Running head: MERE EXPOSURE EFFECT IN AD POSTPRINT Normal mere exposure effect with impaired recognition in Alzheimer’s disease

We investigated the mere exposure effect and the explicit memory in Alzheimer’s disease (AD) patients and elderly control subjects, using unfamiliar faces. During the exposure phase, the subjects estimated the age of briefly flashed faces. The mere exposure effect was examined by presenting pairs of faces (old and new) and asking participants to select the face they liked. The participants were then presented with a forced-choice explicit recognition task. Controls subjects exhibited above-chance preference and recognition scores for old faces. The AD patients also showed the mere exposure effect but no explicit recognition. These results suggest that the processes involved in the mere exposure effect are preserved in AD patients despite their impaired explicit recognition. The results are discussed in terms of Seamon et al.’s proposal (1995) that processes involved in the mere exposure effect are equivalent to those subserving perceptual priming. These processes would depend on extrastriate areas which are relatively preserved in AD patients. Mere Exposure Effect in AD 3 Normal mere exposure effect with impaired recognition in Alzheimer’s disease The mere exposure effect refers to the increase of positive attitude which results from the repeated exposure to previously novel stimuli (Zajonc, 1968). This effect has proven to be a robust, reliable phenomenon and has been shown on a variety of stimuli (including nonsense words, line drawings, ideographs, faces, possible and impossible three-dimensional objects, melodies etc.), procedures (e.g., forced-choice preference judgments, liking ratings, pleasantness ratings, behavioural indices of stimulus preference) and populations (for reviews, see Bornstein, 1989; Harrison, 1977). In particular, a number of neuropsychological studies have demonstrated a normal mere exposure effect in amnesic or prosopagnosic patients despite a profound inability to recognize the exposed stimuli (Johnson, Kim and Risse, 1985; Johnson and Multhaup, 1992; Greve and Bauer, 1990; see however Redington, Volpe and Gazzaniga, 1984, for contradictory results). These findings offer a strong argument for postulating that the mere exposure effect is an implicit memory manifestation, independent of explicit memory (Kihlstrom, Mulvaney, Tobias et al., 2000). Similarly, two studies have explored the mere exposure effect in Alzheimer’s disease (AD). Winograd, Goldstein, Monarch et al. (1999) found that a mere exposure effect for unfamiliar faces was present in mild to moderate AD patients, although they exhibited impaired recognition memory compared to controls. More recently, Halpern and O’Connor (2000) failed to observe a mere exposure effect for unfamiliar melodies. As the earlier study used the visual modality for stimuli exposition, the authors hypothesised that implicit memory for auditory information, especially nonverbal material, may be defective in AD due to neural degeneration in auditory areas in the temporal lobes. However, Verfaillie, Keane and Johnson (2000) recently found a normal priming effect in AD patients in the auditory perceptual identification of words, despite impaired recognition. As suggested by Verfaillie et al., this discrepancy within the same modality between intact implicit memory for words and impaired Mere Exposure Effect in AD 4 implicit memory for melodies may reflect a dissociation between an intact auditory wordform system and an impaired melodic processing system. These results also suggest that AD partients’ implicit memory performance might not depend exclusively on the modality of presentation, but also on the type of stimuli. Besides this modality difference, other differences make both studies hardly comparable. First, different types of preference measures were used: a forced-choice preference judgment between an exposed and a novel stimulus in Winograd’s et al. (1999) study and a bipolar rating scale (with the endpoints labelled least pleasant ‘1’ and most pleasant ‘5’) in that of Halpern and O’Connor (2000). In fact, pointing at the preferred stimulus does not require the same processes as those involved in translating one’s affective attitude towards a stimulus on a rating scale. Rating might involve more complex processes, such as understanding the correspondence between the scale and the feeling intensity, metacognitive processes, maintaining the choice in working memory long enough to make a rating, etc. All these processes might be too complex or impaired in AD patients, leading to a biased estimation of the mere exposure effect. On the contrary, pointing is a relatively easy and direct way to express the preference. In addition, both studies also differ in the number of exposed stimuli (19 in Winograd et al.’s study vs. 8 in Halpern and O’Connor’s study) as well as in the encoding procedure: Halpern and O’Connor used only two presentations for each melody, and only one encoding task (judging the melody speed), while Winograd et al. used three five-second duration exposures for each face with three different judgments about different face features (eye colour, hair colour, and age). This last condition certainly leads to a more precise perceptual encoding. Another important contrast between these studies is the type of processing involved as a function of the stimulus type. A face presented during 5 sec can be processed as a whole, with all parts being presented in the same time. On the contrary, a 5 sec melody is, by definition, sequential and its constituting parts cannot be processed Mere Exposure Effect in AD 5 simultaneously, but in succession. It is, therefore, possible that the AD patients’ deficit in melodic preference was the consequence of either the demands associated to the preference judgment task or the characteristics of the encoding condition. Finally, a diffrence of age existed between the AD patients groups (78.7 +/6.2 in Halpern & O’Connor’s study and 69 +/10 in Winograd et al.’s study). Even if both groups were relatively equivalent regarding the dementia severity, the age difference between samples might have been a final factor involved in the discrepancy of results between of both studies. To summarise, the evidence concerning the mere exposure effect in Alzheimer’s disease is rather sparse, especially when compared to the numerous studies devoted to the perceptual or conceptual expressions of implicit memory (see Fleischman and Gabrieli, 1998; Meiran and Jelikic, 1995). Furthermore, the only two existing studies have provided contradictory results, and their results are not easily comparable. It is, therefore, hard to characterise the pattern of success and failure of AD patients on the mere exposure effect, when considering the differences between these studies. Finally, a last problem should be mentioned concerning the Winograd et al. (1999) study. Indeed, in their research, preference judgments (Experiment 1) and recognition scores (Experiment 2) were obtained in separate groups of AD patients. Both groups were not statistically different in age, education, or dementia severity, but, because of the neuropsychological heterogeneity of AD, the observed dissociation between a preserved mere exposure effect for faces and an impaired face recognition performance should be interpreted with caution. In addition, in Experiment 2, the healthy control subjects were matched for age with the AD patients but had completed more years of education, which also makes the AD patients’ recognition deficit questionable. Considering all these problems, it clearly appears that further studies are needed in order to better characterise the mere exposure effect in AD. The purpose of the present Mere Exposure Effect in AD 6 research was to re-examine the mere exposure effect for unfamiliar faces in AD patients. Similarly to the Winograd et al.’s study, we used preference and recognition tasks with a same format (forced-choice judgments). However, contrary to that study, preference judgment and recognition performance were explored in the same group of patients and control subjects. Furthermore, very short exposure times and a unique encoding task (age estimation) were used in order to make explicit recognition especially difficult and consequently, to optimise the possible dissociation between recognition performance and preference judgments. Despite these modifications in the encoding condition, we expected the mere exposure effect to be preserved in AD patients as it was in Winograd et al.’s study. Indeed, mere exposure effect does not seem to be favoured by long exposure duration or detailed perceptual encoding (Bornstein, 1989).