Preserved topography, lateralization, selectivity, and functional connectivity of the language network in older brains

Nature Communications Preserved topography, lateralization, selectivity, and functional connectivity of the language network in older brains

Healthy aging is associated with structural and functional brain changes. However, cognitive abilities vary in how they change with age: executive functions decline, while aspects of linguistic processing remain relatively preserved. This heterogeneity predicts differences among brain networks in whether and how they change with age. To evaluate this prediction, we used precision fMRI to examine the language-selective network and the Multiple Demand network, which supports executive functions, in older adults (N equals sixty-four) relative to young controls (N equals four hundred eighty-three). The Multiple Demand network of older adults shows weaker, less spatially extensive, and more topographically variable activations during an executive function task and reduced within-network functional connectivity. In contrast, we find remarkable preservation of the language network in older adults: it responds during language comprehension as strongly and selectively as in younger adults, with similar left-hemispheric lateralization and within-network functional connectivity. These findings align with behavioral preservation of language comprehension in healthy aging.

Healthy aging is associated with slower information processing and a reduction in working memory and attentional resources. These behavioral changes in executive abilities have been argued to result from changes in the structure and function of different large-scale brain networks. Some have reported age-related changes in the brain areas that support executive functions. Others have argued for more widespread effects on multiple networks. For example, one popular claim has been that brain networks become generally less clearly segregated with age. This reduced segregation can manifest as less clearly distinguishable functional profiles -de-differentiation-due to a combination of weaker responses in functionally specialized areas and compensatory recruitment of brain regions not typically engaged in those functions in younger individuals (often, the frontal executive regions). Reduced segregation can also manifest as weaker dissociations in functional connectivity, with reduced within-network connectivity and increased between-network connectivity.

In contrast to the well-documented changes in executive abilities, language is often described as one of the better-preserved functions in older adults. In spite of occasional word-finding difficulties and changes in the organization of semantic memory, vocabulary keeps growing with age, the ability to extract meaning from texts continues to improve, and the ability to predict upcoming words remains relatively intact. However, whether and how the neural infrastructure for language processing changes with age remains debated. For example, some researchers have argued that the general topography of the language network-a set of temporal and frontal brain areas that respond selectively during language processing-is similar between younger and older adults, but others have argued for more bilateral activation with age. Similarly, the magnitude of response in some language regions has been argued to increase with age, to decrease with age, or to show no change. And for within-network connectivity, some have reported a general reduction with age, others an increase or a decrease depending on the specific pair of brain regions, yet others found no change.

These inconsistencies could be due to several factors. First, past studies have used an array of language tasks-including object naming, auditory sentence comprehension, monitoring of a target word, and passive narrative reading-making across-study comparisons challenging. Moreover, many commonly used paradigms conflate language processing and general task demands. Because such paradigms recruit both the language-selective network and the domain-general Multiple Demand network, which supports executive functions, the observed effects are difficult to attribute to a particular network. This issue is compounded by the reliance of most past fMRI studies on group analyses, where activation maps or functional connectivity maps are averaged voxel-wise across participants. Because of substantial inter-individual variability in the precise locations of functional areas, including language areas and Multiple Demand areas, and the proximity of the two networks to each other in the association cortex, this group-averaging approach blurs network boundaries, leading to interpretive challenges. This blurring of functional boundaries, inherent in the group-analytic approach, is the reason that the field of cognitive neuroscience has been increasingly moving toward 'precision imaging', where the relevant functional areas are identified within individuals, either using established 'localizer' paradigms or bottom-up clustering of voxel timecourses based on functional connectivity data.

To resolve some of the controversies about whether or how the language network changes with age, we here use a precision imaging approach with a 'localizer' paradigm that has been established in prior studies to reliably identify the language network, which selectively supports comprehension and production. For comparison, we also examine the domain-general Multiple Demand network, which supports executive abilities, such as attention, working memory, and cognitive control-functions that are widely agreed to exhibit age-related decline. Examining these two cognitive networks in parallel enables a direct test of whether age-related brain changes are uniform across networks or instead exhibit network-specific profiles. In addition, this approach allows us to evaluate the claim about network de-differentiation introduced above. In line with increasing emphasis in the field on robustness and replicability, we i) examine a relatively large cohort of older adults (O A, N equals sixty-four), along with a large cohort of young adults (Y A, N equals four hundred eighty-three); ii) include several commonly used measures of brain function, and iii) use several statistical analytic approaches. To foreshadow our results, we find that a) for the language network, the topography, including the extent of activation and degree of lateralization, as well as the strength of within-network connectivity are similar between older and younger adults, and only the magnitude of response to language slightly increases with age, and b) for the Multiple Demand network-which remains robustly segregated from the language network in older adults-the topography is more variable across individuals and the magnitude and extent of activation as well as the within-network connectivity decrease substantially with age.

Results

Our participants consisted of sixty-four older adults (average age: fifty-nine point seven; range forty-one to eighty; thirty-six females; fifty-nine right-handed, three left-handed, one ambidextrous and one with missing handedness data) and four hundred eighty-three younger adults (average age: twenty-three point eight; range seventeen to thirty-nine; two hundred seventy-eight females; four hundred thirty-one right-handed, twenty-eight left-handed, eleven ambidextrous and thirteen with missing handedness data). The older adult cohort includes data from two independent cohorts that were combined to increase statistical power. General cognitive functioning was assessed in one of the older adult cohorts using the MMSE-Two Thousand One, and all participants scored within the normal range. Each participant completed two extensively validated localizer tasks, which reliably and selectively engage the target networks of interest-the language and the multiple demand networks-in spite of their close proximity within the left frontal lobe. A subset of the participants additionally completed one or two so-called 'naturalistic-cognition' paradigms: a resting state scan and a story listening paradigm. These data were used in the analyses of functional connectivity, and the story listening data were used to test for sensitivity to linguistic complexity in the language network's responses.

Before presenting the critical results, we describe the two brain networks of interest in more detail. The language network, as defined here, is a set of left-lateralized frontal and temporal brain regions that jointly support linguistic processing, including lexical retrieval, syntactic structure building, and semantic composition. Although this network necessarily interacts with other brain systems during real-life language use, its regions are strongly selective for language over diverse non-linguistic inputs and tasks. The Multiple Demand network is a set of bilateral frontal and parietal brain regions that jointly support executive functions, including attention, working memory, and cognitive control. The Multiple Demand network is engaged by diverse cognitively demanding tasks, across stimulus and task types. It is important to note that although we use particular localizers to define our networks of interest, each of these paradigms has been shown to generalize across many variants that use alternative materials and tasks. Furthermore, both of these networks can also be defined in individual participants based on patterns of functional connectivity, and the networks thus identified correspond closely with the activations elicited by validated localizers, like those used here.

In the analyses, we focus on the neural measures that have been most commonly examined in past studies. In particular, for both networks, we first consider the typicality of the activation patterns by comparing individual maps to probabilistic activation atlases from large numbers of participants scanned on the same paradigms in prior studies. We additionally examine changes in the extent of activation and, for the language network only, the degree of lateralization. Lastly, we investigate the magnitude of neural response for the relevant contrasts and the strength of inter-regional functional correlations during the naturalistic paradigms. To ensure robustness to analytic choices, we report both categorical effects of age group and continuous effects of age across the full set of participants.