Brain oscillatory correlates of working memory constraints

Abstract

It has been claimed that the coordination of neuronal oscillations differing in frequency is relevant for cognition. However, the validity of this claim has scarcely been investigated. Recent studies revealed that cross-frequency phase coupling and modulations of alpha-power dissociate between retention of relevant and suppression of irrelevant information in visual working memory (WM). We summarize these important results, and discuss possible implications for understanding the neural mechanisms of WM constraints.

Introduction

Mainly based on behavioral assessments, cognitive theories traditionally assume a fixed upper limit for the available storage space in WM (Luck and Vogel, 1997, Miller, 1956). Estimates for capacity limits thereby range from 4 (± 1) (Cowan, 2001, Luck and Vogel, 1997) to 7 (± 2) items (Lisman and Idiart, 1995, Miller, 1956). Recent neuroscientific investigations point to interacting neuronal processes that jointly limit performance in WM tasks (D'Esposito, 2007, Zimmer, 2008). Accordingly, efficient performance on WM tasks requires at least the maintenance of relevant information over a short time-period and, at the same time, inhibition of distracting, non-task relevant information from entering visual awareness (Gazzaley et al., 2005, Vogel et al., 2005).

From the mid 1990s onwards, the so-called ‘change-detection paradigm’ (CD-paradigm, e.g. Luck and Vogel, 1997) has become a popular tool for studying properties of WM in neuroscience. In CD tasks, observers are presented with a memory display that contains a variable amount of information (e.g. colored squares). After a brief retention interval (e.g., around 1 s) a probe display asks the participant to judge whether the currently presented information has changed relative to the representation retained in WM. Since the initial report (Luck and Vogel, 1997), this paradigm has been repeatedly applied to investigate neuronal activity related to WM retention (for a recent review see, Drew et al., 2006).

In a recent development, Vogel and Machizawa (2004) included cues prior to the onset of the memory display to direct the observers' attention to information in only one visual hemifield while simultaneously presenting to-be-ignored information in the uncued hemifield (e.g., Fig. 1A). This procedure enabled the identification of a negative posterior EEG slow-wave that reflects neuronal activity associated with WM retention (Ikkai et al., 2010) and the ability to filter distracting information (Vogel and Machizawa, 2004), the so-called contra-lateral delay activity (CDA). Nevertheless, due to strong interactions between perceptual and attentional demands in CD tasks (Bays and Husain, 2008, Emrich et al., 2009, Fukuda and Vogel, 2009), it remained difficult to separate mechanisms related to item selection from mechanisms related to retention proper.

Although the working memory system is often conceptualized to comprise mechanisms for maintenance and utilization of information (e.g., Baddeley, 1992, Baddeley, 2000), here, we strongly focus on the retention aspect and its corresponding neural correlates. More precisely, the present review article reports on recent evidence for possible oscillatory neuronal correlates of inhibition and retention mechanisms relevant for successful WM performance as derived from MEG/EEG research (for a recent review on physiological studies please refer to Duzel et al., 2010). In the following, we specifically focus on studies relating the amplitude of alpha oscillations to inhibition of distracting information and the synchronized co-activation of low-frequency theta and high-frequency beta and gamma oscillations to WM-maintenance. Finally, the two phenomena reported in recent WM-studies are then related to an intermediate model of working memory constraints.