Elsevier

Neurobiology of Aging

Volume 57, September 2017, Pages 18-27
Neurobiology of Aging

Regular article
Prefrontal-parietal effective connectivity during working memory in older adults

https://doi.org/10.1016/j.neurobiolaging.2017.05.005Get rights and content

Abstract

Theoretical models and preceding studies have described age-related alterations in neuronal activation of frontoparietal regions in a working memory (WM) load-dependent manner. However, to date, underlying neuronal mechanisms of these WM load-dependent activation changes in aging remain poorly understood. The aim of this study was to investigate these mechanisms in terms of effective connectivity by application of dynamic causal modeling with Bayesian Model Selection. Eighteen healthy younger (age: 20–32 years) and 32 older (60–75 years) participants performed an n-back task with 3 WM load levels during functional magnetic resonance imaging (fMRI). Behavioral and conventional fMRI results replicated age group by WM load interactions. Importantly, the analysis of effective connectivity derived from dynamic causal modeling, indicated an age- and performance-related reduction in WM load-dependent modulation of connectivity from dorsolateral prefrontal cortex to inferior parietal lobule. This finding provides evidence for the proposal that age-related WM decline manifests as deficient WM load-dependent modulation of neuronal top-down control and can integrate implications from theoretical models and previous studies of functional changes in the aging brain.

Introduction

Aging is associated with a decline in working memory (WM) as indicated by a large number of studies comparing WM performance between younger and older adults (for reviews, see Craik and Salthouse, 2011, Verhaeghen and Cerella, 2002). In tasks including different levels of WM load, age-related performance decrements were found to be most apparent at high WM load (e.g., Nagel et al., 2011, Nyberg et al., 2009, Schneider-Garces et al., 2010). For several years, WM performance in older adults has been studied during functional neuroimaging, in particular using functional magnetic resonance imaging (fMRI), to examine neural correlates of the age-related decline in WM. Several earlier studies have reported pronounced activation mainly in frontoparietal areas during verbal WM tasks in older adults compared to younger adults (Cabeza et al., 2004, Logan et al., 2002, Reuter-Lorenz et al., 2000). However, when reviewing the literature, both hyper- and hypoactivations in frontoparietal regions including dorsolateral prefrontal cortex (DLPFC), lateral premotor cortex (LPMC), and inferior parietal lobule were detected during WM in older participants compared to their younger counterparts (Rajah and D'Esposito, 2005). It has been suggested that part of these conflicting results could be integrated by taking WM load into account; thus, hyperactivations were mainly found at relatively low WM load, whereas hypoactivations were most frequently reported at high WM load (Heinzel et al., 2014a, Nagel et al., 2011, Schneider-Garces et al., 2010). To replicate such previous findings and to confirm prior region of interest (ROI)-based analyses from a subsample of the present study (Heinzel et al., 2014a), we performed fMRI whole-brain analyses in an extended sample.

The “Compensation-Related Utilization of Neural Circuits Hypothesis” (CRUNCH) by Reuter-Lorenz and Cappell (2008) describes age-related hyperactivations at low WM load in terms of reduced neural efficiency and hypoactivations at high WM load in terms of reduced neural capacity (Barulli and Stern, 2013). In a more general framework (the “Scaffolding Theory of Aging and Cognition”, this pattern of both reduced neural efficiency and capacity has been described as reduced neural “adaptivity” (Park and Reuter-Lorenz, 2009, Reuter-Lorenz and Park, 2014). In the context of WM, this resembles the notion that younger adults are able to adapt task-related activity according to increasing WM load, whereas older adults require compensational activation already at low WM load and are unable to further recruit neural resources at high WM load. In line with previous n-back research (e.g., Nagel et al., 2011, Nyberg et al., 2009), “low WM load” refers to 1-back, whereas “high WM load” refers to 3-back in the present study. When testing interindividual differences within older participants, previous research (for review see Grady, 2012, Nyberg et al., 2012, Stern et al., 2005) suggests that participants with less performance decrements in challenging cognitive tasks may also show more “youth-like” (Nagel et al., 2011) brain activation patterns. Better WM performance was associated with higher WM load-dependent adaptivity of neural activations (i.e., low activation at low WM load [1-back] and high activation at high WM load [3-back], Nagel et al., 2011, Nyberg et al., 2009). Although this concept of adaptivity has been operationalized and investigated predominantly in terms of neural activity, underlying network dynamics that may govern these neural activities remain largely unknown. Recently, a few studies have tested correlation-based functional connectivity in older age, indicating deficient frontoparietal coupling that may relate to changes in neural activity (Heinzel et al., 2014a, Matthäus et al., 2012, Nagel et al., 2011, Steffener et al., 2012). It has been suggested that these alterations might be due to an age-related deficit in prefrontal top-down control over posterior regions (Gazzaley et al., 2005, Gazzaley and D'Esposito, 2007).

However, no models of effective connectivity, such as dynamic causal modeling (DCM, Friston et al., 2003), supporting this proposal have been tested in older adults to date. In younger adults, increasing WM load-dependent effective connectivity from DLPFC to IPL (“top-down”/“backward”) best explained fMRI data acquired during a numeric “n-back” WM task (Deserno et al., 2012). Until now, it has not been investigated how aging may alter this architecture of WM load-dependent prefrontal to parietal effective connectivity. Thus, the aim of the present study was to go beyond correlational connectivity approaches and to test competing directional neuronal models of WM-dependent frontoparietal effective connectivity in older age for the first time. Results could improve the understanding of underlying mechanisms in terms of network dynamics of age-related WM deficits and integrate notions from compensation and top-down control models of aging. Therefore, we applied DCM (Friston et al., 2003) complemented by Bayesian model selection (BMS) techniques (Stephan et al., 2009) to fMRI data of younger and older adults performing a numeric n-back WM task as used in previous studies (Deserno et al., 2012, Heinzel et al., 2014a, Owen et al., 2005).

According to previous work (Nee et al., 2013, Owen et al., 2005), 3 nodes were identified to represent a simplified model of frontoparietal top-down control during WM: DLPFC, LPMC, and IPL. During WM, IPL was found to be involved in processes of information storage (Chein and Fiez, 2010, Christophel et al., 2012, Guerin and Miller, 2011, Todd and Marois, 2004), and DLPFC has been suggested to play a key role in top-down control of WM storage (Curtis and D'Esposito, 2003, Edin et al., 2009) potentially by guiding the selection of relevant and suppression of irrelevant information (Gazzaley et al., 2007, McNab and Klingberg, 2008, Montojo and Courtney, 2008). The LPMC has been associated with the attention-based rehearsal in WM (Baddeley, 2003, Bledowski et al., 2010, Curtis and D'Esposito, 2003) and was found to be most strongly involved at high task demands during WM updating (Nee et al., 2013, Wager and Smith, 2003).

In the present study, the following hypotheses were tested:

  • 1.

    Older adults show lower WM performance, specifically at high WM load.

  • 2.

    Older adults show increased neural activation at low (1-back) and decreased activation at high (3-back) WM load in frontoparietal WM regions, indicating reduced WM load-dependent adaptivity of neural activations.

  • 3.

    FMRI data are best explained by a “backward” model, comprising a modulatory connection from DLPFC to IPL, supporting notions from “top-down” control models.

  • 4.

    Older adults show a reduced WM load-dependent modulation of DLPFC to IPL connectivity, thus proposing an extension of the concept of WM load-dependent adaptivity in terms of effective connectivity.

  • 5.

    High-performing older adults show higher WM load-dependent modulation of DLPFC to IPL connectivity compared to low-performing older adults.

Section snippets

Participants and screening instruments

Thirty-four older participants (age: 60–75 years) and 18 younger participants (age: 21–30 years) were recruited via newspaper and online announcements in Berlin, Germany. All participants were native German speakers, right-handed, had normal or corrected-to-normal vision, no history of any neurological or psychiatric diseases, and did not take any psychiatric medication. Mini-Mental Status Examination (Folstein et al., 1975) was 27 or above in all participants. All participants were suitable

N-back performance in younger and older adults

To test whether differences between age groups in n-back performance (defined as hit rate minus false alarm rate) were related to WM load, a 2 (age group) by 3 (WM load) ANOVA with scanner site as covariate was conducted. A significant main effect of WM load (F[2,94] = 112.31, p < 0.001, partial η2 = 0.705) showed that n-back performance decreased with higher load levels irrespectively of age. Younger participants outperformed older participants irrespectively of WM load, as demonstrated by a

Discussion

In keeping with previous research, age-related decrements in WM-performance observed in the present study were most pronounced at high levels of WM load. Neural activation in frontoparietal regions showed a WM load by age interaction, indicating both higher activations during 1-back and lower activations during 3-back in older compared to younger participants. On the level of effective connectivity, fMRI data of both age groups were best explained by a “backward” model, including WM-dependent

Disclosure statement

The authors have no actual or potential conflicts of interest.

Acknowledgements

This work was supported by German National Academic Foundation scholarships to SH and RCL, the German Ministry for Education and Research (BMBF 01QG87164 and 01GS08195 and 01GQ0914), the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, FOR 1617: grant RA1047/2-1 and in part by the DFG SPP 1772: grants RA1047/4-1 and HE 7464/1-1), and by a MaxNetAging award to M.A.R. LD is supported by the Max Planck Society. Fractions of the current work are part of an unpublished thesis. The

References (58)

  • J.A. Maldjian et al.

    An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets

    NeuroImage

    (2003)
  • C.A. Montojo et al.

    Differential neural activation for updating rule versus stimulus information in working memory

    Neuron

    (2008)
  • L. Nyberg et al.

    Memory aging and brain maintenance

    Trends Cogn. Sci.

    (2012)
  • C. Rottschy et al.

    Modeling neural correlates of working memory: a coordinate-based meta-analysis

    NeuroImage

    (2012)
  • K.E. Stephan et al.

    Bayesian model selection for group studies

    NeuroImage

    (2009)
  • P. Verhaeghen et al.

    Aging, executive control, and attention: a review of meta-analyses

    Neurosci. Biobehav. Rev.

    (2002)
  • A. Baddeley

    Working memory: looking back and looking forward

    Nat. Rev. Neurosci.

    (2003)
  • C. Bledowski et al.

    What “works” in working memory? Separate systems for selection and updating of critical information

    J. Neurosci. Off. J. Soc. Neurosci.

    (2009)
  • R. Cabeza et al.

    Task-independent and task-specific age effects on brain activity during working memory, visual attention and episodic retrieval

    Cereb. Cortex

    (2004)
  • J.M. Chein et al.

    Evaluating models of working memory through the effects of CONCURRENT irrelevant information

    J. Exp. Psychol. Gen.

    (2010)
  • T.B. Christophel et al.

    Decoding the Contents of visual short-term memory from human visual and parietal cortex

    J. Neurosci.

    (2012)
  • J.D. Cohen et al.

    Temporal dynamics of brain activation during a working memory task

    Nature

    (1997)
  • F.I.M. Craik et al.

    The Handbook of Aging and Cognition

    (2011)
  • L. Deserno et al.

    Reduced prefrontal-parietal effective connectivity and working memory deficits in schizophrenia

    J. Neurosci. Off. J. Soc. Neurosci.

    (2012)
  • I.T.Z. Dew et al.

    Where is ELSA? The early to late shift in aging

    Cereb. Cortex

    (2012)
  • F. Edin et al.

    Mechanism for top-down control of working memory capacity

    Proc. Natl. Acad. Sci. U. S. A.

    (2009)
  • A. Gazzaley et al.

    Age-related top-down suppression deficit in the early stages of cortical visual memory processing

    Proc. Natl. Acad. Sci. U. S. A.

    (2008)
  • A. Gazzaley et al.

    Top-down suppression deficit underlies working memory impairment in normal aging

    Nat. Neurosci.

    (2005)
  • A. Gazzaley et al.

    Top-down modulation and normal aging

    Ann. N. Y. Acad. Sci.

    (2007)
  • Cited by (0)

    View full text