Links among cognitive test scores shifted modestly with training
A secondary analysis of a randomized trial in 4,471 adults found modest changes in test-score links over about 70 days, differing by starting performance.
Limit: Only 50.9% completed follow-up, potentially disrupting the balance from randomization
Edited by Boris Djordjevic · AgingResearch.News
The analysis included 4,471 adults aged 18–80 from a nonclinical population recruited through the Lumosity website. Their mean age was 38.7 years; 2,404 were female. The original randomized trial compared daily Lumosity game sessions, typically about 15 minutes, with 15-minute crossword sessions. Participants took seven cognitive tests before training and about 70 days later.
Researchers grouped participants by starting performance and mapped statistical links among their test scores. Both activities were associated with higher overall scores. Among lower performers, Lumosity users gained more than crossword users. Lower-performing Lumosity users’ score networks showed greater overall integration—a measure of how efficiently different parts of a network connect. Higher-performing participants showed less clustering and lower efficiency of connections within smaller parts of their networks. The network changes were modest.
Why it matters
The authors suggest that measures of how cognitive skills relate to one another may help investigate why people differ in their vulnerability to age-related cognitive decline.
Caveats
This exploratory secondary analysis used a retained subset of the randomized sample, so randomization may not have been fully preserved. The networks captured group-level test-score relationships, not brain connections, in nonclinical website users.
From the paper
- Participants
After outlier removal, 4471 participants were included in the analysis (age 38.7 ± 15.1, range = 18–80; 2404 female).
- Who took part
This randomized controlled trial (NCT-02367898) involved a nonclinical population recruited through the Lumosity website (www.lumosity.com).
- Duration
Participants were instructed to train for 10 weeks, after which they were invited to complete the post-assessment battery.
- Analyzed for cognitive function
Cluster 1 included 2767 individuals, and cluster 2 1704.
- Result
There was a significant main effect of time, indicating that individuals with low performance improved from pre- to post-training in both treatments (β = 0.14, 95% CI [0.10, 0.17], p < 0.001).
- Limitation
First, group-level aggregation restricted inferential power and precluded modeling of individual trajectories.
- Cognitive function
The time × training group interaction was significant (β = 0.15, 95% CI [0.10, 0.20], p < 0.001), indicating that Lumosity users demonstrated larger gains in GI performance over time than crossword users.
Quoted word for word from the full text on link.springer.com.
The paper
University of Pavia; University of Trento
GeroScience, Jul 10, 2026, Open access
