A network perspective on cognition in individuals with Parkinson's disease

In neuropsychological diagnostics, the assignment of cognitive tests to domains is usually not empirically based. Hence, we aimed to assess the dimensionality structure of cognition in individuals with Parkinson's disease (PD) and conceptually replicate the findings in cognitively healthy individuals (CHIs).

We performed Exploratory Graph Analysis (EGA) for dimensionality analysis of cognitive test scores in N = 698 individuals with PD from the DEMPARK/LANDSCAPE study. Redundancy was reduced based on Unique Variable Analysis (UVA) before re-performing EGA. CHI data (N = 60,398) served as a conceptual replication base.

EGA identified five dimensions. After removing redundancy identified by UVA, EGA identified a unidimensional structure of cognitive test scores. The findings were conceptually replicated in CHIs.

The findings imply the need to re-evaluate the composition of cognitive test batteries to reduce redundancy and improve the validity of cognitive diagnostics. Cognition may be better described as a network of interrelated cognitive functions rather than a factorial structure of latent cognitive domains.

Cognitive test scores of the same paradigm were strongly associated with each other.This finding indicates redundancy in the cognitive test battery.After removing redundancy, scores were best represented by unidimensional structures.The findings in Parkinson's disease were conceptually replicated in healthy controls.The results suggest that cognition should be viewed as a complex "network" of interrelated functions.