پردازش جانبی گرا از حافظه کاذب و تعصب توجه در پیری

Aging is associated with higher propensity to false memories and decreased retrieval of previously studied items. When young adults (YA) perform on a lateralized version of the Deese–Roediger–McDermott (DRM) paradigm, the right cerebral hemisphere (RH) is more sensitive than the left (LH) to false memories, suggesting hemispheric imbalance in the cerebral mechanisms supporting semantic and episodic memory processes. Since cerebral asymmetries tend to be reduced with age, we surmised that behavioral asymmetries in the generation of false memories would be diminished with aging. To probe this hypothesis, a lateralized version of the DRM paradigm was administered to healthy older adults (OA) and YA. During the encoding phase, lists of semantically associated words were memorized. During the retrieval session, targets (previously seen words), lures (LU) (never seen strongly semantically related words) and distracters (never seen, unrelated words) were briefly displayed either in the left or right visual fields, thus primarily stimulating the RH or LH, respectively. Participants had to decide whether the word was previously studied (Old/New), but also whether they had a strong episodic recollection (Remember) or a mere feeling of familiarity (Know) about Old words. In line with our predictions, false memories were globally higher in OA than YA, and vivid false recollections (i.e., Remember responses) were higher when LU were presented in the RH in YA, but not in OA. Additionally, we found significant correlations between YA participants’ Familiarity scores and leftward attentional bias as previously evidenced using a visuospatial landmark task (Schmitz and Peigneux, 2011), an effect not present in OA. This result is in line with the hypothesis of an interplay between attentional resources allocated to visuospatial and memory processes, suggesting a memory pseudoneglect phenomenon that would be altered with aging.

Originally introduced by Deese (1959) then updated by Roediger and McDermott (1995), the Deese/Roediger–McDermott (DRM) task is probably the most widely used paradigm to investigate the development of false memories (Brainerd et al., 2011, Gallo, 2010 and Pezdek and Lam, 2007). In the DRM procedure, participants have to study thematic lists of words semantically related to a critical, but never-presented lure (e.g., the non-presented lure sleep is semantically related to a network of words to be learned such as bed, rest, night, awake, etc.). In subsequent recall and/or recognition tests, participants tend to erroneously accept the critical lure as being part of the list of learned words, a false memory effect consistently observed across hundreds of studies (for reviews see Gallo, 2010 and Roediger and McDermott, 2000). Even when informed about the risk to produce erroneous memories in an attempt to reduce hit and false alarm rates, participants persist in the false recognition of the critical lure ( Gallo et al., 2001 and McDermott and Roediger, 1998), demonstrating the robustness of this effect. Moreover, the critical lure can be associated with a strong feeling of recollection, i.e., participants’ recognition can be accompanied by specific details of the study episode instead of a simple feeling of familiarity ( Gallo and Roediger, 2003 and Roediger and McDermott, 1995). The false memory phenomenon observed in the DRM paradigm can be explained using two complementary, not mutually exclusive theories. According to the Fuzzy Trace Theory (FTT; Brainerd and Reyna, 2002 and Brainerd et al., 2008), both verbatim and gist traces are encoded and activated during the study phase. Verbatim traces represent the surface details of the stimuli and gist traces constitute the core meaning of the stimuli without the perceptual details. Thus, accurate memory of studies items is principally based on verbatim traces whereas false memory is mainly driven by gist traces. Alternatively, according to the Activation-Monitoring Account (AMA; Gallo, 2010, McDermott and Watson, 2001 and Roediger et al., 2001), two opposing processes give rise to false memories. During the encoding of the thematic lists, the critical lure is activated either consciously by elaborative processes or automatically through spreading activation within an associative network. During the testing phase when a critical lure is activated, participants have to differentiate between the activation generated by the actual presentation of the item and its previous internal activation. Reality-monitoring confusions may then appear for critical lures (LU) as compared to semantically unrelated distracters, leading to the false memory phenomenon. At the functional neuroanatomical level, prefrontal and temporal regions appear to play both common and different roles in the generation and expression of false memories. Indeed, neuroimaging data have highlighted left prefrontal (PFC) and lateral temporal cortices activations during lists encoding, likely reflecting semantic associations processing (Burton et al., 2003 and McDermott et al., 2003). However, left PFC activation during encoding is predictive of participants’ performances on both studied items and critical LU (Kim and Cabeza, 2007a and Kubota et al., 2006), whereas left medial temporal lobe activity is associated to true memories only (Kim and Cabeza, 2007a), implying elaborative semantic processes and the storage of real events, respectively. Nonetheless, the anterior temporal lobe may also contribute in semantic activation since magnetic (Gallate et al., 2009) or electrical (Boggio et al., 2009) stimulation at encoding reduces false memory formation. Besides, it has been shown that hippocampal activity during lists encoding is predictive of the occurrence of list-related false memories up to three days later (Darsaud et al., 2011). During the retrieval phase, hippocampal and the left ventrolateral PFC activities may reflect the recovery of semantic information whereas dorsolateral PFC may subtend source-monitoring (Cabeza et al., 2001). Moreover, high confidence judgments in true and false memories have been associated with medial temporal lobe and fronto-parietal activities, respectively (Kim and Cabeza, 2007b). Overall, temporal and prefrontal activations during encoding and retrieval are in accordance with the two distinct processes of semantic activation and source-monitoring proposed in the AMA model (Gallo, 2010). Using a modified version of the DRM paradigm, hemispheric differences have been also highlighted in the processing of true and false memories (Bellamy and Shillcock, 2007, Ben-Artzi et al., 2009, Faust et al., 2008, Ito, 2001 and Westerberg and Marsolek, 2003). Indeed, when primarily targeted using a divided visual fields procedure, the left hemisphere (LH) tends to be more accurate for studied items (Ito, 2001) and to reject more easily the critical LU (Bellamy and Shillcock, 2007 and Westerberg and Marsolek, 2003). Moreover correct rejections of critical LU in the LH are associated with higher confidence levels (Westerberg and Marsolek, 2003). Overall, these results suggest that the LH better discriminates true from false memories than the right hemisphere (RH). However, the LH may be more susceptible to critical LU when encoding is composed of dominants-meaning lists as compared to subordinates-meaning lists (Faust et al., 2008). Hence, LH-dependent false memory may be enhanced by the semantic strength of the material presented at encoding. Interestingly, as compared to the LH, the RH generates less false memories than the LH and is not affected by the strength of the associates at encoding. Nevertheless, the RH becomes more prone to false memory when participants have to study short texts containing semantically associated words (Ben-Artzi et al., 2009). A common explanation for these LH/RH differences fits in the framework of the Fine-Coarse Semantic Coding Theory (FCT; Beeman, 1998 and Jung-Beeman, 2005), according to which each hemisphere differently processes semantic information during word processing, in that semantic fields are more focused in the LH and more diffused in the RH. Consequently, the RH is more likely than the LH to activate a concept connected by distant semantic relations, and in the DRM task, a thematic list (e.g., bed, rest, night, awake, etc.) is more susceptible to activate the critical lure (i.e., sleep) in the RH. Accordingly, critical LU are more confidently rejected when primarily presented in the LH ( Westerberg and Marsolek, 2003). Likewise, the LH produces more critical LU when coming from dominant-meaning lists being strongly activated, whereas the RH is sensitive to the same degree to subordinate- and dominant-lists ( Faust et al., 2008). A contrario, short texts containing semantically associated words will prime more false memories in the RH since this hemisphere is more efficient for the storage of long term and wider semantic fields ( Ben-Artzi et al., 2009). Additionally, the FCT proposes that the temporal lobe involves semantic activation and integration during word processing, whereas the semantic selection of different meanings depends on the PFC ( Jung-Beeman, 2005), a proposal fitting well with previous DRM studies that showed the crucial involvement of these regions during both encoding ( Boggio et al., 2009, Burton et al., 2003, Darsaud et al., 2011, Gallate et al., 2009, Kim and Cabeza, 2007a and McDermott et al., 2003) and retrieval ( Cabeza et al., 2001 and Kim and Cabeza, 2007b). In this respect, the FCT ( Jung-Beeman, 2005) may be considered as an extension of both FTT ( Brainerd and Reyna, 2002) and AMA ( Gallo, 2010) conceptualizations, at least regarding the semantic activation and monitoring of the critical lure in the DRM. Aging processes affect performances in DRM tasks, with reduced accurate retrieval and increased false memories in older adults (OA) as compared to young adults (YA; e.g., Balota et al., 1999, Dehon, 2006 and Dehon and Bredart, 2004; for a review see Schacter et al., 1997). Different patterns of cerebral activations were observed in OA performing the DRM (Dennis et al., 2007 and Dennis et al., 2008), with a reduction of hippocampal activity both at encoding (Dennis et al., 2007) and retrieval (Dennis et al., 2008) of previously presented items, indicating possible verbatim trace impairments during these two memory stages. Conversely, increased left temporal activity was associated with false memory encoding and retrieval in OA. These age-related activations suggest a particular sensitivity of OA to the activation of the gist traces within semantic networks. Interestingly, OA exhibit decreased activity in the left PFC together with increases in its right homologous parts during encoding of items that will be correctly retrieved at test (Dennis et al., 2007). This balanced activation pattern in the PFC may reflect compensatory mechanisms and/or difficulties during semantic information monitoring. According to the hemispheric asymmetry reduction in older adults (HAROLD) model (Cabeza, 2002 and Daselaar and Cabeza, 2005), OA tend to recruit less lateralized PFC activity than YA in order to counteract cognitive decline. Considering hemispheric differences in false memory processing in YA (Bellamy and Shillcock, 2007, Ben-Artzi et al., 2009, Faust et al., 2008, Ito, 2001 and Westerberg and Marsolek, 2003) and the reduction of cerebral asymmetries coupled with additional LH activation during both encoding and retrieval in OA (Cabeza, 2002, Dennis et al., 2007 and Dennis et al., 2008), one should expect an age-related reduction of behavioral asymmetries for false memories in OA. In the present study, we tested this prediction in YA and OA using a lateralized version of the DRM to investigate the propensity of each cerebral hemisphere to false memories. Additionally, we used a Remember/Know procedure to assess participants’ levels of recollection and familiarity (Tulving, 1985; for reviews see Gardiner, 2008 and Yonelinas, 2002). As a reminder, recollection makes reference to the retrieval of vivid memories within their encoding context, whereas familiarity refers to the feeling that an event happened but without any contextual information. DRM studies have shown that LU’ retrieval may be associated with the recollection of never happened contextual details (Roediger and McDermott, 1995). In this context, we predicted less false recollections in the LH that seems less sensitive to critical LU (Bellamy and Shillcock, 2007 and Westerberg and Marsolek, 2003) and more confident in rejection (Westerberg and Marsolek, 2003). Additionally, the LH might generate more recollections for previously encoded items (Ito, 2001). From an aging perspective and according to the HAROLD model, hemispheric asymmetry in recollections should be reduced or even abolished in OA. Finally, over and above the HAROLD model (Cabeza, 2002), it has been suggested that normal aging is associated with a RH decline (Dolcos et al., 2002). We recently obtained results supporting this hypothesis, showing an age-related suppression of the pseudoneglect phenomenon in a perceptual landmark (LDM) task (Schmitz and Peigneux, 2011). Pseudoneglect is a slight but consistent attentional leftward bias (LWB) commonly observed in young healthy populations (Bowers and Heilman, 1980 and Jewell and McCourt, 2000), purportedly explained by an RH dominance in the attentional fronto-parietal network (e.g., Cicek et al., 2009). Besides, attentional mechanisms – especially related with activity in parietal regions – are believed to contribute in episodic memory processes (Cabeza, 2008, Cabeza et al., 2008, Cabeza et al., 2012, Ciaramelli et al., 2008 and Wagner et al., 2005). Given that the right posterior parietal cortex also plays a crucial role in the pseudoneglect phenomenon (e.g., Cicek et al., 2009, Ellison et al., 2004, Ghacibeh et al., 2007 and Weiss et al., 2003), we wondered whether attentional biases in YA and OA would be predictive for laterality effects in DRM performance. To probe this additional, exploratory hypothesis, we tested the same young and aged populations as in our pseudoneglect study (Schmitz and Peigneux, 2011).