Foreign Language Learning in Older Adults Modifies Resting‐State Functional Connectivity Between the Subcortical Structures and the Cortex

Giovanna Bubbico; Federica Tomaiuolo; Carlo Sestieri; Golnoush Akhlaghipour; Alberto Granzotto; Antonio Ferretti; Mauro Gianni Perrucci; Stefano L Sensi; Stefano Delli Pizzi

doi:10.1002/agm2.70073

. 2026 Apr 23;9(2):99–110. doi: 10.1002/agm2.70073

Foreign Language Learning in Older Adults Modifies Resting‐State Functional Connectivity Between the Subcortical Structures and the Cortex

Giovanna Bubbico ^1,^2,^✉, Federica Tomaiuolo ^2,³, Carlo Sestieri ^1,², Golnoush Akhlaghipour ⁴, Alberto Granzotto ^1,⁵, Antonio Ferretti ^1,^2,⁶, Mauro Gianni Perrucci ^1,², Stefano L Sensi ^1,^2,^5,⁷, Stefano Delli Pizzi ^1,^2,^5,^✉

PMCID: PMC13163939 PMID: 42130753

ABSTRACT

Objectives

Neuroimaging studies suggest that Foreign Language Learning (FLL) influences resting‐state functional connectivity (rs‐FC) within the neocortex, but its effects on subcortical–cortical connectivity remain unclear. We hypothesize that FLL may specifically affect the rs‐FC of subcortical regions involved in integrating cortical networks, leading to adaptive changes in overall brain connectivity.

Methods

Twenty‐seven healthy older adults were randomly assigned to either a 16‐week FLL intervention or a control condition. Resting‐state fMRI data were analyzed using seed‐based connectivity methods targeting seven subcortical structures. Group differences in rs‐FC were examined alongside correlations with maps of the expression of receptors/genes related to plasticity and behavioral performance.

Results

FLL induced specific and distributed changes in rs‐FC between subcortical and cortical areas. Notably, thalamocortical circuits showed both increases and decreases in connectivity with regions of the default mode network. Increases of striato‐cortical connectivity targeted frontal regions. Changes of connectivity spatially overlapped with cortical areas enriched in CB1 receptor density and NTRK2 gene expression. Importantly, FLL also modulated pallidal connectivity, a change that was significantly associated with improvements in semantic memory.

Conclusions

The FLL program selectively reorganizes subcortico–cortical networks, particularly those involving the thalamus and the caudate, in a way that supports the biological plausibility of FLL‐induced functional plasticity. The alignment with plasticity‐related molecular markers and the observed cognitive improvements further support the potential of FLL as an ecologically valid strategy for preserving brain function in later life.

Keywords: aging, CB1 receptor, NTRK2 gene expression, resting‐state functional connectivity, semantic memory, subcortical networks

Foreign language learning in older adults reorganizes thalamic and caudate connectivity, in a process topographically overlapping with the expression of receptors and genes related to neuroplasticity.

graphic file with name AGM2-9--g004.jpg

Abbreviations

AAL‐3: Automated Anatomical Labeling Atlas 3
AMN: Action Mode Network
ARC: Activity‐Regulated Cytoskeleton‐associated protein
ART: Artifact Detection Tool
BDNF: Brain‐Derived Neurotrophic Factor
BOLD: Blood Oxygenation Level‐Dependent
CB1: Cannabinoid receptor type 1
CSF: Cerebrospinal Fluid
dACC: Dorsal Anterior Cingulate Cortex
DMN: Default Mode Network
FAB: Frontal Assessment Battery
FDR: False Discovery Rate
FLL: Foreign Language Learning
fMRI: Functional Magnetic Resonance Imaging
FOV: Field of View
GLM: General Linear Model
GRIA1/GRIA2: AMPA receptor subunit genes
IFG: Inferior Frontal Gyrus
MCI: Mild Cognitive Impairment
mGluR5: Metabotropic Glutamate Receptor 5
MMSE: Mini‐Mental State Examination
mPFC: Medial Prefrontal Cortex
MRI: Magnetic Resonance Imaging
NTRK2: Neurotrophic Receptor Tyrosine Kinase 2 (BDNF receptor)
PASA: Posterior–Anterior Shift in Aging
PCC: Posterior Cingulate Cortex
PET: Positron Emission Tomography
rs‐FC: Resting‐State Functional Connectivity
rs‐fMRI: Resting‐State Functional Magnetic Resonance Imaging
SBC: Seed‐Based Connectivity
T1/T2: Pre‐intervention and Post‐intervention MRI sessions
TMT: Trail Making Test

1. Introduction

An increasing number of studies have examined the long‐term benefits of Foreign Language Learning (FLL) on cognition, with particular focus on its potential to counteract cognitive decline associated with aging [1]. Specifically, learning one or more foreign languages can enhance cognitive flexibility, improve working memory, and strengthen executive functions, such as concentration, problem‐solving, and planning [2, 3, 4]. Furthermore, recent evidence suggests that bilingualism and multilingualism may slow the onset of neurodegenerative diseases by improving the brain's resilience to age‐related cognitive decline [5]. These protective effects are attributed to the “brain plasticity” induced by language learning, which stimulates brain circuits that are crucial for higher cognitive functions.

Functional neuroimaging has proven to be an effective tool for assessing the effects of FLL in older adults. To date, the available studies in the literature suggest that FLL alters the spontaneous brain activity, measured in terms of resting‐state functional connectivity (rs‐fMRI), within the so‐called executive control [3, 6], language [6], and cingulo‐opercular network [4]. In contrast, except for one investigation that examined the cerebellar‐cortical functional coupling [2], no study has investigated the effects of FLL on the connectivity between subcortical and cortical areas. However, it is now well established that several subcortical structures, such as the hippocampus, striatum, thalamus, and amygdala, support multiple cognitive and affective functions (e.g., memory, motor control, emotional processing, and attention regulation) through their interaction with associative neocortical areas [7]. Interestingly, with aging, these structures and their associated circuits undergo profound anatomical and functional changes, with a potential effect on both cognition and emotion.

Specifically, the hippocampus, which is crucial for the formation of long‐term memories [8, 9], decreases in volume and loses connectivity with the cortex, a phenomenon associated with deficits of episodic memory (e.g., recall of specific events) [10] and spatial memory (orientation in the environment, [11]). The striatum, which plays a key role in motor control, learning, and emotional regulation [12], is also affected by aging. The dysfunction in striatal circuits has been linked to declines in executive functions, such as planning and decision‐making [13]. The thalamus, which is often regarded as a “hub” for sensory and motor information, also plays a key role in the integration of neocortical networks, impacting higher cognitive functions such as memory and executive functions [14]. Notably, atrophy of thalamic nuclei can occur at high and heterogeneous rates during aging, even in individuals with no overt cognitive impairment [15]. Finally, the amygdala, essential for emotional regulation [16, 17], may exhibit dysfunctional activity with aging, leading to increased emotional reactivity and difficulties in emotion regulation. Of note, an increase in emotional stress, which is common in older adults, can negatively affect cognitive well‐being, and recent studies have observed that structural changes in the amygdala can be predictive of the development of dementia [18].

Thus, different subcortical structures are not only crucial for daily functioning but also appear to play a role in brain resilience during aging. In previous work, we have observed changes in functional connectivity between subcortical and cortical structures that may represent compensatory mechanisms associated with pathological aging (e.g., MCI patients) [19, 20]. The working hypothesis of the present study is that FLL may promote plasticity mechanisms in subcortical regions, fostering adaptive processes associated with changes in connectivity between subcortical and cortical areas. In particular, we expect thalamocortical connectivity to show detectable modifications induced by the FLL intervention, given its involvement in both learning and memory [21], as well as in declarative and procedural linguistic memory processes [22, 23], fluent reading [24], and word recognition [25]. Furthermore, we expected that changes in connectivity would target cortical areas that are part of the “Action Mode Network” (AMN) [26], a new conceptualization of the previously described cingulo‐opercular network [27]. This hypothesis is grounded in recent theoretical models that describe the AMN as a control network supporting adaptive, goal‐directed behavior and bridging cognitive and motor functions. Since the AMN encompasses cortical areas such as the inferior frontal gyrus (IFG), anterior insula, dorsal anterior cingulate cortex (dACC), and supplementary motor areas, regions critically involved in language, attention, and motor planning, it represents a plausible target for FLL‐induced plasticity.

In this pilot study, a cohort of 27 healthy older adults underwent Blood Oxygenation Level Dependent (BOLD) resting‐state functional MRI (rs‐fMRI) before and after a 16‐week FLL program (n = 14) or a no‐intervention condition (n = 13). A seed‐based connectivity (SBC) approach [28] was used to investigate whether the FLL intervention modulated resting‐state functional connectivity (rs‐FC) between subcortical structures and the neocortex, as well as among subcortical regions. A network‐level analysis assessed the spatial overlap between FLL‐induced changes in the connectivity of subcortical regions and the distribution of several receptors implicated in neuroplasticity during aging [29]. These included the cannabinoid receptor CB1 [30, 31], the GABAa receptor [32, 33], the NMDA receptor [34], and the metabotropic glutamate receptor mGluR5 [35]. A second network‐level analysis examined the overlap with the distribution of the main genes involved in synaptic plasticity: (a) GRIA1 and GRIA2, encoding AMPA receptor subunits critical for learning and memory [36]; (b) ARC, a gene mediating activity‐dependent synaptic modifications and essential for long‐term memory formation [37]; (c) BDNF, which supports neuronal growth, differentiation, and synaptic plasticity [38]; and (d) NTRK2, encoding the BDNF receptor, which plays a key role in synaptic function and neurogenesis [39]. Finally, an exploratory analysis investigated whether FLL‐related changes in rs‐FC were associated with improvements in cognitive performance.

2. Materials and Methods

This study draws upon a dataset previously used in earlier publications by our research team [2, 3]. The novelty of the present approach lies in the application of a seed‐based rs‐FC approach to further investigate how FLL influences subcortical–cortical resting‐state functional connectivity. Importantly, the present study introduces several key innovations relative to our earlier work [2, 3, 6]. First, we extend the analysis of subcortical–cortical connectivity to include pallidal and amygdalar seeds in addition to thalamic and caudate regions. Second, we combine resting‐state functional connectivity with cortical maps of receptor density, including CB1, and gene expression, including NTRK2, to provide a molecular context for the observed functional changes. The study was approved by the Local Ethics Committee (number: 1711), and participants provided written informed consent. All procedures were conducted in accordance with the ethical standards outlined in the Declaration of Helsinki.

2.1. Experimental Design

A total of 27 older adults were randomly assigned to two groups in approximately 1:1 ratio and completed the study. Fourteen participants underwent a 16‐week beginner‐level English course, while 13 formed the no‐intervention group. Participants were assigned to the FLL or control group using simple randomization, without stratification for age, sex, or education. The FLL group attended weekly 2‐h sessions, including a 90‐min class with a native instructor, a 15‐min break, and 30 min of assigned homework. The course focused on fundamental English skills, cultural aspects, and practical language use, incorporating team‐based projects to enhance engagement. Progress was qualitatively assessed at both the beginning and end of the program. The no‐intervention group did not receive formal training but completed the same pre‐ and post‐tests. To ensure they maintained their usual activities, participants were contacted via monthly phone calls. Further study details, including clinical histories and criteria, were reported in our previous study [6].

2.2. Cognitive Evaluation

All participants underwent a neuropsychological evaluation before and after the FLL intervention. The assessment encompassed multiple cognitive domains, including executive functions, episodic memory, verbal fluency, and working memory. Global cognitive decline was measured using the Mini‐Mental State Examination (MMSE). Sustained attention (TMT A), divided attention (TMT B), and cognitive flexibility (TMT B‐A) were assessed with the Trail Making Test (TMT). Phonological access and executive functioning were evaluated using the Verbal Fluency Test, while episodic memory was measured with the Babcock Memory Test. Executive function performance was further assessed using the Frontal Assessment Battery (FAB). Detailed demographic and neuropsychological profiles of both groups are available in our previous study [6]. Raw scores from each neuropsychological test were first converted into age‐standardized z‐scores based on the means and standard deviations from a normal reference subset specific to each neuropsychological outcome [40, 41, 42].

2.3. MRI Data Collection

Participants underwent an MRI scan both before (T₁) and after (T₂) the intervention. Imaging was conducted using a Philips Achieva 3 Tesla scanner (Philips Medical Systems, Best, Netherlands). Structural MR images were acquired with a 3D fast field echo T1‐weighted sequence (dimensions: 170 × 240 × 240 mm³, slice thickness = 1 mm, in‐plane voxel size = 1 × 1 mm). Structural images were used to generate gray/white matter and cerebrospinal fluid (CSF) masks, which were used for confounding removal in the preprocessing step. For each scan, BOLD rs‐fMRI images were collected with eyes closed using a gradient‐echo T2*‐weighted echo‐planar sequence (FLL group: 3 runs; dimensions: 80 × 80 × 35 × 180; FOV 240 mm; TR = 1.8 s, voxel size: 3 × 3 × 3 mm³; no‐intervention group: 4 runs; dimensions: 64 × 64 × 30 × 145; FOV = 256 mm; TR = 1.67 s; voxel size: 4 × 4 × 4 mm³). Participants were awake and alert during the MRI sessions. Alertness was periodically confirmed by asking participants to make a small movement with their right foot during the intervals between different acquisition runs.

2.4. Rs‐fMRI Data Analysis

The data were processed and analyzed using the Conn toolbox (version 22), [43]. Preprocessing steps were performed according to the methods previously described in our previous studies [28, 44]. Specifically, the preprocessing step included slice‐timing correction via sinc interpolation, motion correction, outlier detection for scrubbing based on the ART method, and spatial smoothing using an 8 mm FWHM kernel. The preprocessed functional images were then co‐registered with the corresponding anatomical scans, normalized to MNI space, linearly detrended, and band‐pass filtered (0.008–0.09 Hz, post‐regression). Seed‐based rs‐FC analysis (https://web.conn‐toolbox.org/fmri‐methods/connectivity‐measures/seed‐based) was performed, deriving Fisher‐transformed bivariate correlation coefficients between each seed and all other brain voxels. Seven subcortical regions—including the accumbens, amygdala, caudate, hippocampus, pallidus, putamen, and thalamus—were defined using the Conn‐integrated Automated Anatomical Labeling atlas 3 (AAL‐3) [45] and selected as seeds for standard SBC analysis. The rs‐FC between these seed regions and individual voxels across the entire brain was then analyzed to identify rs‐FC patterns.

2.5. Spatial Correlation With Receptor and Gene Density

Session‐related changes (delta: T2—T1) were compared between the FLL and no‐intervention groups to assess the net effect of the FLL intervention (i.e., delta FLL—delta no‐intervention). Unthresholded statistical maps were used to extract z‐scores from the 100 cortical parcels defined by the Schaefer atlas (https://github.com/ThomasYeoLab/CBIG/tree/v0.14.3‐Update_Yeo2011_Schaefer2018_labelname/stable_projects/brain_parcellation/Schaefer2018_LocalGlobal/Parcellations). These contrast maps were designed to isolate effects specifically attributable to the FLL program, disentangled from no‐intervention‐related mechanisms potentially affecting neurotransmission. Non‐thresholded z‐statistic maps generated during CONN preprocessing were projected onto the fsaverage5 cortical surface using FreeSurfer (https://surfer.nmr.mgh.harvard.edu/). Mean z‐values were then extracted for each of the 100 Schaefer parcels in the 7‐network scheme.

For the seed‐based connectivity maps showing significant results, two independent network‐level correlation analyses were conducted. First, we explored the spatial correspondence between resting‐state fMRI activity and neurochemical signatures derived from a PET‐based receptor atlas [46]. Based on the rationale outlined in the introduction, we focused on a subset of four receptors implicated in the modulation of brain function: CB1, GABAa, NMDA, and mGluR5. Receptor and transporter density values were extracted for each of the 100 cortical regions according to the Schaefer 7‐network parcellation [47], accessible at https://github.com/netneurolab/hansen_receptors.

Second, we assessed the spatial correlation between resting‐state fMRI maps and transcriptomic profiles from the MAGICC atlas, focusing on genes linked to neuroplasticity—specifically GRIA1 and GRIA2 (which encode AMPA receptor subunits), as well as ARC, NTRK2, and BDNF, which are critical modulators of synaptic plasticity. The MAGICC atlas, built on data from the Allen Human Brain Atlas, includes expression measures for 20,781 genes across 1304 cortical tissue samples from the left hemisphere of six postmortem brains. These expression levels are mapped to approximately 30,000 cortical surface vertices, enabling high‐resolution spatial analysis. The atlas provides gene expression data across 50 left‐hemisphere cortical regions, based on the 7‐network Schaefer parcellation [47], available at https://github.com/kwagstyl/MAGICC.

To account for spatial autocorrelation inherent in cortical topography, we employed the BrainSMASH framework [48], implemented via the “neuromaps” Python package (https://github.com/netneurolab/neuromaps). This method creates 1000 surrogate receptor maps that retain the spatial autocorrelation structure, mean, and variance of the original data. Analyses were conducted separately for each hemisphere. These surrogate datasets facilitated the construction of two null models: one that disregarded spatial structure and one that preserved it. Accordingly, we computed two types of non‐parametric p‐values: one spatially naïve and one spatially adjusted. Moreover, Bonferroni correction was applied to account for multiple comparisons across receptors (corrected p = 0.05/4 receptors/2 hemispheres = 0.006) and genes (corrected p = 0.05/5 genes = 0.01) tested for each map for gene maps.

2.6. Statistical Analysis

By using a General Linear Model (GLM), for each voxel, we estimated rs‐FC changes—defined as the difference in z‐Fisher transformed connectivity strength between post‐treatment (T₂) and pre‐treatment (T₁)‐with groups (FLL and no‐intervention) as the independent variables and age and educational level as covariates. Voxel‐level hypotheses were tested using multivariate statistics, taking into account random effects and covariance across subjects. Cluster‐level inferences were made using Gaussian Random Field theory [49] with thresholds of p < 0.001 for voxel‐level clusters and p < 0.05 for FDR‐corrected cluster size [50]. Pearson's correlations were then used within the FLL group to examine the presence of significant associations between significant changes (T₂‐T₁) in rs‐fMRI metrics and neuropsychological scores; a significant difference between the two groups was found (MMSE and Babcock Memory Test). Of note, connectivity change estimates were derived from GLM models that included age and education as covariates. Therefore, the resulting connectivity change values used in brain–behavior correlations were already adjusted for these demographic factors.

3. Results

3.1. Changes of rsFC of Subcortical Regions Induced by FLL

The seed‐based connectivity analysis revealed significant alterations in rs‐FC between the neocortex and several subcortical structures, including the right thalamus, left caudate nucleus, right globus pallidus, putamen and left amygdala Table S1. The power analysis results reveal robust and statistically significant effects of the FLL intervention on functional connectivity across several subcortical regions, with large effect sizes (Cohen's d = 1.330–1.780, Table S2).

The main finding was the observation of a decrease in rs‐FC between the right thalamus (Figure 1A) and both the posterior cingulate cortex (PCC) and the medial prefrontal cortex (mPFC). These two regions represent the posterior and anterior portions of the so‐called Default Mode Network (DMN) [51], a task‐negative system commonly affected by aging [52]. Additionally, the right thalamus exhibited reduced connectivity with the posterior insula and increased connectivity with the supramarginal gyrus, which are part of the sensorimotor system and the previously described AMN, respectively. The left thalamus exhibited decreased connectivity with the visual cortex, specifically the cuneus (Figure 1B). Furthermore, the left caudate nucleus showed increased connectivity with the superior frontal regions, which are part of the DMN (Figure 1C).

FLL‐induced changes in thalamic (A,B) and caudate (C) functional connectivity with the neocortex. The seed regions used in the seed‐to‐voxel functional connectivity analysis are highlighted in green. The cortical surface maps illustrate the topographic distribution of clusters showing FLL‐induced changes in functional connectivity. Red–yellow clusters indicate regions with increased connectivity following FLL, while violet–blue clusters represent areas of decreased connectivity. Violin plots display the group distribution of functional connectivity values within significant clusters (red: FLL; blue: Controls). The boxplots represent the quartiles of the dataset; whiskers indicate the range of values within 1.5 times the interquartile range (IQR), and values outside this range are displayed as potential outliers. MPFC = medial prefrontal cortex; PCC = posterior cingulate cortex; pINS = posterior insula; SFC = superior frontal cortex; SMG = supramarginal gyrus.

When looking at changes of connectivity between subcortical structures, we observed that the left amygdala displayed increased rs‐FC with the ventrolateral subregion of the thalamus (Figure 2A), while the right globus pallidus showed increased connectivity with the posterior thalamus and the precentral gyrus (Figure 2B).

FLL‐induced changes in functional connectivity between the amygdala (A) and pallidum (B) with rest of the brain. Seed regions used for the seed‐to‐voxel analysis are shown in green. The cortical maps display the topographic distribution of clusters exhibiting FLL‐induced connectivity changes. Red–yellow clusters indicate regions with increased connectivity. Violin plots illustrate the group‐wise distribution of connectivity values within significant clusters (red: FLL; blue: Control group). MC = motor cortex.

No other significant changes in the connectivity of subcortical regions were observed.

3.2. Relationship Between Changes of Subcortical–Cortical Rs‐FC and the Spatial Distribution of Receptors and Gene Expression

FLL‐related changes reported in the rs‐FC maps showed a marked leftward lateralization (Figure 3); Tables S3–S10. At the network level, significant correlations were observed between the maps of connectivity changes of the left thalamus (SA uncorrected p < 0.001; p = 0.002; R = 0.405) and left caudate nucleus (SA uncorrected p = 0.001; SA corrected p = 0.004; R = 0.364) and the spatial distribution of CB₁ receptor expression. A similar, though non‐significant, trend was observed for the right thalamus (SA uncorrected p = 0.002; SA corrected p = 0.016; R = 0.414). Moreover, while the map of the connectivity changes of the left caudate showed a significant correlation with the expression of NTRK₂ (SA uncorrected p = 0.004; SA corrected p = 0.009; R = 0.397), a trend toward a significant association was observed between the map of the left thalamus and the expression of NTRK₂ (SA uncorrected p = 0.004; SA corrected p = 0.022; R = 0.024).

Spatial relationships between changes in functional connectivity and receptor/gene expression. Scatterplots show the spatial associations between rs‐fMRI connectivity changes (T2–T1 delta in the FLL vs. no‐intervention groups) and receptor/gene expression maps. Panels A–C depict the overlap between rs‐FC changes and the distribution of CB1 receptor density and NTRK2 gene expression, respectively.

3.3. Relationships Between Changes of Subcortical–Cortical rsFC and Neuropsychological Outcomes

Table S11 reports baseline demographic characteristics (age, sex, education) and baseline cognitive scores (e.g., MMSE and Babcock Memory Recall Test) for the FLL and control groups, together with the corresponding statistical comparisons.

As already shown in our previous studies [2, 3], at the behavioral level, FLL participants showed a significant improvement in semantic memory performance compared with controls, as indexed by the Babcock Memory Recall Test, whereas changes in MMSE scores were modest and should be interpreted with caution, given the lack of alternative forms and the screening nature of this measure.

The linear regression analysis identified significant associations between changes in rs‐FC, particularly of the pallidum, and improvements in semantic long‐term memory, as measured by the Babcock Memory Recall Test. Notably, increased rs‐FC between the right pallidum and the precentral gyrus (β = 0.737, p = 0.002; Figure 4A) and decreased rs‐FC between the right pallidum and the thalamus (β = −0.792, p < 0.001; Figure 4B) were both significantly associated with better memory performance. These results remained significant after correction for multiple comparisons, indicating that pallidum‐related connectivity changes may play a specific role in supporting the consolidation of semantic memory processes.

Scatterplots illustrate the correlation between the strength of pallidum (A,B) and thalamic (C) functional connectivity and the performance changes on the Babcock Memory Recall Test scores (T2–T1). MC = motor cortex; mPFC = medial prefrontal cortex.

In contrast, although the rs‐FC between the right thalamus and the medial prefrontal cortex (mPFC) showed a positive association with Babcock scores (β = 0.531, p = 0.019; Figure 4C), this effect did not survive correction for multiple comparisons and should therefore be interpreted with caution.

The direction of the β coefficients helps clarify the underlying relationships: a positive β (pallidum–precentral gyrus; thalamus–mPFC) indicates that greater functional coupling in these pathways is associated with better semantic memory performance, whereas a negative β (pallidum–thalamus) suggests that reduced functional coupling is linked to improved memory outcomes, possibly reflecting increased network efficiency or reduced interference. Together, these findings suggest that the FLL program modulates distinct connectivity pathways that may facilitate the encoding and retrieval of semantic information.

4. Discussion

This pilot study investigated the effects of FLL on rs‐FC in healthy older adults. The results revealed that the intervention elicited significant changes in the pattern of connectivity of several subcortical regions. The most prominent alterations involved the connectivity of the thalamus and the caudate nucleus, with additional changes observed for the amygdala and the pallidum. These modifications were directionally heterogeneous, encompassing both increases and decreases in connectivity, and were localized within specific neuroanatomical circuits. Taken together, these findings suggest a coordinated reorganization of large‐scale subcortico–cortical dynamics consistent with contemporary models of network‐level plasticity in aging.

Importantly, several of these changes spatially overlapped with the cortical distributions of neurotransmitter systems known to support synaptic plasticity, particularly the CB1 and the NTRK2. This pattern is consistent with previous findings from our group [3], which demonstrated that the FLL program in older adults modulates the functional connectivity of specific cerebellar subregions with cortical areas rich in CB1 receptor expression. These hemodynamic‐molecular associations are especially relevant given the established roles of CB1 and NTRK2 in neural plasticity. Specifically, CB1 is critically involved in modulating synaptic transmission and plasticity and has been linked to learning, memory consolidation, and cognitive flexibility across the lifespan [30, 31]. Similarly, NTRK2, the high‐affinity receptor for BDNF, is essential for activity‐dependent synaptic strengthening, neuronal survival, and large‐scale functional network remodeling [39]. Thus, the convergence between functional connectivity changes and the cortical expression patterns of these molecular markers suggests that the observed reorganization of thalamocortical and caudate–cortical networks may be associated, at least in part, with experience‐dependent neuroplastic signaling mechanisms. Nonetheless, since our receptor and gene expression data come from normative sources rather than individual‐level molecular imaging, this convergence should be interpreted with caution and viewed as a hypothesis‐generating observation rather than conclusive evidence of causal relationships.

Interestingly, the thalamus exhibited prominent FLL‐related changes in rs‐FC, characterized by both increases and decreases in connectivity with distributed cortical regions. This dual modulation (i.e., reduced connectivity with the DMN and increased connectivity with the AMN) is consistent with the known integrative role of the thalamus in dynamically coordinating the activity of large‐scale cortical networks [53]. Notably, these connectivity changes followed well‐established anatomic‐functional constraints. Reduced rs‐FC was observed between the thalamus and medial cortical regions, including the medial prefrontal cortex and posterior cingulate cortex, which are both core hubs of the DMN [51]. This is especially relevant given that these neocortical regions are commonly affected by aging [52] and are structurally and functionally connected to thalamic nuclei, such as the mediodorsal nucleus and the pulvinar [54, 55].

The observed reduction in thalamic‐DMN connectivity aligns with the broader role of the thalamus as an integrative hub that facilitates communication between posterior and anterior cortical regions. In older adults, age‐related cognitive decline is often accompanied by compensatory functional reorganization, typically characterized by increased activation in prefrontal areas and reduced activity in posterior cortices. Such reorganization is thought to represent the brain's adaptive response to structural and functional deterioration associated with aging. In our previous work [3], we reported increased frontal activity following an FLL intervention, which we linked to cognitive enhancement—particularly in long‐term verbal memory, a domain in which our participants had shown significant improvement. These findings are consistent with the Posterior‐to‐Anterior Shift in Aging (PASA) model [56], which posits that aging is associated with a shift in neural activity from posterior to anterior brain regions as a compensatory strategy, a view increasingly supported by empirical evidence. Within this framework, the thalamic connectivity changes observed in the present study gain further significance. Specifically, the reduction in connectivity between the thalamus and medial cortical regions—including the medial prefrontal cortex and posterior cingulate cortex—may reflect a functional reorganization aimed at optimizing information transfer across distant cortical systems, thereby enhancing computational efficiency during cognitively demanding tasks such as language learning [57]. In this sense, reduced thalamo‐DMN coupling may enable a more dynamic reallocation of neural resources toward circuits involved in attentional control and executive processing.

Conversely, increased rs‐FC was observed between the thalamus and the supramarginal gyrus, which is a posterior component of the AMN [28]. The AMN, also referred to as the cingulo‐opercular network, is strongly associated with attentional control and vigilance. The opposing directionality of thalamic connectivity changes with the DMN and AMN fits a coherent theoretical model of reallocation between internally and externally oriented systems—an organization increasingly recognized as central to cognitive plasticity in aging. Notably, the AMN and DMN often operate in opposition, regulating the balance between externally and internally directed mental states, a dynamic metaphorically described as a neural “yin and yang” [28]. The opposing directionality of thalamic connectivity changes with DMN and AMN regions is, therefore, both anatomically grounded and functionally meaningful. This pattern points to a broader reorganization of thalamocortical dynamics, potentially reflecting compensatory or homeostatic mechanisms. Specifically, reduced integration with introspective, self‐referential DMN regions may be offset by increased coupling with the AMN, which supports goal‐directed behavior and environmental engagement. These connectivity shifts may be driven by underlying molecular systems that promote neural plasticity and adaptive reconfiguration, enabling the brain to maintain functional efficiency in the context of age‐related or intervention‐induced change.

FLL was associated with increased rs‐FC between the caudate and superior frontal regions. This finding aligns with the well‐established role of the caudate in goal‐directed behavior, associative learning, and cognitive control [58, 59]. The caudate's involvement suggests subcortical contributions to compensatory recruitment of frontal systems during cognitively demanding learning experiences. Notably, this effect appears to counteract patterns typically observed in aging, as fronto‐striatal circuits are particularly vulnerable to age‐related decline, both in terms of structural integrity and functional connectivity [60, 61]. As observed for the thalamus, caudate‐related rs‐FC changes also showed spatial overlap with the cortical distribution of the CB1 receptor and the NTRK2 gene. This convergence reinforces the interpretation that these connectivity alterations may reflect adaptive, plasticity‐driven processes. Both CB1 and NTRK2 are implicated in experience‐dependent learning and long‐term synaptic remodeling, suggesting that molecular mechanisms of plasticity may underpin the functional reorganization observed in response to FLL.

The decrease in rs‐FC between the pallidum and the thalamus, and the increase in connectivity between the pallidum and the precentral gyrus, which is a region involved in the control of movements, were associated with better performance in semantic memory. These associations remained significant after correction for multiple comparisons, suggesting a specific link between changes in pallidal connectivity and memory performance. Although the pallidum is traditionally associated with motor regulation [62], recent evidence indicates its involvement in cognitive processes, particularly through its integration within cortico–basal ganglia–thalamic loops [63]. In this context, the observed modulation may reflect a reorganization of pallidal circuits that supports certain cognitive outcomes, even in tasks without an explicit motor component. These results extend the same network‐level interpretation, highlighting how different nodes of the basal ganglia–thalamocortical loop may undergo coordinated reorganization that supports specific cognitive gains. While these findings highlight the potential cognitive relevance of subcortical network changes, further research is needed to clarify the functional significance of pallidal connectivity in relation to specific aspects of memory and learning. Importantly, these observations offer preliminary but promising evidence for a broader role of the basal ganglia in supporting high‐level cognitive functions, particularly in older adults.

Although the amygdala is not typically central in models of cognitive training, FLL was associated with increased functional coupling between the amygdala and the thalamus. While this change was not directly linked to behavioral outcomes, it may reflect broader effects of language learning on arousal, motivation, or emotional salience. Emotional engagement is known to facilitate both encoding and retention, and the amygdala plays a key role in assigning salience to stimuli and modulating attentional resources [16, 64]. This neurofunctional modulation may help explain why FLL—being inherently social and emotionally engaging—could surpass other cognitive training modalities in fostering adherence and transfer effects. Within a unified framework, this modulation may reflect the engagement of emotional‐salience systems, which can facilitate learning through enhanced motivation and attentional allocation. Moreover, age‐related alterations in amygdala structure and connectivity have been linked to affective dysregulation, heightened stress sensitivity, and increased risk of dementia [18]. Our findings suggest that the amygdala remains functionally plastic in older adults and may be indirectly modulated through socially embedded learning experiences such as FLL.

A notable feature of the present findings is the predominance of left‐hemisphere effects. This lateralization is consistent with well‐established models of language dominance, particularly for semantic retrieval and language‐related cognitive control, which are strongly left‐lateralized in healthy adults [65]. Given that the FLL program emphasized vocabulary learning and semantic association, enhanced engagement of left‐hemisphere cortico–subcortical circuits is theoretically expected. An additional consideration is that the molecular maps used in the receptor and gene‐level correspondence analyses (CB1 receptor density and NTRK2 expression) are derived from left‐hemisphere data from the Allen Human Brain Atlas. As a result, these exploratory correlations inherently reflect the organization of left‐hemisphere cortical surfaces and may accentuate apparent left‐lateralized associations. Importantly, the same statistical thresholds and correction procedures were applied bilaterally in all rs‐FC analyses, meaning that the lateralization cannot be attributed to asymmetric thresholding. Instead, the observed pattern likely arises from a combination of functional specialization for language, task demands related to semantic learning, and asymmetries in the available molecular datasets. These factors should be considered when interpreting the hemispheric distribution of the effects.

This study has several limitations that should be considered. First, although the FLL group showed an increase in MMSE scores, this result should be interpreted with great caution. The MMSE is a brief global screening tool with limited sensitivity to specific cognitive domains, and alternate forms were not used at post‐test, making practice effects a plausible explanation for at least part of the improvement. For these reasons, our discussion focuses on more specific gains in semantic memory, which are theoretically consistent with—and empirically supported by—the observed patterns of subcortico–cortical connectivity change.

Second, the FLL group was older and showed a lower education level and general cognitive decline (e.g., MMSE) at baseline than the no‐intervention group. Although this profile would typically predict a less favorable cognitive trajectory and lower cognitive reserve in aging, in contrast with our findings, these baseline differences represent a limitation and constrain the strength of causal inferences. To mitigate their potential impact, age and years of education were explicitly included as covariates in all GLM analyses, and all cognitive results—particularly those involving global measures—should be interpreted with appropriate caution.

Third, the exploratory design and relatively small sample size limit statistical power and generalizability of the results. In particular, the sample is relatively small for studies involving voxel‐wise fMRI and multiple correlation analyses. However, the observed group differences, particularly in key subcortical clusters, were associated with large effect sizes (Cohen's d > 1.3), indicating that the study was sufficiently powered to detect meaningful changes in the most responsive circuits. Nevertheless, a larger sample would be required to detect more subtle effects and increase the reproducibility of the findings.

Fourth, the absence of long‐term follow‐up precluded the assessment of the durability of cognitive gains, especially in the memory domain. Future longitudinal studies are therefore needed to determine whether these effects persist over time. Fifth, individual variability, such as prior language experience, age of acquisition, and intrinsic motivation, may influence the responsiveness to FLL. Examining these moderating factors could help tailor interventions to individual cognitive profiles.

Related to this, the use of a passive control group represents a major methodological limitation, as we cannot disentangle the specific effects of language learning from non‐specific influences associated with participation in an active, socially embedded intervention. Factors such as increased social engagement, expectancy, structured stimulation, task novelty, and motivational involvement are known to contribute to far‐transfer outcomes and psychological well‐being in older adults and may therefore have partially contributed to the cognitive and neural changes observed in the FLL group. Future studies should incorporate an active control condition—matched for social contact, time commitment, and expectancy—to more precisely isolate the specific contribution of FLL to observed functional and cognitive changes.

Sixth, while the spatial correspondence between rs‐FC changes and molecular expression patterns (e.g., the CB1 receptor, NTRK2 gene) is intriguing, these associations were derived from normative atlases rather than individual‐level data. Therefore, these results should be interpreted as exploratory and hypothesis‐generating, rather than indicative of direct mechanistic relationships.

Seventh, gene expression data were only available for the left hemisphere, which prevented analysis of potential lateralization effects and limited full‐brain transcriptomic mapping. Given the region‐ and hemisphere‐specific expression patterns of many plasticity‐related genes, this represents a notable constraint. Future work incorporating age‐matched PET data or, ideally, subject‐specific receptor imaging would enhance the spatial precision and interpretability of receptor–connectivity associations.

Finally, resting‐state fMRI data were acquired using slightly different protocols across groups due to scanner hardware and software upgrades during data collection. We cannot fully exclude the possibility that such differences may have contributed to the observed patterns of connectivity. To mitigate this issue, all data were processed using a unified preprocessing pipeline (CONN toolbox), including normalization to MNI space and resampling to a common voxel resolution (2 × 2 × 2 mm). The minor difference in TR (0.13 s) resulted in approximately 16 fewer volumes per session for the FLL group (≈7.4% reduction). However, because resting‐state analyses focused on low‐frequency fluctuations (0.008–0.09 Hz), which are adequately sampled under both TRs, this discrepancy is unlikely to have meaningfully affected the results.

In conclusion, this study indicates that a structured FLL intervention is associated with anatomically specific changes in resting‐state functional connectivity (rs‐FC) between subcortical and cortical regions in healthy older adults. The most prominent effects involved the thalamus and the caudate nucleus, two integrative hubs implicated in cortico‐cortical interactions and known to be vulnerable to age‐related decline. The direction and topography of these connectivity changes suggest a discriminating modulation of distinct thalamocortical and striatocortical circuits. Importantly, the spatial correspondence between these neural changes and the cortical distribution of CB1 receptor density and NTRK2 gene expression supports the biological plausibility of FLL‐induced functional plasticity. Overall, the findings highlight the potential of socially embedded interventions such as FLL to engage potential neuroplastic mechanisms in the elderly. Future research should investigate the long‐term stability of these effects and evaluate their generalizability to additional cognitive domains and to broader aging populations. Moreover, multimodal approaches combining resting‐state fMRI with diffusion MRI and task‐based fMRI will be essential to validate the network‐level interpretations proposed here and to delineate how FLL shapes both the structural and functional architecture of the aging brain.

Author Contributions

G.B., C.S., and S.D.P., performed the conceptualization; G.B., F.T., and S.D.P., carried out the formal analysis; G.B. and S.D.P., conducted the investigation; G.B., C.S., and S.D.P., drafted the original manuscript; G.B., conducted the data curation and performed the project administration; G.B., A.F., M.G.P., and S.D.P., provided the resources; F.T., G.A., A.G., A.F., M.G.P., and S.L.S., contributed to writing, review and editing; S.L.S., and S.D.P., supervised the project. All the authors mentioned above made substantial contributions to the content of the paper and approved the final version of the manuscript. All the authors have read and approved the final version of the manuscript and agree with the order of presentation of the authors.

Funding

This work was supported by Ministero dell’Università e della Ricerca (MUR), Departments of Excellence. European Commission, D73C22000840006, D73C22001930006.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: The seed‐based connectivity analysis results.

Table S2: Effect Sizes.

Table S3: BrainSMASH output for the association between the rs‐FC map of the left thalamus and the selected receptor maps.

Table S4: BrainSMASH output for the association between the rs‐FC map of the right thalamus and the selected receptor maps.

Table S5: BrainSMASH output for the association between the rs‐FC map of the left caudate and the selected gene maps.

Table S6: BrainSMASH output for the association between the rs‐FC map of the right pallidum and the selected gene maps.

Table S7: BrainSMASH output for the association between the rs‐FC map of the right pallidum and the selected gene maps.

Table S8: BrainSMASH output for the association between the rs‐FC map of the right thalamus and the selected gene maps.

Table S9: BrainSMASH output for the association between the rs‐FC map of the left caudate and the selected gene maps.

Table S10: BrainSMASH output for the association between the rs‐FC map of the right pallidum and the selected gene maps.

Table S11: Mean and Standard deviation for baseline demographic characteristics (age, sex, education) and baseline cognitive scores.

AGM2-9--s001.docx^{(39.3KB, docx)}

Acknowledgments

A.F., and S.L.S., acknowledge financial support from the Strengthening of Research Structures and the Creation of RandD Innovation Ecosystems initiative under the National Recovery and Resilience Plan (NRRP), Mission 4, Component 2, Investment 1.5, funded by the European Union–NextGenerationEU–VITALITY, ECS00000041 (D73C22000840006). G.B., acknowledges partial funding from the Departments of Excellence program of the Ministero dell'Università e della Ricerca (MUR). This work was also partially supported by the Departments of Excellence 2023–2027 initiative of the Italian Ministry of Education, University, and Research for the Department of Neuroscience, Imaging, and Clinical Sciences (DNISC) at the University of Chieti–Pescara. Additionally, support was provided by the NRRP—Mission 4 Education and Research—Component 2 From Research to Business—Investment 1.2, through the Funding Projects for Young Researchers–Seal of Excellence, Next Generation EU (SOE_0000062, CUP D73C22001930006). Finally, we sincerely thank all the participants for their time, commitment, and willingness to contribute to this research. Their dedication was invaluable to advancing our understanding of foreign language learning and brain plasticity. Open access publishing facilitated by Universita degli Studi Gabriele d'Annunzio Chieti Pescara, as part of the Wiley ‐ CRUI‐CARE agreement.

Contributor Information

Giovanna Bubbico, Email: giovanna.bubbico@unich.it.

Stefano Delli Pizzi, Email: stefano.dellipizzi@unich.it.

References

1. Antoniou M., Gunasekera G. M., and Wong P. C., “Foreign Language Training as Cognitive Therapy for Age‐Related Cognitive Decline: A Hypothesis for Future Research,” Neuroscience and Biobehavioral Reviews 37, no. 2 (2013): 2689–2698, 10.1016/j.neubiorev.2013.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
2. Bubbico G., Tomaiuolo F., Sestieri C., et al., “Foreign Language Learning in Older Adults Modifies the Pattern of Resting‐State Activity in the Medial Prefrontal Cortex,” Journal of the Neurological Sciences 475 (2025): 123568, 10.1016/j.jns.2025.123568. [DOI] [PubMed] [Google Scholar]
3. Bubbico G., Tomaiuolo F., Sestieri C., et al., “Learning a Foreign Language in Older Adults Shapes the Functional Connectivity of Distinct Cerebellar Sub‐Regions With Cortical Areas Rich in CB1 Receptor Expression,” Brain and Behavior 15, no. 5 (2025): 70565, 10.1002/brb3.70565. [DOI] [PMC free article] [PubMed] [Google Scholar]
4. Teubner‐Rhodes S., K. I. Vaden, Jr. , Dubno J. R., and Eckert M. A., “Cognitive Persistence: Development and Validation of a Novel Measure From the Wisconsin Card Sorting Test,” Neuropsychologia 102 (2017): 95–108, 10.1016/j.neuropsychologia.2017.05.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
5. Perani D., Farsad M., Ballarini T., et al., “The Impact of Bilingualism on Brain Reserve and Metabolic Connectivity in Alzheimer's Dementia,” Proceedings of the National Academy of Sciences 114, no. 7 (2017): 1690–1695, 10.1073/pnas.1610909114. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Bubbico G., Chiacchiaretta P., Parenti M., et al., “Effects of Second Language Learning on the Plastic Aging Brain: Functional Connectivity, Cognitive Decline, and Reorganization,” Frontiers in Neuroscience 13 (2019): 423, 10.3389/fnins.2019.00423. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Janacsek K., Evans T. M., Kiss M., Shah L., Blumenfeld H., and Ullman M. T., “Subcortical Cognition: The Fruit Below the Rind,” Annual Review of Neuroscience 45 (2022): 361–386, 10.1146/annurev-neuro-110920-013544. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Scoville W. B. and Milner B., “Loss of Recent Memory After Bilateral Hippocampal Lesions,” Journal of Neurology, Neurosurgery and Psychiatry 20, no. 1 (1957): 11–21, 10.1136/jnnp.20.1.11. [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Squire L. R., “Memory and the Hippocampus: A Synthesis From Findings With Rats, Monkeys, and Humans,” Psychological Review 99, no. 2 (1992): 195–231, 10.1037/0033-295X.99.2.195. [DOI] [PubMed] [Google Scholar]
10. Tulving E., “Memory and Consciousness,” Canadian Psychology 26, no. 1 (1985): 1–12, 10.1037/h0080017. [DOI] [Google Scholar]
11. Leal S. L. and Yassa M. A., “Neurocognitive Aging and the Hippocampus Across Species,” Trends in Neurosciences 38, no. 12 (2015): 800–812, 10.1016/j.tins.2015.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Shohamy D., “Learning and Motivation in the Human Striatum,” Current Opinion in Neurobiology 21, no. 3 (2011): 408–414, 10.1016/j.conb.2011.05.009. [DOI] [PubMed] [Google Scholar]
13. Cox K. M., Aizenstein H. J., and Fiez J. A., “Striatal Outcome Processing in Healthy Aging,” Cognitive, Affective, and Behavioral Neuroscience 8, no. 3 (2008): 304–317, 10.3758/CABN.8.3.304. [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Fama R. and Sullivan E. V., “Thalamic Structures and Associated Cognitive Functions: Relations With Age and Aging,” Neuroscience and Biobehavioral Reviews 54 (2015): 29–37, 10.1016/j.neubiorev.2015.03.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
15. Choi S. H., Bylykbashi E., Chatila Z. K., et al., “Combined Adult Neurogenesis and BDNF Mimic Exercise Effects on Cognition in an Alzheimer's Mouse Model,” Science 361, no. 6406 (2018): eaan8821, 10.1126/science.aan8821. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Davis M. and Whalen P. J., “The Amygdala: Vigilance and Emotion,” Molecular Psychiatry 6, no. 1 (2001): 13–34, 10.1038/sj.mp.4000812. [DOI] [PubMed] [Google Scholar]
17. LeDoux J. E., The Emotional Brain: The Mysterious Underpinnings of Emotional Life (Simon and Schuster, 1996). [Google Scholar]
18. Padulo C., Sestieri C., Punzi M., et al., “Atrophy of Specific Amygdala Subfields in Subjects Converting to Mild Cognitive Impairment,” Alzheimer's and Dementia (New York) 9, no. 4 (2023): e12436, 10.1002/trc2.12436. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Delli Pizzi S., Punzi M., and Sensi S. L., “Functional Signature of Conversion of Patients With Mild Cognitive Impairment,” Neurobiology of Aging 74 (2019): 21–37, 10.1016/j.neurobiolaging.2018.10.004. [DOI] [PubMed] [Google Scholar]
20. Delli Pizzi S., Granzotto A., Bomba M., Frazzini V., Onofrj M., and Sensi S. L., “Acting Before; A Combined Strategy to Counteract the Onset and Progression of Dementia,” Current Alzheimer Research 17, no. 9 (2020): 790–804, 10.2174/1567205017666201203085524. [DOI] [PubMed] [Google Scholar]
21. Perry B. A. L., Mendez J. C., and Mitchell A. S., “Cortico‐Thalamocortical Interactions for Learning, Memory and Decision‐Making,” Journal of Physiology 601, no. 1 (2023): 25–35, 10.1113/JP282626. [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Crosson B., “The Role of the Thalamus in Declarative and Procedural Linguistic Memory Processes,” Frontiers in Psychology 12 (2021): 682199, 10.3389/fpsyg.2021.682199. [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Green D. W. and Abutalebi J., “Language Control in Bilinguals: The Adaptive Control Hypothesis,” Journal of Cognitive Psychology 25, no. 5 (2013): 515–530, 10.1080/20445911.2013.796377. [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Freedman L., Zivan M., Farah R., and Horowitz‐Kraus T., “Greater Functional Connectivity Within the Cingulo‐Opercular and Ventral Attention Networks Is Related to Better Fluent Reading,” NeuroImage: Clinical 26 (2020): 102214, 10.1016/j.nicl.2020.102214. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. K. I. Vaden, Jr. , Kuchinsky S. E., Cute S. L., Ahlstrom J. B., Dubno J. R., and Eckert M. A., “The Cingulo‐Opercular Network Provides Word‐Recognition Benefit,” Journal of Neuroscience 33, no. 48 (2013): 18979–18986, 10.1523/JNEUROSCI.1417-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Dosenbach N. U. F., Visscher K. M., Palmer E. D., Raichle M. E., and Gordon E. M., “The Brain's Action‐Mode Network,” Nature Reviews Neuroscience 26 (2025): 158–168, 10.1038/s41583-024-00895-x. [DOI] [PubMed] [Google Scholar]
27. Dosenbach N. U. F., Visscher K. M., Palmer E. D., et al., “A Core System for the Implementation of Task Sets,” Neuron 50, no. 5 (2006): 799–812, 10.1016/j.neuron.2006.04.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Delli Pizzi S., Tomaiuolo F., Ferretti A., et al., “Modulation of Cerebellar‐Cortical Connectivity Induced by Modafinil and Its Relationship With Receptor and Transporter Expression,” Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 11 (2024): 10, 10.1016/j.bpsc.2024.11.010. [DOI] [PubMed] [Google Scholar]
29. Burke S. N. and Barnes C. A., “Neural Plasticity in the Ageing Brain,” Nature Reviews Neuroscience 7, no. 1 (2006): 30–40, 10.1038/nrn1809. [DOI] [PubMed] [Google Scholar]
30. Bilkei‐Gorzo A., “The Endocannabinoid System in Normal and Pathological Brain Ageing,” Philosophical Transactions of the Royal Society B 367 (2012): 3326–3341, 10.1098/rstb.2011.0388. [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Bilkei‐Gorzo A., Albayram O., Draffehn A., et al., “A Chronic Low Dose of Delta(9)‐tetrahydrocannabinol (THC) Restores Cognitive Function in Old Mice,” Nature Medicine 23 (2017): 782–787, 10.1038/nm.4311. [DOI] [PubMed] [Google Scholar]
32. Rissman R. A., De Blas A. L., and Armstrong D. M., “GABA(A) Receptors in Aging and Alzheimer's Disease,” Journal of Neurochemistry 103 (2007): 1285–1292, 10.1111/j.1471-4159.2007.04832.x. [DOI] [PubMed] [Google Scholar]
33. Rissman R. A. and Mobley W. C., “Implications for Treatment: GABAA Receptors in Aging, Down Syndrome and Alzheimer's Disease,” Journal of Neurochemistry 117 (2011): 613–622, 10.1111/j.1471-4159.2011.07237.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Hunt D. L. and Castillo P. E., “Synaptic Plasticity of NMDA Receptors: Mechanisms and Functional Implications,” Current Opinion in Neurobiology 22 (2012): 496–508, 10.1016/j.conb.2012.01.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Carvalho T. G., Alves‐Silva J., de Souza J. M., et al., “Metabotropic Glutamate Receptor 5 Ablation Accelerates Age‐Related Neurodegeneration and Neuroinflammation,” Neurochemistry International 126 (2019): 218–228, 10.1016/j.neuint.2019.03.020. [DOI] [PubMed] [Google Scholar]
36. Lu W., Shi Y., Jackson A. C., et al., “Subunit Composition of Synaptic AMPA Receptors Revealed by a Single‐Cell Genetic Approach,” Neuron 62, no. 2 (2009): 254–268, 10.1016/j.neuron.2009.02.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Korb E. and Finkbeiner S., “Arc in Synaptic Plasticity: From Gene to Behavior,” Trends in Neurosciences 34, no. 11 (2011): 591–598, 10.1016/j.tins.2011.08.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Cohen‐Cory S., “Brain‐Derived Neurotrophic Factor and the Development of Structural Synaptic Connectivity,” Developmental Neurobiology 70, no. 5 (2010): 271–288, 10.1002/dneu.20774. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Castrén E. and Rantamäki T., “Role of Brain‐Derived Neurotrophic Factor in the Aetiology of Depression: Implications for Pharmacological Treatment,” CNS Drugs 24, no. 1 (2010): 1–7, 10.2165/11530010-000000000-00000. [DOI] [PubMed] [Google Scholar]
40. Edmonds E. C., Delano‐Wood L., Clark L. R., et al., “Susceptibility of the Conventional Criteria for Mild Cognitive Impairment to False‐Positive Diagnostic Errors,” Alzheimer's and Dementia 11, no. 4 (2015): 415–424, 10.1016/j.jalz.2014.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
41. Edmonds E. C., Eppig J., Bondi M. W., et al., “Heterogeneous Cortical Atrophy Patterns in MCI Not Captured by Conventional Diagnostic Criteria,” Neurology 87, no. 20 (2016): 2108–2116, 10.1212/WNL.0000000000003326. [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Mapstone M., Cheema A. K., Fiandaca M. S., et al., “Plasma Phospholipids Identify Antecedent Memory Impairment in Older Adults,” Nature Medicine 20, no. 4 (2014): 415–418, 10.1038/nm.3466. [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Whitfield‐Gabrieli S. and Nieto‐Castanon A., “Conn: A Functional Connectivity Toolbox for Correlated and Anticorrelated Brain Networks,” Brain Connectivity 2, no. 3 (2012): 125–141, 10.1089/brain.2012.0073. [DOI] [PubMed] [Google Scholar]
44. Tomaiuolo F., Delli Pizzi S., Sestieri C., et al., “Data‐Driven Analysis of Whole‐Brain Intrinsic Connectivity in Patients With Chronic Low Back Pain Undergoing Osteopathic Manipulative Treatment,” NeuroImage: Clinical 43 (2024): 103659, 10.1016/j.nicl.2024.103659. [DOI] [PMC free article] [PubMed] [Google Scholar]
45. Rolls E. T., Huang C.‐C., Lin C.‐P., Feng J., and Joliot M., “Automated Anatomical Labelling Atlas 3,” NeuroImage 206 (2020): 116189, 10.1016/j.neuroimage.2019.116189. [DOI] [PubMed] [Google Scholar]
46. Hansen J. Y., Shafiei G., Markello R. D., et al., “Mapping Neurotransmitter Systems to the Structural and Functional Organization of the Human Neocortex,” Nature Neuroscience 25, no. 11 (2022): 1569–1581, 10.1038/s41593-022-01186-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
47. Schaefer A., Kong R., Gordon E. M., et al., “Local‐Global Parcellation of the Human Cerebral Cortex From Intrinsic Functional Connectivity MRI,” Cerebral Cortex 28, no. 9 (2018): 3095–3114, 10.1093/cercor/bhx179. [DOI] [PMC free article] [PubMed] [Google Scholar]
48. Burt J. B., Helmer M., Shinn M., Anticevic A., and Murray J. D., “Generative Modeling of Brain Maps With Spatial Autocorrelation,” NeuroImage 220 (2020): 117038, 10.1016/j.neuroimage.2020.117038. [DOI] [PubMed] [Google Scholar]
49. Worsley K. J., Marrett S., Neelin P., Vandal A. C., Friston K. J., and Evans A. C., “A Unified Statistical Approach for Determining Significant Signals in Images of Cerebral Activation,” Human Brain Mapping 4, no. 1 (1996): 58–73, 10.1002/(SICI)1097-0193(1996)4:1<58::AID-HBM4>3.0.CO;2-O. [DOI] [PubMed] [Google Scholar]
50. Chumbley J., Worsley K., Flandin G., and Friston K., “Topological FDR for neuroimaging,” NeuroImage 49, no. 4 (2010): 3057–3064, 10.1016/j.neuroimage.2009.10.090. [DOI] [PMC free article] [PubMed] [Google Scholar]
51. Raichle M. E., MacLeod A. M., Snyder A. Z., Powers W. J., Gusnard D. A., and Shulman G. L., “A Default Mode of Brain Function,” Proceedings of the National Academy of Sciences 98, no. 2 (2001): 676–682, 10.1073/pnas.98.2.676. [DOI] [PMC free article] [PubMed] [Google Scholar]
52. Guo C. C., Tan R., Hodges J. R., et al., “Network‐Selective Vulnerability of the Human Cerebellum to Alzheimer's Disease and Frontotemporal Dementia,” Brain 139, no. 5 (2016): 1527–1538, 10.1093/brain/aww003. [DOI] [PMC free article] [PubMed] [Google Scholar]
53. Shine J. M., Lewis L. D., Garrett D. D., and Hwang K., “The Impact of the Human Thalamus on Brain‐Wide Information Processing,” Nature Reviews Neuroscience 24, no. 7 (2023): 416–430, 10.1038/s41583-023-00701-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
54. Guedj C. and Vuilleumier P., “Functional Connectivity Fingerprints of the Human Pulvinar: Decoding Its Role in Cognition,” NeuroImage 221 (2020): 117162, 10.1016/j.neuroimage.2020.117162. [DOI] [PubMed] [Google Scholar]
55. Yuan R., Di X., Taylor P., et al., “Functional Topography of the Thalamocortical System in Humans,” Brain Structure and Function 221, no. 4 (2015): 1971–1984, 10.1007/s00429-015-1018-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
56. Davis S. W., Dennis N. A., Daselaar S. M., Fleck M. S., and Cabeza R., “Que PASA? The Posterior–Anterior Shift in Aging,” Cerebral Cortex 18, no. 5 (2008): 1201–1209, 10.1093/cercor/bhm155. [DOI] [PMC free article] [PubMed] [Google Scholar]
57. Hwang K. and D'Esposito M., “The Thalamus in Cognitive Control,” in The Thalamus: Volume 1. Organization, Development, and Function, vol. 16, ed. Halassa M. M. (Cambridge University Press, 2022), 10.1017/9781108674287.017. [DOI] [Google Scholar]
58. Basaia S., Zavarella M., Rugarli G., et al., “Caudate Functional Networks Influence Brain Structural Changes With Aging,” Brain Communications 6, no. 2 (2024): 116, 10.1093/braincomms/fcae116. [DOI] [PMC free article] [PubMed] [Google Scholar]
59. Marquand A. F., Haak K. V., and Beckmann C. F., “Functional Corticostriatal Connection Topographies Predict Goal Directed Behaviour in Humans,” Nature Human Behaviour 1, no. 8 (2017): 0146, 10.1038/s41562-017-0146. [DOI] [PMC free article] [PubMed] [Google Scholar]
60. Rieckmann A., Johnson K. A., Sperling R. A., Buckner R. L., and Hedden T., “Dedifferentiation of Caudate Functional Connectivity and Striatal Dopamine Transporter Density Predict Memory Change in Normal Aging,” Proceedings of the National Academy of Sciences 115, no. 40 (2018): 10160–10165, 10.1073/pnas.1804641115. [DOI] [PMC free article] [PubMed] [Google Scholar]
61. O'Rawe J. F. and Leung H.‐C., “Topographic Organization of the Human Caudate Functional Connectivity and Age‐Related Changes With Resting‐State fMRI,” Frontiers in Systems Neuroscience 16 (2022): 966433, 10.3389/fnsys.2022.966433. [DOI] [PMC free article] [PubMed] [Google Scholar]
62. Karube F., Takahashi S., Kobayashi K., and Fujiyama F., “Motor Cortex Can Directly Drive the Globus Pallidus Neurons in a Projection Neuron Type‐Dependent Manner in the Rat,” eLife 8 (2019): e49511, 10.7554/eLife.49511. [DOI] [PMC free article] [PubMed] [Google Scholar]
63. Haber S. N. and Calzavara R., “The Cortico‐Basal Ganglia Integrative Network: The Role of the Thalamus,” Brain Research Bulletin 78 (2008): 69–74, 10.1016/j.brainresbull.2008.09.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
64. Domínguez‐Borràs J. and Vuilleumier P., “Amygdala Function in Emotion, Cognition, and Behavior,” Handbook of Clinical Neurology 187 (2022): 359–380, 10.1016/B978-0-12-823493-8.00015-8. [DOI] [PubMed] [Google Scholar]
65. Ries S. K., Dronkers N. F., and Knight R. T., “Choosing Words: Left Hemisphere, Right Hemisphere, or Both? Perspectives on the Lateralization of Word Retrieval,” Annals of the New York Academy of Sciences 1369, no. 1 (2016): 111–131, 10.1111/nyas.12993. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials