Brain correlates of linguistic-cognitive stimulation in neurotypical and Atypical older adult populations: A systematic review

BACKGROUND: Grounded in the perspective of language as a high-order cognitive organizer, language engages multiple domains, suggesting that linguistic stimulation may strengthen cognitive resources holistically. Despite its potential for promoting neuroplasticity and mitigating decline, the underlying neural mechanisms and their correlation with behavioral outcomes remain unclear. OBJECTIVE: To synthesize neuroimaging evidence on the effects of linguistic-cognitive interventions on brain structure and function across the cognitive aging continuum (healthy, SCD, MCI, and AD). METHODS: Following PRISMA 2020, we searched articles produced between 2009 and 2024 with randomized controlled trials evaluating neuroimaging outcomes following linguistic-cognitive interventions. Ten studies met inclusion criteria. Data were independently extracted by four reviewers, and risk of bias was assessed with Cochrane RoB 2.0. Neuroimaging modalities included structural measures (VBM, DTI) and functional measures (rs-fMRI, PET). RESULTS: Samples ranged from 30 to 160 participants, aged 64-82 years, with 3.4-17.1 years of education. Six studies reported significant post-intervention brain changes (five functional, one structural). Functional effects were mainly observed in the default mode network, frontoparietal regions, and hippocampal circuits. Structural findings included gray matter increases and improved white matter integrity. Neuroimaging effects were more frequent in MCI/AD and in in-person versus online/hybrid interventions. However, six studies found no significant behavioral improvement despite neural changes. Risk of bias was high in seven studies. CONCLUSIONS: Linguistic-cognitive interventions can modulate brain connectivity and, less consistently, structure in aging. Neuroimaging may be a sensitive biomarker of intervention efficacy, but larger, longitudinal, multimodal trials with ecologically valid outcomes are needed.