Implicit and explicit procedural learning in patients recently remitted from severe major depression

Abstract

In acute depression a high prevalence of deficits in learning and memory performance has been reported. Still, it is unclear whether these cognitive deficits are present after remission of clinical symptoms of depression. The present study compared 20 inpatients recently remitted from severe major depressive disorder (MDD) with 20 healthy matched control participants on two sequence learning tasks: a modified serial reaction-time task (SRT) for implicit learning, which is sensitive to subcortical and frontal impairments, and a serial generation task (SGT) for explicit learning. As compared with performance in healthy controls, implicit and explicit learning were not impaired in recently remitted inpatients with depression. Intentional acquisition of new information was related to the severity of depressive symptoms as patients with higher scores on Beck's Depression Inventory (BDI) showed poorer explicit learning. In contrast to findings in acute depression, our results suggest a normal degree of learning in remitted depression; these findings are consistent with unimpaired fronto-striatal functioning. However, although not statistically significant, patients remitted from melancholic MDD revealed poorer implicit learning performance compared with patients remitted from non--melancholic MDD. Longitudinal studies in patients with melancholic vs. non-melancholic MDD are needed to investigate the course of cognitive functioning during the recovery from MDD.

Introduction

Encoding and recognition of information have consistently been reported to be impaired in acute depression (meta-analysis by Zakzanis et al., 1998) and in bipolar depression (e.g. Bearden et al., 2006), but the evidence of cognitive dysfunctions in remitted depression is less clear-cut. An increasing number of studies have reported an unsatisfactory degree of remission in unipolar depression, and moreover, residual symptoms upon remission have been shown to have a strong prognostic value (cf. Fava et al., 2007). Only a few studies have examined cognitive functioning in patients recently remitted from major depressive disorder (MDD); however, a recent prospective study reported low episodic verbal memory performance to be a premorbid marker of depression (Airaksinen et al., 2007).

In this context awareness at retrieval seems to be a critical factor determining the degree of impairment, as depression has been demonstrated to impair effortful processing while interfering only minimally with automatic processing (review by Hartlage et al., 1993). Reber (1967) coined the term ‘implicit learning’ to describe participants' profiting from a probabilistic association between stimuli. This could be measured as a more efficient (faster) responding, i.e. performance increase is concomitant with practice even in the absence of awareness of the learned routine or skill.

To assess implicit memory performance independent of the current state of mood [mood-congruent memory effects as described by Bower (1981) are not reviewed here], neutral verbal stimuli were presented in most studies. Some of these studies reported impaired implicit learning of depressed participants (e.g. Elliot and Greene, 1992, Bradley et al., 1995, Tarsia et al., 2003), while a nearly equal number of studies revealed no significant differences between control subjects and depressed participants (e.g. Danion et al., 1991, Danion et al., 1995, Watkins et al., 1992, Watkins et al., 1996, Ellwart et al., 2003) or individuals with anhedonia (Mathews and Barch, 2006). On the one hand, severity of depressive symptoms might contribute to these inhomogeneous findings, as it is related to cognitive performance (e.g. review by Hartlage et al., 1993). On the other hand, divergent results might reflect different experimental paradigms tapping different aspects of implicit learning, as implicit learning has been demonstrated to be impaired in a conceptual task, but spared in a data-driven perceptual task (Jenkins and McDowall, 2001).

The serial reaction-time task (SRT) introduced by Nissen and Bullemer (1987) allows an assessment of implicit learning without the confounding factor of language skills. Typically, mean reaction-time (RT) in the SRT decreases markedly when random stimulus presentation is replaced by a “hidden” cyclically repeated sequence of stimuli. As the RT savings occur even though participants are not aware of the repeating sequence, improved RTs during sequential trials reflect the degree of implicit sequence learning. Naismith et al. (2006) reported reduced implicit sequence learning in acute depression when implementing a standard SRT paradigm. The lower rates of implicit learning in depressive patients were associated with poorer mental flexibility, slower visuo-motoric speed as well as longer duration of the depressive episode. However, in a modified non-spatial SRT version, no performance deficits in implicit learning were obtained in depressed patients (O'Connor et al., 2005).

The only imaging study implementing a SRT paradigm in (geriatric) depressive patients revealed significantly decreased prefrontal activation and increased striatal activation during explicit sequence learning, but no significant differences in activation during implicit learning compared with controls (Aizenstein et al., 2005). Regarding the behavioral data, the depressive patients were impaired in implicit sequence learning (accuracy), but revealed spared explicit learning. Imaging studies in healthy subjects have consistently shown the basal ganglia, in particular the striatum, in interaction with cortical and cerebellar structures, to be implicated in the non-conscious acquisition of sequences (Rauch et al., 1995, Rauch et al., 1997, Rauch et al., 1998, Berns et al., 1997, Peigneux et al., 2000, Schendan et al., 2003, Thomas et al., 2004). Moreover, the primary motor cortex, the premotor cortex, and the supplementary motor area seem to be important motor-associated components of this cortico-striatal circuit subserving implicit learning in the SRT (Grafton et al., 1995, Grafton et al., 2002, Hazeltine et al., 1997, Exner et al., 2006). The critical role of the cortico-striatal circuitry for SRT performance has also been established in patients with striatal pathology, such as Parkinson's disease (e.g. Pascual-Leone et al., 1993, Doyon et al., 1998).

To our knowledge, this study was the first to investigate implicit and explicit sequence learning in a group of remitted MDD patients. We implemented procedurally equivalent paradigms, a prerequisite to examine a dissociation of implicit and explicit performance. Still, the two conditions might differ with respect to their psychometric properties (cf. Chapman and Chapman, 2001). Additionally, we focussed on the possible contamination of implicit performance by the use of intentional retrieval strategies (Richardson-Klavehn and Bjork, 1988), as participants cannot be prevented from utilizing explicit memory strategies.

The aim of the present study was (1) to examine implicit and explicit sequence learning in recently remitted major depressive patients compared with healthy control subjects, and (2) to investigate whether implicit and explicit learning performance were related to neurocognitive functioning, severity of symptoms or melancholic subtype.