Correlations

Pearson correlations between the memory task and questionnaires for the whole sample are reported in Table 3. The main results indicated significant correlations between scores on the checking subscale of the OCI-R and the proportion of MOTOR–

VISUAL confusions, as well as between the total DES score and the MOTOR–I.MOTOR bidirectional score. A significant negative correlation was observed between scores on the checking subscale of the OCI-R and the proportion of hits for MOTOR-encoded actions.

In order to ensure that correlations between scores on the checking subscale of the OCI-R, DES total score and task indices were not due to anxiety or depression but, rather, were specific to checking (and not, for example, to washing), we computed partial correlations, controlling for scores on the STAI-S, STAI-T, BDI-II and the washing subscale of the OCI-R. These partial correlations confirmed the zero-order correlations previously described.

Table 3. Pearson correlations between questionnaires and reality monitoring task OCI-R OCI-R

checking OCI-R

washing STAI-S STAI-T BDI-II DES

Hits MOTOR –.05 –.32** .13 –.28* –.09 –.10 –.17

MOTOR – I.MOTOR –.01 .03 –.09 –.01 –.07 –.18 .16

MOTOR – VISUAL .05 .31** .03 .01 .08 .16 .13 MOTOR – I.VISUAL .12 .08 .03 –.03 –.03 .08 –.01 Bidirec

MOTOR – I.MOTOR –.00 –.04 –.03 –.13 –.15 –.09 .25*

Note: * p < .05 ; ** p < .01; MOTOR–I.MOTOR: proportion of MOTOR-encoded actions recognized as I.MOTOR; MOTOR–VISUAL: proportion of MOTOR-encoded actions recognized as VISUAL; MOTOR–

I.VISUAL: proportion of MOTOR-encoded actions recognized as I.VISUAL; Bidirec MOTOR–I.MOTOR:

bidirectional score of confusions between MOTOR and I.MOTOR; OCI-R washing: Obsessive-Compulsive Inventory – Revised, washing subscale

4. Discussion

The main purpose of this study was to re-examine the issue of motor memory and reality monitoring abilities in checking-prone participants by using a task that presented actions as stimuli. Our results can be summarized as follows. First, and in line with previous findings (Anderson, 1984; Engelkamp, 1998; Larøi et al., 2005), the sample as a whole made fewer attribution errors after a “real” condition of encoding (MOTOR and VISUAL) than an “imaginary” condition of encoding (I.MOTOR and I.VISUAL).

Furthermore, fewer attribution errors were made in the “imaginary” compared to the VERBAL encoding condition.

When comparing checking-prone with non-checking-prone individuals, no differences were found between the two groups in terms of participants’ evaluations of the task (i.e., incidental instructions). However, checking-prone individuals recognized more VERBAL-encoded actions than non-checking-prone participants. This finding is in accordance with previous studies that observed better recognition in checkers than in non-checkers (e.g., MacDonald et al., 1997). It is also in line with works reporting the absence of a verbal memory deficit in OCD (e.g., Radomsky & Rachman, 1999) or even a better word recognition ability in checkers (Sheffler Rubenstein et al., 1993).

Nevertheless, we observed that checking-prone participants recognized fewer MOTOR-encoded actions than non-checking-prone individuals, thus confirming previous studies (e.g., Ecker et Engelkamp, 1995; Sheffler Rubenstein et al., 1993; Sher et al., 1983, 1984, 1989). This motor memory difficulty, as well as the subjects’ good performance for VERBAL-encoded actions, also emerged when they used the R/K/G judgment. That is, checking-prone individuals gave more R responses for VERBAL-encoded actions and fewer R responses for MOTOR-encoded actions than non-checking-prone participants.

Overall, these results seem to confirm the presence of a specific motor memory difficulty in checking-prone participants, as observed by Ecker and Engelkamp (1995) in clinical OCD patients. It should also be noted that the low number of R responses for

MOTOR-encoded actions could be due to a difficulty in either the encoding or the retrieval (recollection) of actions. However – and in accordance with Reed’s (1977, 1985) hypothesis – it seems more likely that it is related to checkers’ difficulties in accurately encoding MOTOR actions. Anyway, this interpretation merits further examination. From this perspective, one must emphasize the possible role of state anxiety in the motor memory deficit. Indeed, our results indicate that scores on the STAI-S were negatively correlated with the number of MOTOR actions correctly recognized. In other words, it appeared that the high anxiety state experienced by checking-prone participants could make encoding and/or retrieving of MOTOR-encoded actions more difficult. This relationship could be tentatively interpreted by suggesting that anxiety causes individuals to pay particular attention to their bodily state and physical sensations (Ludwick-Rosenthal & Neufeld, 1985; Zoellner & Craske, 1999). Consequently, it could be argued that the focus on body sensations may disturb the encoding and/or the recollection of the kinesthetic and motor characteristics of actions. This interpretation is more generally in line with Rachman’s (2002) model, which postulates that state anxiety, enhanced by what he calls “multipliers” (the perceived responsibility, the perceived probability of the feared harmful event and the perceived severity of the feared harmful event), impairs the recollection of situations that are likely to elicit checking. According to Rachman, when OCD patients are anxious, they focus most of their attention on the threat and on a scanning of their emotional reactions to the threat, which leads to a poor recording of the specific details of their checking. Regarding our study, it is useful to remember that checkers’ difficulty in recognizing MOTOR actions cannot be totally attributed to their anxiety state, but that there also exists a direct link between checking and reality monitoring, as showed by the partial correlations.

With regard to the encoding of modality attributions, our results revealed first that checking-prone participants were more likely to confuse MOTOR with VISUAL-encoded actions and MOTOR- with I.VISUAL-encoded actions than non-checking-prone participants. In other words, checking-prone individuals seemed to be in greater difficulty when required to distinguish between an action they had performed themselves, and an action the experimenter performed – whether it was a real or an imagined action. These confusions might be interpreted in light of Reed’s hypothesis (Reed, 1977, 1985): in attributing their own actions to the experimenter, checking-prone individuals seem to remember events from the viewpoint of a non-participant observer.

Second, and contrary to our hypothesis, our results indicated that checking-prone individuals did not confuse MOTOR with I.MOTOR conditions more often than non-checking-prone individuals. This might suggest that checkers’ reality monitoring abilities are preserved. However, we also observed that the MOTOR–I.MOTOR confusion, although not correlated with the checking subscale of the OCI-R in the sample as a whole, clearly correlated with the dissociation scale. This finding confirms the link between dissociation, checking, and memory deficits postulated by some authors (e.g., Watson et al., 2004). It also suggests that reality monitoring difficulties may be linked to checking

through dissociation. More specifically, it can be postulated that the tendency to dissociate may make the encoding of actions less accurate, which in turn leads to more impersonal memories for actions (Reed, 1977). In addition, it should be noted that the more frequent MOTOR–I.VISUAL confusions observed in checking-prone participants, which involve a confusion between a real condition (MOTOR) and an imagined one (I.VISUAL), could also be interpreted as reflecting a reality monitoring deficit in checking-prone participants. However, this interpretation should be cautiously considered, given that significance was not quite achieved. Furthermore, we must emphasize that the MOTOR–I.VISUAL confusion did not correlate with the dissociation score.

In sum, differences between checking-prone and non-checking-prone participants for MOTOR–I.VISUAL confusions, but not for MOTOR–I.MOTOR confusions, as well as the different pattern of correlation between dissociation and these two types of confusions, suggest that the mechanisms involved in checkers’ reality monitoring difficulties are complex and multi-determined. Apart from the possible role of dissociation, anxiety could also be considered as a factor influencing reality monitoring performance. As mentioned earlier in connection with the motor memory difficulty, state anxiety may decrease the attention paid to the encoding of action in favor of bodily sensations and emotion regulation. Another possible factor, consistent with our data, could be checkers’ difficulty in encoding MOTOR actions. Thus, the motor memory deficit could cause individuals to have less precise and distinctive memory traces of actions, which in turn could result in a difficulty deciding whether the action was really performed or just imagined. Furthermore, reality monitoring deficits may also be due to a difficulty affecting the retrieval of action memories. More specifically, it could be suggested that intrusions or obsessional thoughts interfere with the retrieval of the information that would permit the individual to distinguish between imagined and real actions. Finally, in accordance with Brown et al. (1994), an additional interpretation could assume that checkers have particularly vivid mental images of actions, which could impair their reality monitoring.

Before concluding, we must point out two limitations on the present study. One concerns the relative insensitivity of the memory task we used, leading to ceiling effects and to small or tendency-based differences between groups. The other concerns the fact that we explored non-clinical checkers. Accordingly, our study would benefit from a replication with a clinical sample using a more sensitive task.

In conclusion, the findings of this study suggest the existence of a motor memory deficit in checking-prone individuals. They also seem to confirm Reed’s (1977, 1985) hypothesis postulating that checkers’ memories for actions take an observer’s point of view. Finally, they suggest that the issue of reality monitoring deficits in checking is more complex than had previously been presumed. Different factors could influence the appearance of these deficits, such as dissociation tendency, anxiety, motor memory deficits, retrieval difficulties due to intrusions, or vivid mental images. Taken as a whole,

this study sheds new light on motor memory and reality monitoring deficits in OCD checkers and contradicts previous works that recommend abandoning the memory deficit hypothesis. If replicated with a clinical sample, this study could highlight the importance of basing therapeutic interventions on memory enhancement, such as the doubt-reduction treatment suggested by Tallis (1993).

ETUDE 3

Phenomenological characteristics of autobiographical memories and imagined events in sub-clinical obsessive-compulsive checkers

Abstract

Phenomenological characteristics of autobiographical memories and imagined experiences were examined in checking and non-checking-prone individuals. Participants were asked to retrieve a positive, negative and neutral memory, and to imagine a positive, negative and neutral experience. They were then requested to evaluate each event according to characteristics such as sensory and contextual details. Main results revealed that non-prone participants reported more general vividness than checking-prone individuals for real events. In addition, non-checking-checking-prone individuals reported more visual details and vividness for real compared to imagined experiences, while no difference was found for checking-prone participants between real and imagined events.

These results suggest that checking-prone participants report poor memories of real events, which could in turn explain difficulties distinguishing between real and imagined events.

23 This study is a reprint of the article: Zermatten, A., Van der Linden, M., d’Argembeau, A., & Ceschi, G.

(2006). Phenomenological characteristics of autobiographical memories and imagined events in sub-clinical obsessive-compulsive checkers. Manuscript accepted with minor revision in Applied Cognitive Psychology.

http://www3.interscience.wiley.com/cgi-bin/jhome/4438?CRETRY=1&SRETRY=0

24 This study was supported by the Swiss National Science Foundation, grant no. 1114-067135.01/01. We would like to thank Maude Chollet and John Kadous for their help with the data collection.

1. Introduction

According to Reed (1977, 1985), individuals presenting Obsessive-Compulsive Disorders (OCD) with checking compulsions (i.e. checkers) tend to remember actions from the perspective of a non-participant observer and to rely more on visual imagery, whereas normally, one tends to emphasise subjective experience and to rely on images from all modalities including the kinaesthetic. These memory characteristics could cause checkers to have a poor memory for actions. More specifically, checkers may find it difficult to figure out whether they have really performed an action (such as locking the door) instead of merely imagined it, and consequently engage in their checking rituals.

In their theoretical framework, Johnson and Raye (1981) described the ability to discriminate between perceived and imagined events as the reality monitoring capacity.

According to these authors, this capacity depends on an evaluation of the event’s memory characteristics. In short, memories of perceived events contain more details and more perceptual and contextual information than memories of imagined events (e.g. Suengas &

Johnson, 1988). Thus, in order to determine whether an event was perceived or imagined, individuals normally base their judgements on contextual and perceptual information contained in the memory. Concerning OCD, Brown et al. (1994) evoked the possibility that checkers had a lower sensitivity to the difference between the memory of a percept and the memory of a mental image than non-checkers. In other words, checkers might present either poor and not very detailed memories of real events, or very vivid and detailed representations of imagined events (as found by de Silva, 1986). Both these possibilities could lead to a difficulty distinguishing a real from an imagined event.

The reality monitoring deficit in checkers has been explored experimentally by a number of studies, the outcomes of which were contradictory. Some research showed reality monitoring deficits in clinical (Ecker & Engelkamp, 1995) or sub-clinical (Sheffler Rubenstein et al., 1993) checkers, whereas others did not show any deficit in clinical checkers (Brown et al., 1994; Constans et al., 1995; Hermans et al., 2003; McNally &

Kohlbeck, 1993; Merckelbach & Wessel, 2000). As we pointed it out in one of our earlier studies (Zermatten, Van der Linden, Larøi, & Ceschi, 2006c), some methodological issues could explain these discrepancies (e.g. material including words for some studies and actions for others, ceiling or floor effects in some of the recognition tasks used). These methodological limitations were circumvented in our earlier study (Zermatten et al., 2006c) through the implementation of an ecological procedure based on the memorisation of actions. More specifically, we proposed to 75 undergraduate students evaluated by the Obsessive Compulsive Inventory – Revised (Foa et al., 2002a) to encode daily actions in five different modalities (to perform the action, to imagine oneself performing it, to see the experimenter performing it, to imagine the experimenter performing it and to verbally repeat a description of the action). In a subsequent recognition test, participants had to say in which modality each action was encoded. Our findings revealed that checking-prone individuals recognised fewer performed actions (i.e. motor-encoded actions) than non-checking-prone participants, suggesting a motor

memory deficit. We also observed that checking-prone participants were more likely to confuse actions they performed themselves with actions the experimenter performed, confirming Reed’s hypothesis of an observer point of view in checkers’ memories of actions. Finally, reality monitoring abilities were found to be partly linked to checking, but also to dissociation.

Another way of studying reality monitoring deficits, is to explore checkers’ subjective representations of real and imagined events, by investigating phenomenological (or qualitative) characteristics of these events. Johnson, Foley, Suengas and Raye (1988) developed the Memory Characteristics Questionnaire (MCQ) in order to evaluate these characteristics. In this questionnaire, participants are requested to remember true autobiographical memories or to generate simulated autobiographical events, and to rate these events on different Likert scales evaluating different characteristics of the imagined or recalled events, such as sensory details (visual, tactile, auditory, gustatory, and olfactory), spatial information, temporal information or associated emotions. Using this questionnaire, Johnson et al. (1988) showed that perceived events were given higher ratings than imagined events on various characteristics such as perceptual or contextual information.

Several studies have used a modified version of the MCQ with a non clinical population. Destun and Kuiper (1999), for example, observed that autobiographical memories of positive events were characterised by more detailed sensorial and contextual information than memories of stressful events (see also Larsen, 1998; Raspotnig, 1997).

More recently, D’Argembeau, Comblain and Van der Linden (2003) observed that autobiographical memories of positive events contained more sensorial and contextual details than those for either negative or neutral events, while no differences were found between memories of neutral and negative events. These authors also evaluated the perspective from which memories were retrieved. According to Nigro and Neisser (1983), when one remembers a situation, one may ‘see’ the situation from one’s own point of view (Field memories, F) or one may ‘see’ the situation as if one was an external observer (Observer memories, O). In their study, D’Argembeau et al. observed that positive and negative events were more often recollected with a field perspective than neutral events.

To the best of our knowledge, the phenomenological details of memories for real and imagined autobiographical events have not yet been examined either in sub-clinical or in clinical checkers. This exploration could lead to have an idea of the subjective experience of real and imagined events in checkers, and in turn, to better understand their reality monitoring abilities. In order to explore this issue, we gave undergraduate students, whose checking symptoms were evaluated, a modified version of the MCQ to assess the phenomenological characteristics of negative, neutral and positive autobiographical memories, as well as representations of negative, neutral and positive imagined experiences. The characteristics explored were the sensory details (visual, tactile, auditory, gustatory, and olfactory), the contextual information (spatial placement of objects, place and time of event), the intensity of feelings (positive and negative emotions)

and the general vividness of the real and imagined events. In accordance with Brown et al.

(1994), we hypothesised that checking-prone participants will generally show fewer differences between the number of details for real and imagined events than non-checking-prone individuals. More specifically, and in accordance with works showing memory deficits in checkers (see for example the meta-analysis of Woods et al., 2002), we postulated that this difficulty may be the result of a poorer memory for real events rather than of particularly vivid mental images. In other words, we expected that checking-prone participants will report less details for real events comparing to non checking-prone participants. In addition, we explored if the affective valence of the recalled events affected the memory characteristics of checking-prone and non-checking-prone participants differently. In particular, we were interested to see whether checkers and non-checkers coped differently with negative and positive events, which could in turn affect the characteristics of real and imagined memories. Finally, in order to test Reed’s hypothesis that checkers are characterised by a kinaesthetic deficit and a non-participant observer perspective, we also included a question concerning the kinaesthetic aspects of the events and the point of view adopted by the participants in the events they reported.

2. Method

2.1. Participants

The participants were 79 non clinical participants (48 females, 31 males) aged between 19 and 35 years old with a mean age of 24.96 (SD = 4.24). Their mean years of education amounted to 15.22 (SD = 1.91).

For the statistical analyses, two groups were created based on participants’ checking sub-scores on the revised form of the Obsessive-Compulsive Inventory (Foa et al., 2002a;

French version: Zermatten et al., 2006b). The checking-prone group consisted of individuals presenting a score within the top quartile of the distribution (score > or = 3; N

= 23). As for the non-checking-prone group, it included participants within the lowest quartile of the distribution, that is with a score of 0 (N = 32). However, more than a quarter of the participants presented a score of 0, which explains why the non-checking-prone group is a bit larger than the other group. The general characteristics of the two groups are presented in Table 1. No significant differences were found between groups for age and education. However, significant differences were observed in terms of obsessive-compulsive symptoms, anxiety and depression. In addition, one can observe that the

= 23). As for the non-checking-prone group, it included participants within the lowest quartile of the distribution, that is with a score of 0 (N = 32). However, more than a quarter of the participants presented a score of 0, which explains why the non-checking-prone group is a bit larger than the other group. The general characteristics of the two groups are presented in Table 1. No significant differences were found between groups for age and education. However, significant differences were observed in terms of obsessive-compulsive symptoms, anxiety and depression. In addition, one can observe that the