Was that how it happened? Shaping our memory for personal experiences in conversation with others

In his 1932 book 'Remembering: A study in experimental and social psychology', Bartlett also emphasized the social nature of memory (or remembering). Subsequent research has provided support for his assertion, showing that people often discuss and form memories in the presence of others (Hirst & Echterhoff, 2012), regardless of whether those events are mundane (e.g., everyday events recorded in a diary; Pasupathi, McLean, & Weeks, 2009) or more noteworthy (e.g., criminal events witnessed in person; Paterson & Kemp, 2006; Skagerberg & Wright, 2008). In sum, most remembering is a schema-driven process of reconstruction that often occurs in a social context. As a result, there are many reasons why, over the course of a decade, Brian Williams’ account of his experience over the skies in Iraq might have changed. Here, we will address several explanations based on findings from the literature on conversational remembering.


Considering context and audience: Williams was telling his tale to entertain, not to inform

In the course of a conversation, speakers tend to focus more on evaluative and affective information, preferring to share opinions and feelings about what transpired, rather than recounting events serially (Hyman, 1994). Thus, conversational retellings do not typically involve detailed recollections presented in a stable chronological order (Marsh, 2007). In one study, researchers recorded new parents’ telephone conversations with friends and family regarding the recent birth of a child. Parents often initially discussed topics of greatest interest to the listener, such as the gender of the baby, and its general health – narrative descriptions of events as they unfolded were of secondary concern (Tenney, 1989).

With respect to Williams’ altered recollection of his experience, it is important to note that he was on a show that emphasizes entertainment value. Williams was not behind his news desk. He was telling his tale to entertain, not to inform, and catered to an audience that demanded action and excitement—best evoked when a tale is told from a first-person perspective. Research confirms that people often exaggerate for the purpose of entertaining. For example, Dudukovic, Marsh, and Tversky (2004) had participants read a story and re-tell it either accurately or for the purpose of entertainment. A separate scorer, unaware of each participant’s experimental condition, rated the retellings for accuracy and level of entertainment. Retellings that were rated as highly entertaining tended to be less accurate. Entertaining retellings were told in the language of story-telling—that is, they were told with confidence, and usually in the present tense, with increased exaggerations, omissions and some added details. When retelling events from their own lives, Marsh and Tversky (2004) found that 61% of participants admitted distorting the event – most commonly by omitting certain details and embellishing others. Participants communicated some events repeatedly; of these, 76% were altered across retellings. Participants attributed alterations to the fact that they were recounting the events for different audiences.

Why might the audience of a retelling exert such an effect? According to the principle of co-construction, recollections of memories in conversation are the product of both the speaker and their social environment (Pasupathi, 2001). Attentive and engaged listeners tend to elicit more detailed accounts from speakers, facilitating speaker’s long-term memory for an event (Pasupathi, Stallworth, & Murdoch, 1998). Moreover, speakers alter (or ‘tune’) their retellings of experienced events to suit a particular audience, and these alterations can affect their underlying memory for an event. Higgins and Rholes (1978), for example, had participants read a description of a stimulus person that contained negative, positive, or ambiguous evaluations. Participants then wrote a message describing the person to a receiver whom they were told either liked or disliked the person. Results showed that participants wrote a more negative description when they were told that the receiver disliked the person, and a more positive description when they were told that the receiver liked the person. Additionally, participants’ later recall of the original description of the stimulus person included distortions that were congruent with the message they had written for the receiver.  Thus, participants altered the descriptions in their message to match the receiver’s opinion, and these alterations influenced their later recall; the researchers termed this the saying-is-believing effect, but it has been referred to elsewhere as audience tuning (Hellmann, Echterhoff, Kopietz, Niemeier, & Memon, 2011).In sum,once a memory has been altered through conversation, subsequent retellings are more likely to reflect the most recent retelling, rather than the original memory. This may result from the restructuring of schemas that guide future retellings, as well as selective rehearsal of information (Marsh & Tversky, 2000).


article author(s)