Common and unique components of inhibition and working memory: An fMRI, within-subjects investigation

Abstract

Behavioural findings indicate that the core executive functions of inhibition and working memory are closely linked, and neuroimaging studies indicate overlap between their neural correlates. There has not, however, been a comprehensive study, including several inhibition tasks and several working memory tasks, performed by the same subjects. In the present study, 11 healthy adult subjects completed separate blocks of 3 inhibition tasks (a stop task, a go/no-go task and a flanker task), and 2 working memory tasks (one spatial and one verbal). Activation common to all 5 tasks was identified in the right inferior frontal gyrus, and, at a lower threshold, also the right middle frontal gyrus and right parietal regions (BA 40 and BA 7). Left inferior frontal regions of interest (ROIs) showed a significant conjunction between all tasks except the flanker task. The present study could not pinpoint the specific function of each common region, but the parietal region identified here has previously been consistently related to working memory storage and the right inferior frontal gyrus has been associated with inhibition in both lesion and imaging studies. These results support the notion that inhibitory and working memory tasks involve common neural components, which may provide a neural basis for the interrelationship between the two systems.

Introduction

Working memory and inhibition, described as core executive function domains (Goldman-Rakic, 1987; Hasher & Zacks, 1988), are closely related. Both have been linked to IQ (Horn, Dolan, Elliott, Deakin, & Woodruff, 2003), and performance is associated in both normal (e.g. Conway, Cowan, & Bunting, 2001; Kane and Engle, 2000, Kane and Engle, 2003; Unsworth, Schrock, & Engle, 2004), and clinical populations such as attention deficit hyperactivity disorder (ADHD) (Hervey, Epstein, & Curry, 2004; erté, Geurts, Roeyers, Oosterlaan, & Sergeant, 2006; Willcutt, Doyle, Nigg, Faraone, & Pennington, 2005), high-functioning autism (HFA) and Tourette syndrome (TS) (Verté et al., 2006). It has also been suggested that they may rely on common neural resources (De Fockert, Rees, Frith, & Lavie, 2001; Kane & Engle, 2003). The present study represents an investigation into the extent to which the neural correlates of different forms of inhibition and working memory overlap, and where such commonalities occur.

The right inferior frontal cortex (BA 45/47) has been described as showing the most robust common activation across inhibition tasks (Bunge, Dudukovic, Thomason, Vaidya, & Gabrieli, 2002) and identified as being central to inhibitory control (e.g. Aron, Robbins, & Poldrack, 2004; Chambers et al., 2006; Kelly, Hester, Foxe, Shpaner, & Garavan, 2006). It has been reported to show significantly greater activation for no-go trials in which participants inhibit a preponent response compared to go trials in which participants make a preponent response (Booth et al., 2003; Konishi, Nakajima, Uchida, Sekihara, & Miyashita, 1998; Langenecker & Nielson, 2003), has been strongly implicated in the inhibition of an already initiated manual response, the so-called stop task (Aron, Fletcher, Bullmore, Sahakian, & Robbins, 2003; Aron & Poldrack, 2006; Chambers et al., 2006), and has been described as part of a “shared inhibitory neurocognitive network” involved with both go/no-go (GNG) and stop tasks (Rubia et al., 2001). In a flanker task activity in this region has been found to accompany increases in reaction time associated with incongruent trials (Hazeltine, Poldrack, & Gabrieli, 2000), with the authors arguing that the strong correspondence between this activity and that reported in GNG and set shifting studies suggests that response inhibition may most appropriately characterize the function of this region.

Inferior parietal activation has also been observed in a number of inhibition tasks (Garavan, Ross, Murphy, Roche, & Stein, 2002; Garavan, Ross, & Stein, 1999; Langenecker & Nielson, 2003; Liddle, Kiehl, & Smith, 2001; Peterson et al., 2002; Rubia et al., 2001; Sylvester et al., 2003), although this may be related to the storage of stimulus-response representations rather than inhibition (Hester, Murphy, & Garavan, 2004). Similarly, the dorsolateral prefrontal cortex (DLPFC) activation that has been associated with a range of inhibition tasks (Bellgrove, Hester, & Garavan, 2004; Garavan et al., 2002; Langenecker & Nielson, 2003; Liddle et al., 2001, Sylvester et al., 2003; Van Veen, Cohen, Botvinick, Stenger, & Carter, 2001) might be attributed to working memory demands as significant right DLPFC activation was observed for GNG trials during a counting GNG task, but not during a simple GNG (Mostofsky et al., 2003).

Furthermore, although a relatively consistent set of regions have been implicated in response inhibition paradigms (Wager et al., 2005), different forms of inhibition may be involved in different tasks (Wager et al., 2005). Within this study we adopt the approach taken by Barkley (1997) and consider three inhibition processes (rather than mechanisms); (1) inhibition of an initial prepotent response, (2) stopping an ongoing response or delayed responding, and (3) limiting interference or distractibility. These processes are employed to a different extent in different tasks (Rubia et al., 2001), for example whilst the Stroop and Erikson flanker tasks may rely heavily on limiting interference, GNG and stop tasks are likely to involve a greater reliance on inhibition of a preponent or previously initiated response (although it should be noted that such conflict resolution tasks may well involve other mechanisms besides inhibition, such as facilitation). Correlations between performance on different inhibitory tasks have generally been low (Wager et al., 2005), but sometimes significant (Fan, Flombaum, McCandliss, Thomas, & Posner, 2003; Miyake, Friedman, Emerson, Witzki, & Howerter, 2000). Although it has been suggested that idiosyncrasies between tasks may obscure the results and account for the low behavioral correlations (Friedman & Miyake, 2004), it is also possible that different mechanisms may be involved in different forms of response inhibition. Therefore, when investigating the commonalities between the neural correlates of working memory and inhibition within the present study, three inhibition tasks were used, corresponding to the three inhibitory processes described by Barkley (1997).

Reviews of the working memory literature have implicated the left inferior frontal cortex, DLPFC, premotor cortex, superior frontal cortex, supplementary motor area, and the parietal cortex (for example, Cabeza & Nyberg, 2000; D’Esposito et al., 1998; D’Esposito, Postle, & Rypma, 2000; Smith and Jonides, 1998, Smith and Jonides, 1999). Working memory tasks that use stimuli from different sensory modalities have been shown to activate overlapping cortical regions, in both prefrontal and parietal cortex (Klingberg, 1998; Klingberg, Kawashima, & Roland, 1996), suggested to be supramodal working memory regions. However, a dissociation has also been reported whereby verbal working memory is primarily associated with left hemisphere regions (in particular the left prefrontal cortex (Gabrieli, Brewer, & Poldrack, 1998)) and spatial working memory is linked to right hemisphere regions, even when the same letter stimuli are used in verbal and spatial working memory task conditions (Smith, Jonides, & Koeppe, 1996). Furthermore, Leung and Zhang (2004) suggested that different subsets of the working memory system may be associated with interference resolution in spatial and verbal domains. It is claimed that whereas the left inferior frontal gyrus is involved in overcoming interfering verbal stimuli, the right precentral sulcus and superior parietal lobe are involved in overcoming interfering spatial stimuli. It is also possible that the cortical representation of interference resolution in working memory may be material specific, which could mean that the relationship between inhibition and working memory differs depending upon whether the working memory task (and/or stimulus) is verbal or spatial. For this reason both verbal and spatial working memory tasks were included within this study.

Evidence from patients suggests that the same right inferior frontal region may be involved in both spatial working memory and response inhibition in a stop task, with correlations reported between damage to this region and both stop signal reaction time (SSRT) and spatial working memory (SWM) performance) (Clark et al., 2007). In order to identify commonalities between inhibition and working memory, in healthy participants, one approach has been to draw upon the results of different studies. However, group differences may confound such comparisons. Another approach has been to combine working memory and inhibition demands within the same task, for example Bunge, Ochsner, Desmond, Glover, and Gabrieli (2001) looked at the effects of manipulating the level of proactive interference in a Sternberg-type working memory task, and Kelly et al. (2006) used a task that required inhibition of prepotent responses based on the contents of working memory. However, Hester et al. (2004) reported that maintaining successful inhibitory control under increasing working memory demands tended not to increase activation in overlapping regions, but predominantly in unique inhibition-specific regions. In the present study, working memory and inhibition processes were isolated using different task blocks to avoid confounds associated with the manipulation of inhibition and working memory demands within the same task.

Valid conjunction analysis (Nichols, Brett, Andersson, Wager, & Poline, 2004) was used to localize overlapping activation associated with different forms of inhibition and working memory in the same subjects (to reduce the effects of individual differences). For each task, event related analysis was used to avoid possible confounds associated with using a block design (as described by Aron & Poldrack, 2005), for example unbalanced task maintenance load and switching between blocks which do not include trials that require inhibition and mixed blocks which do include such trials.

Another way in which the present study represents an extension of the approach taken previously is in the design of the GNG and stop tasks. Trials involving inhibition (no-go or stop trials) have typically been compared to go trials in which a preponent stimulus is presented and a preponent response is required. Such an approach may introduce confounds associated with differences in the frequency of presentation of certain stimuli (Aron & Poldrack, 2005), with the stimuli associated with inhibition being presented less frequently than the stimuli that do not involve inhibition. To overcome such confounds this study made use of oddball trials, which were control trials that did not require response inhibition, but were matched to no-go trials in terms of frequency of presentation. By comparing no-go or stop trials to these oddball trials we controlled for differences in familiarity associated with the oddball effect, whereby novel stimuli elicit additional cerebral activity (Halgren, Marinkovic, & Chauvel, 1998). Although such oddball trials may require other forms of inhibition, they do not require inhibition of a response, enabling us to isolate this form of inhibition in contrasts between no-go or stop trails and oddball trials.

Five contrasts were generated: three from inhibition tasks (corresponding to the three different inhibitory processes described by Barkley, 1997) and two from working memory tasks (verbal and spatial), and valid conjunction analysis was used to identify regions of common activation within and between working memory and inhibition domains. The results of the within domain conjunctions were also used to generate ROIs which were used to perform small volume corrected conjunction analyses between domains. In this way we were able to test the hypothesis that there are common regions associated with response inhibition and working memory, observe the extent to which each ROI contributes to each task, and determine whether the choice of inhibition and working memory task influences such commonality. This study represents an extension of the approach by investigating commonalities between three different forms of inhibition and two different forms of working memory and the relative contribution of each ROI to each task, in the same participants, isolating working memory and inhibition within separate task blocks, and using oddball trials in order to control for differences in familiarity within GNG and stop task contrasts.