# Cognitive dedifferentiation in later life: longitudinal findings from the Lothian Birth Cohort 1936

## TL;DR

This study shows that cognitive skills become more interconnected as people age, using data from individuals assessed over time.

## Contribution

The study provides longitudinal evidence of cognitive dedifferentiation and its link to cognitive decline in older adults.

## Key findings

- General cognitive functioning explained increasing variance in cognitive tests as participants aged.
- Cognitive domains like fluid skills converged, while crystallized ability became less influential over time.
- Group-level dedifferentiation closely mirrored individual-level cognitive dispersion with age.

## Abstract

In the cognitive aging literature, the dedifferentiation hypothesis refers to cognitive skills becoming more interrelated in older adulthood. Here, we report evidence for cognitive dedifferentiation in the Lothian Birth Cohort 1936 (LBC1936).

The LBC1936 is a narrow-age cohort assessed at 5 waves between ages 70 and 82. We analyzed data from 418 participants (49% male) who provided cognitive data at all 5 waves.

In single-order structural equation models, the percentage of variance that general cognitive functioning (g) accounted for across 13 cognitive tests increases by wave; w1 to w5: 25%, 27%, 29%, 31%, 36%, and the group-level rate of dedifferentiation closely tracked the group-level rate of cognitive decline (r  =  −.991, p = .001). A hierarchical model, which included 4 cognitive domains as mid-level factors, provides evidence of cognitive dedifferentiation at the cognitive domain level: fluid cognitive domains (Visuospatial Skills, Processing Speed, and Verbal Memory) converged, and Crystallised Ability became less influential on the structure of g over time. We also show that this group-level measure of dedifferentiation reflects the individual-level measure of dispersion (people tend to score more similarly across different cognitive tests with advancing age), r  =  −.989, p = .001.

The current results have implications for longitudinal g modeling choices: it cannot be assumed that g’s composition is the same over time. Future longitudinal research will be important in clarifying the incremental validity, determinants, mechanisms, and implications of cognitive differentiation and dedifferentiation across the lifespan.