Memory and metamemory in obsessive–compulsive disorder

Abstract

Studies on the link between checking and memory problems have produced equivocal results regarding a general memory deficit in patients with obsessive–compulsive disorder and subclinical checkers. However, there is clear and consistent evidence that patients with obsessive–compulsive disorder (OCD) show lack of confidence in their memory performance. The purpose of the present study was to investigate memory and metamemory performance (feeling-of-knowing judgments) for neutral and threat-related material in three groups: OCD patients (OCs), subclinical checkers (SCs), and normal controls (NCs). Participants studied a list of neutral and threat word pairs. After an initial cued-recall test, they provided feeling-of-knowing (FOK) judgments for unrecalled word pairs, followed by a recognition test. The results showed that OCs but not SCs were impaired in both recall and recognition compared to NCs. OCs were also less confident about their future memory performance than the other two groups, as reflected in their lower FOK ratings. Moreover, FOK judgments of the OCs were not reliable predictors of their recognition performance. Finally, neither OCs nor SCs showed any evidence of memory bias for threat-relevant information. The results support the idea of a general memory and a metamemory deficit in OCs.

Introduction

Empirical evidence on the link between compulsive checking and memory problems has been inconsistent. Comparisons of obsessive–compulsive disorder patients (OCs) with normal controls sometimes found no difference in episodic memory tasks of recall and recognition for verbal material (e.g., Foa, Amir, Gershuny, Molnar and Kozak, 1997, MacDonald, Anthony, MacLeod and Richter, 1997) or for actions and objects (e.g., Ceschi, Van der Linden, Dunker, Perroud and Bredart, 2003, Tolin, Abramowitz, Brigidi, Amir, Street and Foa, 2001). However, other researchers reported that patients diagnosed with OCD showed poorer memory than normal controls (Savage, Deckersbach, Wilhelm, Rausch, Baer, Reid and Jenicke, 2000, Tallis, Pratt and Jamani, 1999, Zitterl, Urban, Linzmayer, Aigner, Demal, Semler and Zitterl-Eglseer, 2001). Data from comparisons of subclinical checkers (SCs) with normal controls (NCs) produced similarly inconsistent results with some reporting poorer memory in subclinical checkers than normal controls and some no differences (Rubenstein, Peynircioğlu, Chambless and Pigott, 1993, Sher, Frost and Otto, 1983, Sher, Mann and Frost, 1984). It should be pointed out that a few studies reported a positive memory bias for threat-relevant information. For instance, Radomsky and Rachman (1999) found that OCD patients had better memory for contaminated objects than clean objects, which was not the case for normal control participants. Similar findings have been reported by Constans, Foa, Franklin, and Mathews (1995) and Radomsky, Rachman, and Hammond (2001).

On the basis of their meta-analysis, Woods, Vevea, Chambless, and Bayen (2002) concluded that checkers were poorer in a number of memory tasks including free recall and cued recall of verbal materials but not in recognition. They suggested that the inconsistencies in the literature might have been due to relatively small sample sizes used in these studies, leading to low power to detect differences between groups. It has to be mentioned that Woods et al.’s (2002) main interest was whether checking per se was related to memory deficiency; therefore, they collapsed data from OCs with that from SCs. This is important because it is possible, for instance, that memory deficits may be limited to patients and may not be true for subclinical checkers. Therefore, the differences seen between normals and checkers in that meta-analysis may be difficult to interpret. Moreover, in general, most of the primary studies were based on comparisons between normal controls and either OCD patients or subclinical checkers.

One of the main aims of the present study was to further explore the link between checking and memory deficit. More specifically, in order to look at whether it is checking per se or the OCD diagnosis that is associated with poorer memory, we compared OC patients with SCs and NCs for neutral and threat-relevant items (further divided into checking-related and contamination-related).

There have been suggestions that a central problem in obsessive–compulsive disorder is at the metamemory level. Most of the studies with OCs and SCs have focused on memory confidence in already recalled information or already performed actions. The results have been fairly consistent. For instance, McNally and Kohlbeck (1993) reported that the checker-OCs were less confident than nonchecker-OCs and NCs regarding their recall performance. Tolin et al. (2001) found that OCD patients with primary checking symptoms were less confident in their long-term memory performance than OCs for whom checking was not the primary symptom. In a similar comparison of OCs with and without checking symptoms, MacDonald et al. (1997) reported that the former group was less confident than the latter for both correct and incorrect answers. In sum, there is substantial evidence that OCs demonstrate a memory confidence problem. Indeed, in their meta-analysis, Woods et al. (2002) found the highest effect size for memory confidence. Whether deficit in memory confidence is specific to the threat domain, however, is unclear (e.g., Constans, Foa, Franklin and Mathews, 1995, Woods, Vevea, Chambless and Bayen, 2002).

The second purpose of the study was to look at whether this metamemory deficit extended to other types of metamemory judgments. More specifically, we investigated the strength and accuracy of feeling-of-knowing (FOK) judgments; that is, people’s predictions about remembering currently unrecallable information in a future task (Hart, 1965, Hart, 1967, Nelson, 1994). In FOK studies, people are asked to recall general knowledge information (e.g., What is the capital of Greece?) or asked to recall recently learned verbal material, such as word pairs. For items they are unable to recall they are asked to provide a FOK judgment on the basis of a question like, “Although I cannot recall the answer right now, would I be able to find it among a number of alternatives?” Participants are generally asked to provide a Likert-type rating and following this judgment phase, they are given a recognition test on all the items. The accuracy of the FOK judgments is, then, measured by the correlation between the FOK ratings and the recognition performance. FOK judgments have been used to investigate metamemory processes in several special populations such as chronic fatigue syndrome patients (Lakein et al., 1997), and Korsakoff patients (Shimamura & Squire, 1986), for whom memory and metamemory complaints are integral elements of the symptomatology.

In sum, in addition to recall and recognition performance, we investigated feeling-of-knowing judgments for verbal material of different emotional valence in OCs, SCs, and NCs.