RUNNING HEAD : TRUE & FABRICATED MEMORIES True and Intentionally Fabricated Memories

The aim of the experiment reported here was to investigate the processes underlying the construction of truthful and deliberately fabricated memories. Properties of memories created to be intentionally false – fabricated memories – were compared to properties of memories believed to be true – true memories. Participants recalled and then wrote or spoke true memories and fabricated memories of everyday events. It was found that true memories were reliably more vivid than fabricated memories and were nearly always recalled from a first person perspective. In contrast, fabricated differed from true memories in that they were judged to be reliably older, were more frequently recalled from a third person perspective, and linguistic analysis revealed that they required more cognitive effort to generate. No notable differences were found across modality of reporting. Finally, it was found that, intentionally fabricated memories were created by recalling and then ‘editing’ true memories. Overall, these findings show that true and fabricated memories systematically differ, despite the fact that both are based on true memories. TRUE & FABRICATED MEMORIES Page 3 One of the functions of memory is in imagining. Imagining, for example, how the future might be or how the past might have been otherwise. Indeed, memory and imagining are so interconnected it has been suggested that together they form a remembering-imagining system (Conway, 2009). Although much research has focused on the association between imagining the future and autobiographical memory (Addis, Wong, & Schacter, 2007; D’Argembeau & Van der Linden, 2004; Hassabis, Kumaran, & Maguire, 2007; Hassabis, Kumaran, Vann, & Maguire, 2007; Newby-Clark & Ross, 2003; Schacter & Addis, 2009; Schacter, Addis, & Buckner, 2008; Szpunar & McDermott, 2008; Tulving, 1985; Tulving, 2002), little research has investigated our ability to imagine an alternative past. Therefore, in the present study we directly compare intentionally fabricated autobiographical memories (IFAMs) with autobiographical memories (AMs) the rememberer believes to be true and which they experience as memories. An IFAM is an entirely or partially fabricated memory, consisting primarily, but not exclusively of false facts, as opposed to the expression of false opinions or beliefs (see Newman et al., 2003, for differences in false opinions). IFAMs may arise in forensic contexts, and become particularly pivotal in instances when memory is the only form of evidence available. The sorts of cases in which memories are the only evidence include, what in the UK are termed, cases of ‘historic’ sexual abuse (typically memories dating to childhood recalled by an adult complainant), accident assessments, war, torture, political asylum, and plagiarism. Moreover, reports of fabricated memories are a common feature of forensic interviews and interrogations (Porter & Yuille, 1996, Porter, Yuille and Lehman, 1999) and a number of reasons and motivations for deliberately fabricating memories may exist including revenge, control and monetary gain (Yuille, Tymofievich, & Marxsen, 1995). TRUE & FABRICATED MEMORIES Page 4 However, a small body of work has begun investigating IFAMs, with particular focus falling on systematic differences between real and fabricated memories (Conway, Pleydell-Pearce, Whitecross, & Sharpe, 2003; Merckelbach, 2004; Porter, Peace, & Emmett, 2007; Porter, Yuille, & Lehman, 1999), Although little empirical work has investigated the construction processes of IFAMs, it has been suggested that an IFAM is created by means of a ‘lie script’ . According to this view, a generic representation of an event is generated and is then used in place of a specific memory of a single event. A process that has been termed ‘superficial encoding’ (Porter & Yuille, 1996). This theory has had some support, (Granhag & Strömwall, 1999, Granhag, Strömwall, & Jonsson, 2003) and has led to the view that truth tellers create their accounts through reconstruction, whereas liars attempt to accurately repeat a previously rehearsed script. According to these theories, it is these differences in processing and storage that are responsible for systematic and measureable differences in in truthful and fabricated memory accounts (Colwell, HiscockAnisman, Memon, Taylor, & Prewett, 2007). Porter (1998) further suggests that verbal ideas and images may play a role in IFAM generation, arguing that fabricated memories are “imaginative constructions”. However, Porter (1999), later, proposed that fabricated memories may not be entirely imaginative in nature, but like false memories (untrue memories that, unlike fabricated memories are not known to be false), may incorporate elements of an experienced event with the purpose of enhancing the credibility of an account. In a rather similar and related way it has been suggested that the process of lying involves firstly accessing true beliefs followed by denial and/or distortion of these beliefs (Polage, 2004). Despite this work, to our knowledge, no empirical research has TRUE & FABRICATED MEMORIES Page 5 explicitly addressed the construction processes involved in IFAM construction (with the exception of Conway, et al., 2003). Therefore, the primary aim of the present research is to understand the way in which IFAMs are constructed within the autobiographical memory system. By understanding such processes, we can better understand how IFAMs are stored, rehearsed and recalled, and why differences may occur in their content. The generation or construction of AMs involves the effortful, iterative, access of autobiographical memory knowledge structures and the gradual establishment of patterns of activation/inhibition across distributed neural networks that come to form an AM in an act of remembering (Cabeza & Jacques 2007; Conway, et al., 2003, 2005; Conway & Pleydell-Pearce, 2000). It seems that the construction of IFAMs may involve similar processes, not least because the generation of IFAMs may feature the recall of AMs. It is difficult to conceive of a process of IFAM generation that did not feature, to a least some extent, access of autobiographical knowledge and possibly the generation of specific AMs. The present research is consistent with this and follows from the assumption that IFAM creation involves initially accessing information in long-term memory, followed by a conscious “editing” phase. In other words, an AM is activated and then consciously edited to produce an IFAM. Some evidence supporting this assumption comes from an EEG study by Conway, et al., (2003) contrasting the construction and retention of IFAMs and AMs. In the construction phase no differences in activation were found between IFAMs and AMs and the patterns of activation were highly similar to those observed in a prior TRUE & FABRICATED MEMORIES Page 6 study of AM generation (Conway, Pleydell-Pearce, & Whitecross, 2001) . However, in a retention phase during which IFAMs and AMs were held in mind for 10s a major difference did emerge and that was increased right frontal activation during IFAM retention. This may have reflected the editing of AMs and subsequent difficulties in consciously maintaining the novel IFAM representation. Because of the editing process, and effort required to maintain a novel representation, IFAMs should differ from AMs in some of their recollective qualities. For example, memories not based on real experience may be associated with less vivid mental imagery (Johnson & Raye, 1981), may be placed further back in the past to demonstrate a stable, long-held memory and may be recalled more frequently through an observer perspective, since an observer perspective has been shown to occur following distortion (Freud, 1915; Nigro & Neisser, 1983). Further, AMs may require less cognitive effort to generate (Vrij, et al., 2008) and may also be faster to create. We, therefore, expect to find linguistic constructs within accounts of IFAMs that are indicators of cognitive effort, for example fewer complex words e.g. those over six letters in length and a higher rate of non-fluencies, e.g. “erm”, “umm”, particularly for spoken accounts). Further, we expect that IFAMs will contain more motion words, e.g. walk, go, run. Motion words have been found to be a characteristic of increased cognitive effort, used by a rememberer to reduce the complexity of an account by referring to simple actions rather than expression of emotions or other metacognitive reasoning (Newman, Pennebaker, Berry, & Richards, 2003). Finally, IFAMs may be shorter in length than AMs to avoid unnecessary discussion of detail. TRUE & FABRICATED MEMORIES Page 7 Additional linguistic differences such as an increase in negative emotion words, fewer exclusive and sensation words in IFAMs have been shown by Newman et al. (2003), however we did not expect to find these linguistic differences since our study was investigating fabricated memories rather than fabricated opinions and denials (Newman, et al. , 2003). In summary then, four recollective qualities, vividness, memory / image generation times, retention interval, and perspective and four linguistic constructs, complex words, non-fluencies, motion words and account length, all assumed to reflect cognitive effort and / or an editing process, were investigated. Accounts of the memories were recorded by either typing or by tape recording while they were spoken.