' for the Facebook posts (M = 2.67, SD = 0.98), was still significantly higher than that for the book sentences (M = 1.77, SD = 0.92), t(102) = 6.86, p < .001, even with the length confound reversed, suggesting that the Facebook posts’ advantage was not due to length.1
Another possibility is that Facebook posts may capitalize on perceptual matching of surface-level differences (e.g., Mandler, 1979), since they are littered with irregular typography. To test this, we separated posts that contained emoticons, multiple exclamation points, all letters capitalized, or multiple, repeated letters (e.g., “Del Mar Opening Day is on my birthday this year!!! :) Hellooo HATS”) from posts that did not. A total of 86 posts did not include any of these components. Memory was still significantly higher for these orthographically regular posts (d': M = 2.57, SD = 1.18) than for the book sentences (d': M = 1.95, SD = 1.07), t(284) = 4.32, p < .001.
Memory was substantially higher for Facebook posts than for book sentences. Explanations for this advantage that were based merely on surface-level differences were ruled out: The difference was not due to posts containing emoticons, unique characters, or many or few words; the advantage persisted when all such posts were removed. The posts culled from Facebook showed remarkable memorability; memory for book sentences, on the other hand, reflected more typical memory performance. For example, the memory for sentences found here seems similar to results reported by Belmore (1982) using similar recognition tests for sentences with roughly the same number of targets and lures as in our experiment.
Clearly participants recognized Facebook posts better than ordinary published sentences. To get a further sense of the memorability of such posts, we next compared them to memory for faces. A region of the brain, the fusiform face area, is dedicated to face processing (e.g., Kanwisher, McDermott, & Chun, 1997), suggesting that the brain is specially designed to process and store facial information. While many factors can influence the memorability of a set of faces, faces nonetheless can provide some calibration for the magnitude of Facebook’s memorability, measuring whether memory for Facebook posts is particularly strong or memory for sentences from books is particularly weak. In Experiment 1b, accordingly, participants completed a memory task for faces and for Facebook posts.
Sixteen UCSD undergraduates (age: M = 20.25 years, SD = 1.34; 11 female, five male) participated for course credit.
Four undergraduate assistants used their own Facebook accounts to find a new sample of 200 posts by others2 (see the Appendix). Using a new set of Facebook posts allowed an independent replication of their memorability. For the face memory task, 200 neutral faces (frontal view only) were selected from the Color FERET database (http://face.nist.gov/colorferet/), with 100 of each stimulus type randomly chosen to be targets and 100 to be lures (a different combination of targets and lures per participant).
The procedure was identical to that of Experiment 1a, with two changes: Faces were used instead of book sentences, and a within-subjects rather than between-subjects design was used. Task order was counterbalanced; half of the participants completed the face memory task first, and half completed the Facebook memory task first. Again, the presentation time for each target was 3,000 ms, with a 250-ms ISI. Immediately after the study phase, participants took a self-paced recognition test that consisted of 100 targets randomly intermixed with 100 lures, shown one at a time, and the participants provided old/new responses on a 20-point rating scale.
Experiment 1b further strengthened the idea that memory for these microblog posts is remarkable, with Facebook beating both faces and books. As a frame of reference, across multiple memory tests for lists of words (Manns, Hopkins, Reed, Kitchener, & Squire, 2003), hippocampally damaged amnesic individuals showed an average d' of 0.4, and controls 1.5, a difference of 1.1. Averaging across our experiments, books and faces had a d' of 1.3, and Facebook a d' of 2.5, a difference of 1.2. Thus, Facebook’s advantage over books and faces is on the same scale as the advantage of controls over amnesics.
While several possible mechanisms for the memory strength of Facebook posts have been ruled out, a number of explanations remain. The posts may naturally elicit social thinking and lead to stronger encoding of the posts, whereas sentences written by professional authors and unknown neutral faces may be less likely to naturally elicit such encoding-enhancing elaboration. The more gossipy nature of the Facebook posts may have been inherently more interesting to participants and contributed to this social elaboration (Schiefele & Krapp, 1996). In addition, the Facebook posts may be particularly memorable because they are complete in and of themselves, whereas the book sentences, chosen randomly and out of context, need not have been so.
Another possible explanation is that the relatively unfiltered and spontaneous production of one person’s mind is just the sort of thing that is readily stored in another’s mind. The formality and complexity of well-considered and edited language, while it may have many advantages in accuracy, efficiency, and even beauty, may not be more readily stored. That is, information that people generate easily and naturally may be information that is easily and naturally remembered. Such an explanation is not entirely distinct from the gossip notion, since reports of the personal doings of other people may be closer to our natural speech than are more arcane and abstract reports. Two follow-up experiments explored these explanations for the superiority of Facebook posts. In the first, we examined whether spontaneous social elaboration is the secret to Facebook’s success, at least in the memorability of its posts, by testing whether directed social elaboration impacts memory less for Facebook posts than for other sentences. The second was designed to compare news sentences to news headlines and to readers’ comments to explore the roles of content and completeness in memorability.
Facebook posts may naturally elicit social thinking (e.g., “that is something my friend Emily would post”), while published sentences from books may be much less likely to elicit such elaboration. We tested whether such elaboration underlies the Facebook effect by manipulating the level of processing in an incidental learning task (e.g., Craik & Lockhart, 1972; Nairne et al., 2008). A shallow-encoding condition required participants to count the words of each sentence or post. A deep-encoding condition required participants to think of someone whom they knew (personally or a fictional character) who could have composed each post or sentence. If people spontaneously elaborate Facebook posts but not book sentences, the directions to elaborate should have little memory-enhancing effect on the former and a more profound effect on the latter. If, on the other hand, the Facebook advantage is not due to social elaboration, then adding it to each should have essentially equal effects.
Sixty-four UCSD undergraduates participated for course credit (age: M = 20.34 years, SD = 2.12; 41 female, 23 male). The participants were randomly assigned to Facebook posts or book sentences, as well as to either shallow or deep encoding.
The Facebook posts and book sentences were nearly the same as in Experiment 1a, but to control for length, we used a subset of the book sentences and added more, chosen in the same way (to return the total number to 200 sentences), so that the Facebook posts (M = 11.64, SD = 6.77) would not have more words than the book sentences (M = 11.21, SD = 4.93), t(398) = 0.72, p = .47.
The experiment employed an incidental-learning task with a 2 × 2 design (shallow vs. deep processing, and book sentences vs. Facebook posts). Participants in the shallow-encoding condition counted and entered the number of words per Facebook post or book sentence. In the deep-encoding condition, participants rated (on a 5-point scale, from difficult to easy) the ease with which a Facebook post or book sentence reminded them of something that someone they know would write or say (e.g., themselves, a friend, a family member, or a character in a movie or book). The instructions were adapted from Nairne et al. (2008). After a 10-min distractor task (a game of Space Invaders), the participants took a surprise recognition memory test on the 100 targets randomly intermixed with 100 lures.
In the deep-encoding groups, the average ratings of the ease with which participants could think of a person who would write a sentence (M = 3.30, SD = 0.42) or post (M = 3.20, SD = 0.64) were not significantly different, t(15) = 0.54, p = .595. In the shallow-encoding groups, the numbers of words counted were not significantly different for book sentences (M = 11.24, SD = 0.42) and Facebook posts (M = 11.52, SD = 0.75); t(30) = 1.33, p = .194
If social elaboration were responsible for the memorability of Facebook posts, one would expect that using socially based deep encoding would equalize the d's for book sentences and Facebook posts. However, the memory benefits of Facebook posts and of social encoding were additive, suggesting that the memorability of Facebook posts in our experiment was not simply due to their engendering natural social elaboration during encoding. Furthermore, the ratings of ease of social elaboration did not suggest that the Facebook posts enjoyed any obvious natural advantage there. In the next experiment, we explored whether an item’s content or completeness of ideas might account for the Facebook effect.
To test the hypothesis that Facebook posts might be advantaged because they are coherent and complete ideas, we compared the memorability of sentences drawn from CNN articles to the memorability of the headlines of those articles, which are written to stand alone. To test whether the Facebook advantage might be due to their gossipy nature (e.g., general sharing of details about aspects of others’ lives; Dunbar, 2004), we selected articles drawn either from news or from entertainment. Finally, to test the notion that the unfiltered, largely unconsidered postings of strangers are what are especially memorable, we also compared reader comments drawn from the comment sections at the end of each news or entertainment article to both the headlines and article sentences.
One-hundred-eighty UCSD undergraduates participated in exchange for course credit (age: M = 20.12 years, SD = 2.50; 138 female, 42 male).
The stimuli consisted of text drawn from two categories: CNN’s Breaking News and Entertainment News sections. From each of those categories, we drew three subcategories of stimuli: headlines, sentences, and comments. All were gathered from CNN Twitter feeds (http://twitter.com/CNNBrk and http://twitter.com/CNNshowbiz, respectively) dating from August 12, 2011, through September 15, 2011. For each story, the tweets linked to the online article, from which we copied the headline, a sentence chosen at random from the body of the article, and a randomly chosen reader comment (in the comments section at the bottom of each article). The excluded stimuli were single-word items, items more than 25 words long, articles that contained no reader comments, and comments on comments. We gathered 200 stimuli of each of the six types. The stimuli were presented, and responses recorded, with the E-Prime software.
The participants were instructed that they would be memorizing stimuli gathered from many sources and would be given a memory test afterward. One hundred stimuli were chosen as targets, with 16 or 17 of each type randomly selected for each participant. No participant saw more than one item (headline, sentence, or comment) from a given article. One hundred lures were chosen in the same way. All other procedures were the same as in Experiment 1. Due to an experimental error, some duplicate stimuli were included. All duplicates were removed prior to analysis, resulting in the loss of two targets or lures for each of 43 subjects, and four for each of six subjects.
We also conducted analyses to determine whether sentence length could account for the memorability of the comments and entertainment news. A 2 (category) × 3 (text type) between-group ANOVA was performed on the word counts for all stimuli. The interaction was not significant, F(2, 1182) = 0.57, p > .05. However, we did find a significant main effect for category, with higher word counts for breaking news (M = 11.97, SD = 5.07) than for entertainment news (M = 11.06, SD = 4.84), F(1, 1182) = 13.14, p < .001, as well as a significant main effect for text type (headline, M = 7.79, SD = 1.93, vs. sentence, M = 14.51, SD = 4.48, vs. comment, M = 12.17, SD = 5.21), F(2, 1182) = 270.26, p < .001, with headlines having the lowest word count scores overall, and sentences from articles having the highest. In other words, word length is unlikely to be an explanation for the memorability differences, because the best-remembered stimuli were intermediate in length.
Average proportions (with standard deviations) of low-frequency words per text type in Experiment 3
Both News Types
All text types
The comments, similar to the Facebook posts in the earlier experiments, were remembered exceptionally well, especially relative to the breaking news headlines and sentences from the breaking news and entertainment articles. Their completeness of ideas (as also found in headlines) could contribute to, but not fully explain, the memorability of microblogs. Likewise, the more gossipy nature of Facebook posts (as also captured in entertainment news) could contribute to, but also not fully account for, the memorability of microblogs. It seems likely that both of these characteristics partially contribute to the memorability of microblogs. That is, the natural emanations of the human mind (as expressed in microblogs) are gossipy and have an element of completeness to them.
In this article, we introduced and investigated a new phenomenon—the incredible memorability of microblogs—and explored several potential mechanisms for this effect. The results from the first experiment suggested that memory for Facebook posts is remarkably strong, significantly stronger than that observed either for sentences drawn from published books or for faces. The second experiment showed that this effect was not due to the Facebook posts naturally producing deep social encoding; when people were asked to do just such deep social encoding, it enhanced memorability just as much for the Facebook posts as for the book sentences. If the advantage of the posts were due to people, or at least some reasonable fraction of them, already doing such social encoding, telling them to do so would have had a relatively small effect, not the large effect that was demonstrated. In the final experiment, we tested three factors that could contribute to the superiority of the posts and found some evidence for each. Text that is designed to be complete is indeed remembered better, as was shown by the superiority, in memorability, of headlines over sentences drawn from the bodies of CNN articles. More gossipy text also seems advantaged, with text drawn from entertainment news beating text drawn from breaking news. Finally, it also seems that sentences written casually by lay people, without professional, or perhaps any, editing, are especially readily remembered, as evidenced by the memorability of the comments posted at the ends of entertainment and news articles.
Related to this idea that casually generated language may be remembered better, Keenan, MacWhinney, and Mayhew (1977) examined memory for conversational sentences uttered by people who participants either did or did not know, and that had high or low interactive content (i.e., sentences whose meaning relied, or did not rely, on knowledge of the specific speaker’s intentions with regard to the particular listener), using taped lunchroom conversations. The authors concluded that conversational sentences were remembered better when people knew the speaker and when the sentences were high in interactional content. The memory strength of the Facebook posts and article comments is not exactly parallel, as participants were unlikely to know any of the authors, and so could not tie the utterances to other details of their originators. However, both the posts and comments—and, to some extent, the entertainment stories and headlines—do have a conversational, spontaneous tone (microblogging is, after all, part way to virtual chatting), and thus may not be entirely unrelated to hearing personal details from people whom one knows. Further research would be needed to determine whether this phenomenon applies to all social media (e.g., tweets or text messages), and even whether it applies to notes that one writes oneself, such as diary entries.
Many may consider the Facebook postings and article comments to be vacuous, narcissistic, or vapid, but they are thoughts that their writers considered worth sharing. A philosophical treatise by Immanuel Kant may be more profound, and more edifying to remember, than the average Facebook post or article comment, but his writings may not be tuned so precisely to what our minds effortlessly encode. These especially memorable Facebook posts and reader comments, generated by ordinary people, may be far closer than professionally crafted sentences to tapping into the basic language capacities of our minds. Perhaps the very sentences that are so effortlessly generated are, for that reason, the same ones that are readily remembered. Some sentences—and, most likely, those without careful editing, polishing, and perfecting—are naturally more “mind-ready.” The advantage of entertainment over news, as well as the advantages of Facebook posts over book sentences and of comments over the headlines and text of articles, may reflect such a status. Perhaps these effortlessly occurring proclamations help foster social belongingness that may extend to online communities. As a result, this type of shared information does have a privileged status and is remembered more readily (Nairne & Pandeirada, 2010).
It seems that, with the growth of blogging, text messaging, and the like, written language has moved closer to natural speech, with less editing and contemplation than was needed not only when the writing was done by monks with goose-feather quills or by Gutenberg with moveable type, but even when it is done by authors sitting patiently at their own keyboards. In Japan, several top-selling novels were written in the form of a series of cellphone text messages (www.nytimes.com/2008/01/20/world/asia/20japan.html). In television, likewise, “reality” programming has been on the increase, moving away from careful scripting and plotting to something close to what humans naturally put out, and so perhaps just as naturally take in.
In education, spoken lectures have long been viewed as a useful adjunct to written texts, even when the latter have been more polished, thoughtful, and complete. Consistent with our findings, the nature of the communication may make the information more memorable, with the professor generating natural speech. Perhaps, though, textbooks written as tweets would render the faculty obsolete.
The difference in d's between short (M = 2.35, SD = 1.41) and long (M = 2.67, SD = 0.98) Facebook posts was marginally significant, t(198) = 1.83, p = .057. The same analysis yielded opposite results for the book sentences: Memory was significantly better, as measured by d', for short book sentences (M = 2.15, SD = 1.18) than for long book sentences (M = 1.77, SD = 0.92), t(198) = 2.60, p = .010. Neither effect, of course, can account for the superior memory for Facebook posts relative to book sentences.
We thank J. T. Wixted for his comments on this work, as well as Travis Seale-Carlisle, Kim Huynh, Johanna McElfresh, Michelle Niku, Katelyn Steele, Lisa Teachanarong, Brenda Wu, and Allison Yee for data collection.