Dissociating the effects of Sternberg working memory demands in prefrontal cortex

https://doi.org/10.1016/j.pscychresns.2006.08.002Get rights and content

Abstract

Earlier neuroimaging studies of working memory (WM) have demonstrated that dorsolateral prefrontal cortex (DLPFC) activity increases as maintenance and load demand increases. However, few studies have carefully disambiguated these two WM processes at the behavioral and physiological levels. The objective of the present functional resonance imaging (fMRI) study was to map within prefrontal cortex locales that are selectively load sensitive, delay sensitive, or both. We studied 18 right-handed normal subjects with fMRI at 3 Tesla during a block design version of the Sternberg task. WM load was manipulated by varying the memory set size (3, 5, or 8 letters). The effect of memory maintenance was examined by employing two time delays (1 s and 6 s) between the letter set and probe stimuli. The DLPFC was strongly activated in load manipulation, whereas activation as a function of delay was restricted to the left premotor regions and Broca's areas. Moreover, regions of prefrontal cortex on the right (BA 46) were found to be exclusively affected by load. These results suggest the possibility that top–down modulation of attention or cognitive control at encoding and/or decisionmaking may be mediated by these areas.

Introduction

Working memory (WM) refers to both maintenance of information for a short period of time and to online manipulation of information necessary for higher cognitive functions, such as language, reasoning and problem solving (Baddeley, 1986). Single cell recording studies in non-human primates have demonstrated that specific classes of pyramidal glutamatergic neurons in prefrontal cortex participate in maintenance and manipulation of stimuli during WM tasks (for a review, see Goldman-Rakic, 1999).

Functional neuroimaging studies in humans (McCarthy et al., 1996, Fiez et al., 1996, Courtney et al., 1997, Courtney et al., 1998, D'Esposito et al., 1998, D'Esposito et al., 1999, Zarahn et al., 1999, Haxby et al., 2000, Fletcher and Henson, 2001, Leung et al., 2002) have reported results consistent with findings in non-human primates, suggesting that the DLPFC may be involved in the active maintenance of information. Recent neuroimaging studies have replicated and expanded these findings suggesting that DLPFC activity increases as maintenance demand increases by increasing the length of the retention interval (Haxby et al., 1995, McIntosh et al., 1996, Barch et al., 1997, Grady et al., 1998, Kruggel et al., 2000). Other neuroimaging studies have reported that DLPFC activity increases as load increases by varying the amount of information held in memory (Braver et al., 1997, Jonides et al., 1997, Cohen et al., 1997, Jansma et al., 2000). However, several of the cognitive paradigms used in those studies did not allow separation of these two sub-processes involved in WM. For this reason, other neuroimaging studies have used the Sternberg task, which allows easier disambiguation of encoding, maintenance and retrieval with the possibility of varying the cognitive load of each of these sub-processes.

Block design fMRI studies, using the Sternberg task, have reported increasing activity in DLPFC correlated with memory load increase (Manoach et al., 1997, Rypma et al., 1999, Bunge et al., 2001, Glahn et al., 2002, Druzgal and D'Esposito, 2003). More recent event-related fMRI studies have tried to identify the specific task components (encoding, delay, and retrieval phase) that were sensitive to load manipulation (D'Esposito et al., 2000, Rypma, 2006). These studies have reported mixed results. Rypma and D'Esposito (1999) found that the DLPFC was involved during the encoding phase. Other studies have reported increases in activation of the DLPFC during maintenance (Postle et al., 1999, Leung et al., 2002, Rypma et al., 2002, Cairo et al., 2004, Zarahn et al., 2005, Narayanan et al., 2005, Habeck et al., 2005). A number of studies have also reported DLPFC activity in response to an increase of memory load during retrieval (Rypma and D'Esposito, 2000, Rypma et al., 2002, Veltman et al., 2003, Leung et al., 2004, Cairo et al., 2004) or during all task phases (Bedwell et al., 2005).

Taken together, these findings indicate that the DLPFC, previously thought to be important for executive processes, may be recruited as a function of both increasing load and delay. The aim of this functional resonance imaging (fMRI) study was to examine the relationship between prefrontal regions and WM processes associated with increasing delay and load. We used a block design to examine whether load and delay manipulations rely on common or distinct brain regions within the Sternberg delay-recognition paradigm. The design involved two experimental manipulations. The first allowed exploration of whether activity in the DLPFC increases as a function of working memory load. The second manipulation, obtained by increasing delay, allowed us to address the question of whether activity in the DLPFC increases as a function of delay, even when working memory load is maintained constant.

We reasoned that if the sustained activation during maintenance in DLPFC increases as a function of delay, then one might expect that the activity in the long delay condition should be greater than in the short delay condition of the task. Moreover, DLPFC activity might also increase with the number of items to be remembered. Thus, the DLPFC may not only operate in a load-dependent and delay-dependent manner, but may demonstrate an interaction between load and delay that might lead to a significant increase in activation. If indeed both experimental factors affect WM demands, then one would expect interactive effects, such that the effects of increased demand in one factor would be enhanced at increased levels of demand in the other factor. Although common DLPFC circuits are implicated by increased load and delay demands, we predicted that load and delay manipulations might differ in the extent to which they recruited different pools of neurons in the same or adjacent regions. To our knowledge, this is the first study to compare maintenance and load manipulation in the same group of subjects, so that commonly activated areas as well as differential activations can be identified.

Section snippets

Subjects

The sample comprised of 18 right-handed volunteers (11 males; mean age = 27.4 years). We ensured that all of our subjects were right-handed via self-reports. Moreover, handedness was ascertained by observing hand use during task performance.

All participants were screened for medical, neurological, and psychiatric illnesses, and for use of prescription medications. The study was conducted according to the guidelines of the internal review board at the National Institute of Mental Health. All

Accuracy

All subjects performed significantly above chance (> 70%) at each mnemonic load (Table 1). The data were subjected to repeated measures ANOVA. The subjects' accuracy level decreased with increasing load in the short delay (F(2,32) = 31.34, P < 0.001) and long delay versions of the task (F(2,32) = 5.21, P < 0.01) (Fig. 2).

There was no significant difference in accuracy as a function of short and long delays (F(1,32) = 0.37, P = 0.5). Furthermore, there was no significant interaction between load and delay in

Discussion

The results of the present study indicate that increasing load demands, in the short and long delay versions of the task, were associated with activation of a cerebral network including bilateral DLPFC, inferior prefrontal cortex, and parietal cortex. Significant activation was found in the left supplementary motor area, lateral premotor area and Broca's area as a function of delay manipulation. Each of these regions is consistent with the well-defined neural architecture of WM as published in

Conclusion

In conclusion, we have demonstrated that the prefrontal cortex is differentially engaged by increasing working memory demands. Differentiating the neural underpinnings of WM sub-processes enables greater understanding of WM deficits found in frontal patients (e.g., schizophrenia). The presence of right prefrontal areas activated under conditions of increasing load also suggests the possibility that top–down modulation of attention or cognitive control at encoding and/or decision making may be

Acknowledgements

We thank Brad Zoltick and Saumitra Das for their excellent technical assistance. We are very grateful to Jessica Cohen for her help.

References (53)

  • C.L. Grady et al.

    Age-related changes in regional cerebral blood flow during working memory for faces

    NeuroImage

    (1998)
  • C. Habeck et al.

    An event-related fMRI study of the neural networks underlying the encoding, maintenance, and retrieval phase in a delayed-match-to-sample task

    Brain Research. Cognitive Brain Research

    (2005)
  • J.V. Haxby et al.

    Distinguishing the functional roles of multiple regions in distributed neural systems for visual working memory

    NeuroImage

    (2000)
  • G.D. Honey et al.

    Effects of verbal working memory load on corticocortical connectivity modeled by path analysis of functional magnetic resonance imaging data

    Neuroimage

    (2002)
  • J.M. Jansma et al.

    Specific versus nonspecific brain activity in a parametric N-back task

    NeuroImage

    (2000)
  • F. Kruggel et al.

    Nonlinear regression of functional MRI data: an item recognition task study

    NeuroImage

    (2000)
  • D.S. Manoach et al.

    Identifying regional activity associated with temporally separated components of working memory using event-related functional MRI

    NeuroImage

    (2003)
  • B. Rypma

    Factors controlling neural activity during delayed-response task performance: testing a memory organization hypothesis of prefrontal function

    Neuroscience

    (2006)
  • B. Rypma et al.

    Load-dependent roles of frontal brain regions in the maintenance of working memory

    NeuroImage

    (1999)
  • D.J. Veltman et al.

    Maintenance versus manipulation in verbal working memory revisited: an fMRI study

    NeuroImage

    (2003)
  • E. Zarahn et al.

    Temporal isolation of the neural correlates of spatial mnemonic processing with fMRI

    Brain Research. Cognitive Brain Research

    (1999)
  • E. Awh et al.

    Dissociation of storage and rehearsal in verbal working memory

    Psychological Science

    (1996)
  • A. Baddeley

    Working Memory

    (1986)
  • J.S. Bedwell et al.

    Fuctional neuroanatomy of subcomponent cognitive processes involved in verbal working memory

    International Journal of Neuroscience

    (2005)
  • S.A. Bunge et al.

    Prefrontal regions involved in keeping information in and out of mind

    Brain

    (2001)
  • J.D. Cohen et al.

    Temporal dynamics of brain activation during a working memory task

    Nature

    (1997)
  • Cited by (64)

    • Prefrontal cortex activation during working memory task in schizophrenia: A fNIRS study

      2021, Asian Journal of Psychiatry
      Citation Excerpt :

      The apparent inconsistency in the DLPFC activation during the WM task in SCZ as measured with functional neuroimaging has been explained based on task memory load and task efficiency (Potkin et al., 2009). DLPFC activation has a positive correlation with task memory load and performance until the WM load exceeds the WM capacity (Altamura et al., 2007). Once the WM load exceeds the WM capacity, DLPFC activation declines (Goldberg et al., 1998).

    • Meta-analyses of the n-back working memory task: fMRI evidence of age-related changes in prefrontal cortex involvement across the adult lifespan

      2019, NeuroImage
      Citation Excerpt :

      Functional brain correlates of working memory capacity have been investigated by tasks that manipulate task complexity. Typically, task complexity is manipulated by either increasing the number of display items to be processed (e.g., Sternberg task; Altamura et al., 2007 and colour matching task; Arsalidou et al., 2013a) or by increasing the time interval between a sample stimulus and comparison stimuli – delayed-match-to-sample task (e.g., Simons et al., 2006; Picchioni et al., 2007; Höller-Wallscheid et al., 2017). The Sternberg task (Sternberg, 1966) requires participants to indicate whether one item out of a larger set of items that can vary from 1 to 7 was present in the original set.

    View all citing articles on Scopus
    View full text