The developmental trajectory and compromise of social cognition

Social cognition is a critical mental capacity that allows us to understand and interact with others. Basic social cognitive skills emerge early in infancy, such as joint attention and responsiveness to facial expressions. As we grow, more sophisticated abilities develop, like emotion recognition and mentalizing (the ability to reason about how others think and feel). Adulthood sees the consolidation of these skills before diminishing in old age. Social cognition shapes our interaction within familial, peer, and societal contexts. Alterations in social cognition may be manifested as a wide range of behavioral disturbances, like inappropiate or insensitive responses, and lead to loneliness and social isolation, increasing morbidity and mortality.

Social cognition as a transdiagnostic construct

Multiple clinical conditions, such as autism, depression, and dementia, show significant alterations in social cognition, highlighting its potential as a transdiagnostic marker. Social cognition tasks may inform diagnosis, disease progression, and treatment evaluation. In age-related clinical conditions, social cognition tasks hold promise for the differential diagnosis of neurodegenerative and neuropsychiatric diseases.

Interindividual differences in social cognition and the need for diversity

The high interindividual variability in social cognition task performance is a critical issue to advance the field. Beyond age and diagnosis, other factors have been shown to impact social cognition. Among them are reduced general cognitive abilities, a lower level of education, and lower socioeconomic status.

However, these findings are not without controversy. For instance, evidence also shows that social cognition is independent of general cognitive decline in older adults. These controversies limit our knowledge of social cognition processes and the applicability of current assessments.

A more comprehensive and accurate understanding of the interplay between various factors and social cognition requires including diverse participants in research studies. Social cognition varies across cultural contexts, races, and ethnicities. Persons from underrepresented and underserved backgrounds show non-stereotypical profiles in terms of cognitive decline, educational attainment, and socioeconomic status, and disparity-related factors likely exert a more substantial influence on social cognition than traditional factors.

The present study

Our work was designed to assess combined predictors of social cognition in a diverse sample. We gathered 1,063 participants between 50 and 98 years from nine countries across Latin America and Europe participating in the Multi-Partner Consortium to Expand Dementia Research in Latin America (ReDLat) and other international networks. We included persons with unimpaired cognition and with common pathological conditions associated with aging, namely, subjective cognitive complaints, mild cognitive impairment, Alzheimer’s Disease, and the behavioral variant of frontotemporal dementia.

Participants completed the Mini-Social Cognition and Emotional Assessment (Mini-SEA), a social cognition battery that includes emotion recognition and mentalizing tasks. They also provided clinical and sociodemographic information and performed general cognitive and executive tasks. A subsample of them underwent structural and functional magnetic resonance imaging recordings. These data were used to estimate the brain reserve. We calculated the grey matter volume and functional connectivity of regions that support social cognition and are vulnerable to dementias; the salience network and the default mode network.

We performed a series of machine learning support vector regression models, including diagnosis, demographics, cognitive, and brain reserve features as predictors of social cognition. Simple regression analysis and pairwise comparisons showed the expected effects of advanced age and clinical diagnosis in diminishing social cognition (Figure 1). However, feature selection from machine learning models revealed that age and diagnosis were not the primary predictors of social cognition when assessed jointly with other factors.

Figure 1. Study design and traditional effects of age and diagnosis on social cognition. (A) i. Participants were recruited from high-income and upper-middle-income countries through the ReDLat dementia consortium, the International Network on Social Condition Disorders, and the Geroscience Center for Brain Health and Metabolism. ii. Diagnosis, demographics, cognition, grey matter volume fMRI resting-state functional connectivity of brain networks, and in-scanner motion artifacts were entered into computational models as predictors of social cognition. iii. Data were harmonized across countries. iv. Data were analyzed using support vector regression models with Bayesian optimization for hyperparameter tuning in 70-30 train and test partition and using a bootstrap approach. v. Outcome variables were facial emotion recognition, mentalizing, and a social cognition total score from the Mini-Social Cognition and Emotional Assessment (Mini-SEA) battery. (B) Age significantly predicted worse performance in emotion recognition, mentalizing, and the total score. (C) Participants with mild cognitive impairment (MCI), Alzheimer’s disease (AD), and behavioral variant frontotemporal dementia (bvFTD) performed significantly worse in social cognition relative to the groups of healthy controls (HCs) and subjective cognitive complaints (SCC). 

Cognitive/executive features and education were more important in explaining social cognition variance. The predominance of these features also persisted in most models that included brain reserve features. Socioeconomic status (i.e., country income following the World Bank classification) was also a significant predictor in some models (Figure 2). 

Figure 2. Support vector regression results. (A) Models including diagnosis, demographics, and cognition as predictors of social cognition performance. (B) Models include one level of brain reserve (grey matter volume) and behavioral features as predictors of social cognition performance. (C) Models include resting-state functional connectivity features (brain networks and motion artifacts) as predictors of social cognition performance and behavioral and grey matter volume predictors. DMN: default mode network, EN: executive network, MN: motor network, SN: salience network, VN: visual network, vol: volume.

Take-home messages

The study showed that social cognition in aging is shaped by a heterogeneous array of cognitive and sociodemographic factors beyond the diagnosis and associated brain signatures. A crucial strength of the study is using a large, culturally diverse, and non-stereotypical sample.

The findings have implications for developing tailored predictive models and a more nuanced approach to social cognition assessment. Reducing cognitive demands in stimuli, accounting for potential attentional/comprehension issues, developing culturally-tailored designs, and creating norms adjusted by years of education and country are some steps that may be taken to enhance clinical tools for social cognition assessment in neurodegenerative diseases and other conditions. Future global approaches to social cognition aiming for more accurate predictions and translations must account for sample diversity.

Note: Macarena Espina Díaz designed the poster image.