Emotion/cognition-coupling in word recognition memory of depressive patients: an event-related potential study

Abstract

Depressive patients show deficits in memory functions. However, the underlying mechanisms remain unclear. Experiments with a special emphasis on the link between emotion and cognition appear challenging. The aim of this study is to investigate the influence of the emotional content of words on memory in non-medicated depressive patients (n=11) compared with a control group (n=11) utilizing event-related brain potentials (ERPs). In a continuous word recognition paradigm brain responses to repeated items are characterized by more positive waveforms of ERPs. This recognition effect (‘old/new effect’) has been shown to be sensitive to parameters relevant for memory processing. For the purpose of this ERP experiment visually presented words were classified into three different categories of emotional content. The ERPs for the correctly detected ‘old’ (repeated) words showed an increased positivity beginning approximately 250 ms post-stimulus, concurring with a good recognition performance. In addition, old/new effect and behavioral data were sensitive to words’ different emotional connotations in the control group. In contrast, the depressive patients performed worse and showed no significant old/new effect. Nevertheless, their recognition performance was also enhanced by the emotional content. Furthermore, a differential effect of the emotional content on frontal ERPs was found between groups. In contrast to the control group, a reduced old/new effect indicates a reduced working memory capacity in the moderately depressed patients. This is suggested to be partially due to changes of the emotion/cognition coupling related to ruminations with preferably negative emotional connotation. However, the emotional content also affects recognition performance in the depressive patients.

Introduction

In addition to a lowering of mood, a dominance of negative self-perceptions and thoughts, and reduced psychomotor abilities and motivation, depressive patients frequently complain about diminished memory functions. These subjective accounts have been supported by a number of neuropsychological studies showing that depression may be associated with memory deficits (Robbins et al., 1992 for review; Burt et al., 1995, Bradley et al., 1995, Ilsley et al., 1995, McKay et al., 1995, Calev, 1996, Elliott et al., 1996, Kindermann and Brown, 1997, Purcell et al., 1997, Kalska et al., 1999). However, studies of younger patients with depression have been equivocal, as several studies have failed to find major cognitive deficits in patients whose mean age ranged between 30 and 40 years (DeLuca et al., 1995, Albus et al., 1996, Purcell et al., 1997).

Bower, 1981, Bower, 1983 reported that emotional state may influence performance in recognition tasks, with subjects showing an enhanced performance for stimulus material which is congruent with their mood. He also found a close correlation between the recognition performance and the corresponding emotional states during learning and retrieval. Consistent with this, depressive patients have been shown to have increased recall of depression-relevant words (Bradley et al., 1995, Neshat-Doost et al., 1998). However, these findings were not supported by other investigators (Danion et al., 1995, Calev, 1996). Furthermore, it remains unclear to what extent memory impairment in depression results from specific memory deficits, or more generalized cognitive, e.g. attentional, impairments or emotional factors.

Most of these prior studies have concentrated on the observable components of information processing by measuring verbal and behavioral reactions. While these methods are immensely useful, their limitations must be recognized. First, observable output measures are open to subjects’ biases which are extraneous to the experimental variables. Second, relying solely on inferences from what can be observed to describe the contents and operations of the mind ignores the underlying physiological processes themselves.

One method to assess the brain electricial activity associated with cortical information processing is the recording of event-related brain potentials (ERPs). ERPs are minute voltage fluctuations which can be recorded non-invasively from the intact human scalp occurring in response to stimulus events and cognitive processes (Hillyard and Picton, 1987, Halgren, 1990). Consequently, ERPs have been shown to be sensitive to memory processes (Johnson, 1995, Rugg, 1995a, Rugg, 1995b for review) and to be sensitive to emotional factors on cognitive processing, e.g. face recognition (Begleiter et al., 1983, Johnston et al., 1986, Johnston et al., 1987, Vanderploeg et al., 1987, Lang et al., 1990, Laurian et al., 1991, Pfleger et al., 1993, Halgren and Marinkovic, 1995, Mini et al., 1996, Ito et al., 1998, Orozco and Ehlers, 1998, Oliver-Rodriguez et al., 1999). In addition, a frontal positivity starting at approximately 250 ms post-stimulus has been described for emotional stimuli (Naumann et al., 1993, Naumann et al., 1997) which was more pronounced when subjects were instructed to focus on the emotional connotations of the stimuli.

Explicit memory processes for words have been examined by presenting and repeating words after some intervening items with the subjects’ task being to discriminate between newly presented and repeated words. Rugg and Nagy (1989) and Rugg and Doyle (1992) have shown that under these conditions words evoke a series of negative and positive ERP components which are more positive for repeated words beginning approximately 250 ms after word presentation involving most parts of the scalp. This effect is partially caused by a modification of the negative N400 component and the positive LPC (late positive component, P3b), but has been termed the ‘old/new effect’ to avoid premature identification of the effect with other ERP components (Rugg, 1995a, Rugg, 1995b, Rugg et al., 1996, Rugg et al., 1997).

This old/new effect appears to be influenced by the ‘conscious recollection’ of the stimuli, e.g. of words (Paller and Kutas, 1992, Paller et al., 1995, Rugg, 1995a, Rugg, 1995b, Rugg et al., 1996, Wilding and Rugg, 1996, Wilding and Rugg, 1997), and their frequency of occurrence (Rugg and Doyle, 1992, Rugg and Doyle, 1994). In addition, Fernandez et al. (1998) described a subsequent memory effect between 300 and 700 ms post-stimulus (old/new effect) to be more directly related to episodic memory, whereas a ‘second effect’ (between 700 and 1100 ms) was modulated by the word frequency and thus appeared to be related to processes ‘influencing episodic encoding success indirectly’. Rugg et al., 1998a, Rugg et al., 1998b interpreted the old/new effect to partly reflect implicit memory from 300 ms to 500 ms and explicit memory function later than 500 ms after stimulus onset. Furthermore, Halgren and Marinkovic (1995), Johannes et al. (1999) and Dietrich et al. (2000) demonstrated that normal subjects show an old/new effect for emotionally negative and positive words which is significantly different from that for neutral words.

A number of studies have demonstrated changes of ERP components in depressive patients, e.g. a reduction of the P300-amplitude or a prolonged P300 latency in ‘oddball’ paradigms, which appeared to be rather unspecific (Nandrino et al., 1996). In contrast, Blackburn et al. (1990) modified these ERP experiments to enhance the specificity with regard to depression, and demonstrated that depressive patients exhibit a smaller amplitude of the P300 in response to negatively toned words than to positive words. This was interpreted as being a result of the negative cognitive schemata of the depressive patients (Beck, 1967, Beck et al., 1979, Blaney, 1986) who are predominantly orientated to the processing of negative stimuli, resulting in a higher expectancy of negative stimuli and hence a smaller amplitude of the P300 (P3b).

Based on these results we identified ERPs as a tool to examine memory processing in depressive patients with a special emphasis on electrophysiological correlates of the emotion/cognition coupling, the link between emotional and cognitive aspects of information processing (LeDoux, 1989, LeDoux, 1995, Rolls, 1995). Therefore, we used a previously established ERP paradigm (Johannes et al., 1999, Dietrich et al., 2000) to investigate the influence of the emotional content of words on cognitive processing in non-medicated depressive patients and controls in a continuous word-recognition ERP experiment.

We postulated the old/new effect in the depressive patients would be reduced due to impaired memory functions, and that the emotional connotation of the words would disclose a differential effect on the ERPs due to the specific disturbance of the emotional and cognitive processing in depressives mentioned above.