Dorsolateral prefrontal contributions to human working memory

Abstract

Although neuroscience has made remarkable progress in understanding the involvement of prefrontal cortex (PFC) in human memory, the necessity of dorsolateral PFC (dlPFC) for key competencies of working memory remains largely unexplored. We therefore studied human brain lesion patients to determine whether dlPFC is necessary for working memory function, administering subtests of the Wechsler Memory Scale, the Wechsler Adult Intelligence Scale, and the N-Back Task to three participant groups: dlPFC lesions (n = 19), non-dlPFC lesions (n = 152), and no brain lesions (n = 54). DlPFC damage was associated with deficits in the manipulation of verbal and spatial knowledge, with left dlPFC necessary for manipulating information in working memory and right dlPFC critical for manipulating information in a broader range of reasoning contexts. Our findings elucidate the architecture of working memory, providing key neuropsychological evidence for the necessity of dlPFC in the manipulation of verbal and spatial knowledge.

Introduction

Working memory comprises a system for maintaining, monitoring and manipulating information in short-term memory, providing the interface between perception, long-term memory and action that enables goal-directed behavior (Baddeley, 1998; Baddeley and Petrides, 1996). Although cognitive neuroscience has made remarkable progress in understanding the involvement of the prefrontal cortex (PFC) in human memory, fundamental questions remain regarding the functional organization of the PFC with respect to working memory. One unresolved issue concerns whether subregions within the lateral PFC mediate functionally distinct processes or instead serve a common role in working memory. Anatomically, the lateral PFC consists of multiple subregions that differ in cytoarchitecture and connectivity (Petrides et al., 2012), raising the possibility that these subregions may guide goal-directed behavior through different mechanisms.

A seminal and longstanding debate in cognitive neuroscience has examined this issue, investigating alternative models for understanding the functional organization of the lateral PFC and its role in working memory. Domain-general models posit that the lateral PFC is functionally organized according to the type of working memory operations engaged, with the dorsolateral PFC (dlPFC) embodying computational mechanisms for monitoring and manipulating items in working memory (Owen et al., 1996; Duncan and Owen, 2000; Miller and Cohen, 2001; Koechlin et al., 2003; Petrides, 2000, 2005; Petrides et al., 2012). Monitoring operations are thought to support the active retention of information in working memory and computational mechanisms for manipulating items are recruited for updating (Petrides, 2000) or selecting between these representations (Rowe et al., 2000). In contrast, domain-specific models posit that the lateral PFC is functionally organized according to the domain of information processed. Advocates of this framework propose that dlPFC is functionally specialized to process visuospatial information in working memory, enabling mental representations of coordinates within the spatial domain (Awh et al., 1995; Butters and Pandya, 1969; Butters et al., 1971, 1972; Courtney et al., 1998, 1996, 1997; Goldman-Rakic, 1995; Levy and Goldman-Rakic, 1999; Smith and Jonides, 1999).

The empirical case advanced in support of each model of dlPFC function has relied primarily upon (1) lesion studies in non-human primates demonstrating reliable deficits in working memory due to unilateral dlPFC lesions (Butters and Pandya, 1969; Butters et al., 1971, 1972; Jacobsen and Nissen, 1937; Levy and Goldman-Rakic, 1999) and (2) functional neuroimaging studies in humans reporting activity within the dlPFC for tests of working memory [for meta-analytic reviews, see (Owen et al., 2005; Wager et al., 2004; Wager and Smith, 2003)]. Two key findings from studies of non-human primates performing delayed-response tasks suggest a crucial role for the dlPFC in working memory. First, experimental lesions of the principal sulcus in the dlPFC cause delay-dependent impairments, whereby forgetting increases with the length of the delay (Miller and Orbach, 1972; Bauer and Fuster, 1976; Funahashi et al., 1993). Second, neurophysiological unit recordings from the dlPFC often show persistent, sustained levels of neuronal firing during the retention interval of delayed-response tasks (Funahashi et al., 1989; Fuster and Alexande, 1971; Kubota and Niki, 1971). This sustained activity is thought to provide a bridge between the stimulus cue (e.g., the location of a flash of light) and its contingent response (e.g., a saccade to the remembered location). Such data established a strong link implicating the dlPFC as a crucial node supporting working memory.

Conclusions drawn from these literatures, however, are characterized by the following well-known limitations. First, the precise localization of working memory functions cannot be directly transposed from monkeys to humans due to significant interspecies macroscopic anatomical differences (Petrides et al., 2012). Second, functional neuroimaging (fMRI) studies apply correlational methods and therefore cannot formally demonstrate whether dlPFC is necessary for working memory or instead serves an accessory role (Sarter et al., 1996). As a consequence, the precise localization of working memory function in humans and the contribution of dlPFC to the neural systems underlying working memory remain controversial.

In recent years, lesion studies in humans (Baldo and Dronkers, 2006; D'Esposito and Postle, 1999; D'Esposito et al., 2006; Muller et al., 2002; Ptito et al., 1995; Tsuchida and Fellows, 2009; Volle et al., 2008) and repetitive transcranial magnetic stimulation (rTMS) experiments (Hamidi et al., 2009, 2008; Koch et al., 2005; Postle et al., 2006) have provided key evidence to inform the debate. Human lesion and rTMS research are able to overcome the methodological limitations of earlier non-human primate and functional neuroimaging studies by investigating the anatomical localization of working memory functions in the human brain (Rorden and Karnath, 2004) and evaluating the necessity of the dlPFC for specific components of working memory.

Findings from the contemporary literature, however, have been equivocal, with some investigators reporting specific patterns of working memory deficits (Baldo and Dronkers, 2006; Mottaghy et al., 2002; Ptito et al., 1995; Tsuchida and Fellows, 2009; Volle et al., 2008) and others failing to observe reliable impairment (D'Esposito and Postle, 1999; D'Esposito et al., 2006; Hamidi et al., 2008; Koch et al., 2005; Muller et al., 2002). Difficulty in interpreting the theoretical significance of these findings has resulted from (1) the often diffuse (rather than focal) lesions observed, (2) the lack of comparison subjects carefully matched for pre- and post-injury performance measures, and (3) the limited scope of working memory functions examined. The absence of such data represents a substantial gap in the understanding of both dlPFC function and the neural substrates of working memory. Here, we characterize key competencies of working memory function in a sample of patients with focal brain lesions involving dlPFC.