The Effects of Cognitive Training on Brain Network Activity and Connectivity in Aging and Neurodegenerative Diseases: a Systematic Review

Abstract

Cognitive training (CT) is an increasingly popular, non-pharmacological intervention for improving cognitive functioning in neurodegenerative diseases and healthy aging. Although meta-analyses support the efficacy of CT in improving cognitive functioning, the neural mechanisms underlying the effects of CT are still unclear. We performed a systematic review of literature in the PubMed, Embase and PsycINFO databases on controlled CT trials (N > 20) in aging and neurodegenerative diseases with pre- and post-training functional MRI outcomes up to November 23rd 2018 (PROSPERO registration number CRD42019103662). Twenty articles were eligible for our systematic review. We distinguished between multi-domain and single-domain CT. CT induced both increases and decreases in task-related functional activation, possibly indicative of an inverted U-shaped curve association between regional brain activity and task performance. Functional connectivity within 'cognitive' brain networks was consistently reported to increase after CT while a minority of studies additionally reported increased segregation of frontoparietal and default mode brain networks. Although we acknowledge the large heterogeneity in type of CT, imaging methodology, in-scanner task paradigm and analysis methods between studies, we propose a working model of the effects of CT on brain activity and connectivity in the context of current knowledge on compensatory mechanisms that are associated with aging and neurodegenerative diseases.

Keywords: Aging; Cognitive training; Magnetic resonance imaging; Network; Neurodegenerative diseases; Neuroimaging.

Fig. 1

Flow diagram of the screening process according to PRISMA guidelines

Fig. 2

Overview of the findings of all included studies, irrespective of population or training type, that reported coordinates of brain regions with CT-induced alterations. Each dot represents a single alteration in activity (panel a) or connectivity (in which both the seed and connected region are displayed; panel b). In panel b, the seed and connected regions are classified by resting-state network (parcellation according to Yeo et al., ; Choi et al., ; Buckner et al., 2011) to illustrate within- and between-network connectivity changes. The side views show intra-hemispheric connections. Details about these studies are listed in Table 1. Abbreviations – FPN: frontoparietal network; DMN: default mode network; VAN: ventral attention network; DAN: dorsal attention network; SMN: somatomotor network; n.a.: no network assigned

Fig. 3

Working model. a The inverted U-shaped association between regional brain activity during task performance and task load. Aging and neurodegenerative diseases lead to a shift of the curve to the left, while CT seems to induce the opposite, illustrated by the horizontal arrows. Consequently, at the same task load different neural resources are needed/used. b The association between task-related brain activity and network connectivity and modularity at increasing age or disease stage. The arrows indicate the suggested effect of CT at different stages of aging or disease. Both in panel (a) and (b), (1) indicates training-induced hypo-activity associated with neural efficiency: either a) tasks with lower demand can be performed more efficiently through cognitive training (panel a), or b) CT (partially) restores compensatory hyper-activity that is associated with early stages of aging and neurodegenerative diseases to a more ‘healthy’ state (panel b); (2) indicates that CT leads to hyper-activity that is associated with increased effort and which is needed to successfully fulfill a task with a high cognitive demand