A majority of American adults who use the Internet have social media accounts (e.g., Facebook, Twitter (Greenwood, Perrin, & Duggan, 2016)). Although younger adults are more likely to use social media, use among older adults is increasing (Zickuhr & Madden, 2012). Social media sites are becoming a significant source of news (Hermida, Fletcher, Korell, & Logan, 2012), with 85% of posts on Twitter being headline news or news-like (Kwak, Lee, Park, & Moon, 2010). Thus, how we access the news is evolving, such that traditional news sources, like newspapers, cable, and radio, are becoming less discernible from social media. With the blending of traditional and social media, it is important to assess the extent to which one can accurately recall both the content and the source of information encountered in these media and whether the age of the user matters.
In the present study, younger and older participants studied social media posts (i.e., tweets) and news headlines selected from CNN online that were formatted to appear as items on a Twitter feed or news items on CNN. Item and source memory tests were administered to examine how content and perceived source were remembered. Given the blending of news and social media platforms and the ease with which information can be shared across platforms (e.g., linking a news story on Facebook or Twitter) it is worthwhile to investigate memory differences for social media and news across ages. In this study, we were particularly interested in situations when the content does not match the format of the platform (e.g., a news item processed on social media) and how source memory may be affected if readers are accessing news across multiple sites.
Source memory involves the recollection of details and context surrounding the encoding of an event (e.g., Johnson, Hashtroudi, & Lindsay, 1993). The ability to accurately recall where one learned something is fundamental for evaluating its veracity, or for finding the information at a later date. Relative to item memory, which can be largely familiarity-driven, source memory involves more controlled and recollective processes because it requires binding episodic details. The integration of news on social media (Kwak et al., 2010) might result in source confusion. For example, people misidentified televised news-like advertisements as actual news about 70% of the time (Yegiyan & Grabe, 2007). Thus, people reading news on social media might fail to remember that news-like information presented on social media websites was not actually news.
Factors such as a person’s age and their familiarity with the content can affect the accuracy of source memory. Because of older adults’ deficits in source and associative memory (Old & Naveh-Benjamin, 2008) and their increasing rate of interactions with social media, the proliferation of news items on social media may negatively affect their memory. On episodic memory tests, older adults frequently show deficits in memory relative to younger adults, most noticeably when the tests require use of controlled processes (Balota, Dolan, & Duchek, 2000; Craik & Byrd, 1982; Ferguson, Hashtroudi, & Johnson, 1992). Younger and older adults’ item memory can be similar, especially when the task offers environmental support (e.g., recognition tests), but older adults perform worse in source memory tests (Balota et al., 2000; McIntyre & Craik, 1987; Spencer & Raz, 1995). These deficits in source memory may be attributed to a lack of detail-specific encoding. Johnson et al. (1993) hypothesized older adults were less likely to encode specific contextual details about the to-be-remembered items. Rather, they encode in a more general, automatic manner resulting in poorer retrieval cues and overall impairment in source memory (Rabinowitz, Craik, & Ackerman, 1982). In some cases, however, older adults had better source memory for conceptual details (i.e., whether the speaker was lying or telling the truth) than perceptual details (i.e., the gender of the speaker; Rahhal, May, & Hasher, 2002). According to socio-emotional selectivity theory (Carstensen, 1995), how individuals handle socially relevant interactions is adaptive and dependent on life stages. Older adults in particular might shift their attention away from knowledge-gathering and focus more on processes such as emotion regulation that enhance wellbeing. This shift in focus can then facilitate performance on tasks that emphasize emotional or pro-social processing. For example, older adults tend to remember emotionally and personally relevant information relatively well, and may even compensate for deficits in source memory when given socio-emotional information that is consistent with their current goals. In one study (Rahhal et al., 2002), older adults remembered the source of information as well as younger adults when they were given information about the speakers (e.g., speaker A lies, speaker B tells the truth), suggesting that the socio-emotional information provided allowed them to encode the information better.
Even when the material is not personally relevant or emotional in content, familiarity in some domains may help older adults compensate for deficits in processing resources (Park, 1997). Specifically, older adults have an elaborate semantic memory, that often exceeds that of younger adults (e.g., Salthouse, 2004) and can enhance episodic retrieval. If older adults are able to rely on prior knowledge or capitalize on their experiences, age differences can be reduced or eliminated (see Umanath & Marsh, 2014 for a review). In one study (Matzen & Benjamin, 2013), older adults outperformed younger adults in memory for sentences, a finding the authors attributed to the added experience in reading acquired over a lifetime.
In addition to increased language knowledge and experience, older adults also tend to rely more than younger adults on schemas and schematic knowledge (i.e., stored knowledge that organizes information and specifies the relations between elements). Stereotypes are one form of schematic knowledge. For example, older adults have been shown to be more sensitive to stereotype-consistent information than younger adults, such that the two age groups were equally accurate at attributing stereotype consistent information to the proper source, but older adults were less accurate on stereotype inconsistent content (Mather, Johnson, & De Leonardis, 1999). Even when learning new information, older adults show sensitivity to the compatibility between elements of the stimuli. Specifically, compatible pairings of brand logo graphics and brand names (e.g., pairs that were inherently related) improved associative memory in both younger and older adults and, importantly, eliminated age-related deficits relative to incompatible pairings (Mohanty, Naveh-Benjamin, & Ratneshwar, 2016).
Thus, although older adults frequently do exhibit deficits in source memory, there are circumstances that mitigate these deficits. Specifically, relatedness, prior knowledge, and schematic knowledge can moderate age-related differences in performance. Not surprisingly, younger adults also frequently benefit from these factors as well, indicating that both age groups should benefit when to-be-remembered material is compatible with its source. In the present context, older adults presumably have more experience than their younger counterparts with traditional news media, such as newspaper, radios, and television. Although older adults are increasingly accessing online sources, they still tend to prefer to follow news in print format (Mitchell, 2016). Importantly, their familiarity with sources such as CNN is likely to be quite high. However, social media platforms, such as Twitter, may not be as familiar to older adults. Therefore, they may not be able to capitalize on prior knowledge and integrate this information into existing systems.
The congruency of content and source may also affect older adults’ recollection. Because of older adults’ increased knowledge base and tendency to rely on schematic or semantic knowledge, source memory might be disrupted if the format (which indicates source) does not align with the content (e.g., if a news headline appeared as a Twitter post). If older adults have a richer knowledge of what items are “news” items, seeing these items in the context of a tweet might be particularly disruptive. In other words, when the source does not align with the content, it may become more difficult for older adults to identify the correct source, as they can no longer rely on the integrative nature of a congruent manipulation (e.g., a tweet that appears as a Twitter post).
Credibility is another factor that may impact source memory. Older adults who rate social media posts as less credible or reliable may not devote the processing resources required during the encoding phase to compensate for the age-related deficits. Mutter, Lindsey, and Pliske (1995) examined the effect of repetition on truth judgments (in which repeated statements are rated as more true than non-repeated statements, regardless of statement veracity) in younger and older adults. Older adults were less accurate at recognizing the source of not credible items than the source of credible items. More generally, given the link between attention and memory (Mulligan, 2008), people might have poor memory for social media because it may be perceived as unimportant and they thus devote fewer attentional resources to this content.
Nevertheless, Mickes et al. (2013) found that younger adults had a robust memory for social media content. In their study, college-aged participants were presented with faces, book sentences, or Facebook posts. Participants remembered Facebook posts better than faces and excerpts from books, a finding Mickes et al. attributed to the personal, socially relevant nature of the content. In another experiment, headlines, sentences, and comments were selected from the Entertainment or Breaking News sections of CNN Twitter feeds. People recognized ttems from CNN’s entertainment section better than those from CNN’s news section, suggesting that the social quality of the information - be it the gossipy nature of entertainment news or status updates on Facebook - contributed to memorability. Thus, Mickes et al. concluded that content from social media was remembered better than other types of content, a fact they attributed to the gossipy nature of the information and to its reliance on natural language.
However, Mickes et al. (2013) did not examine the impact of knowing the source of the items. In their study, all items were presented without any source identifiers (i.e., as typed sentences). Including the source of the information - by presenting a social media post in its original format, with a user profile picture, and date and time information - might affect the item’s memorability. Raj and Bell (2010) suggest that binding the contextual information and content enhances the recollection of complex memories. Inclusion of additional formatting information might also increase the level of detail or distinctiveness (Hunt & McDaniel, 1993), thereby improving memory. Although older adults retained less source information than younger adults in a study examining memory for news sources (i.e., print, radio, and TV (Frieske & Park, 1999)), both groups performed better on source tests for richer content, suggesting that the additional detail provided by the formatting might increase performance. Alternatively, familiarity and source credibility may negatively affect the extent to which material is remembered. For example, Lucassen and Schraagen (2013) found that when college-aged students were presented familiar versus unfamiliar material in either a Wikipedia or non-Wikipedia format (i.e., source information was not provided), familiar material was considered less credible when presented in the Wikipedia format, although there was no effect for unfamiliar material. Therefore, if an item is recognized as coming from a social media or user-generated source, it might be considered of less importance or less reliable and thus not be processed as deeply.
In the present study, we extended Mickes et al.’s (2013) findings by including an older adult sample and by examining source memory. The aging sample allowed us to address questions such as the importance of prior experience with social media, the role of controlled processes (Craik & Byrd, 1982), and the role of schematic/semantic support (Umanath & Marsh, 2014). Participants studied items from Twitter and CNN that were digitally altered to appear as tweets or as headlines. We assessed item and source memory for congruent (e.g., tweets formatted as Twitter posts) and incongruent (e.g., tweets formatted as CNN headlines) items. This design allowed us to examine whether the appearance or format of to-be-remembered information, which indicates its source (i.e., Twitter or CNN), contributes to memory above and beyond content and whether congruency between content and format further affects performance.
If the content, specifically the socially relevant nature of social media, is driving memory, tweets should be remembered better than headlines, regardless of format, consistent with Mickes et al. (2013). According to predictions from socio-emotional selectivity theory (Carstensen, 1995), older adults might also show a mnemonic advantage for tweets because of the social content provided by these items. Alternatively, if the social media advantage is driven by content familiarity, older adults might fail to show a mnemonic advantage, due to reduced schematic or semantic support for the content (Umanath & Marsh, 2014). Thus, whereas we expected to replicate the social media advantage in younger adults, predictions about older adults’ performance were less clear. However, if social media sources are deemed as less credible (Lucassen & Schraagen, 2013) and thus devoted fewer attentional resources during encoding (Mutter et al., 1995), then inclusion of the formatting information might eliminate the mnemonic advantage for social media content in both age groups.
Regarding source identification processes, due to older adults’ deficits in source memory coupled with lower familiarity with the platform and possibly the content of the tweets, we expected younger adults to identify the source accurately more often than older adults (Balota et al., 2000), primarily when the stimuli were incongruent. Congruent stimuli were expected to enhance source memory for both younger and older adults, as suggested by prior work showing improved source memory for compatible over incompatible information in both age groups (Mohanty et al., 2016). Given the increased reliance in older adults on factors such as compatibility and relatedness, compared to younger adults, we expected the older participants to be especially challenged when remembering the source of incongruent items.