Touchscreen-Based Cognitive Training Alters Functional Connectivity Patterns in Aged But Not Young Male Rats

Effects of cognitive training or walking on global functional connectivity in young and aged rats. A, B, Whole-brain connectivity maps by scan session and age group for rats that underwent cognitive training versus walking around a track for reward. The brain network maps represent connected nodes, the spheres indicating the strength of node connectivity to the whole network. Edges represent significant correlations in the BOLD fluctuations between two nodes and are depicted by lines when the correlation is z > 0.3. C, D, Whole-brain correlation matrices for the 150 segmented brain regions by scan session and age group for the Cognitive Training or Walking conditions.

Change in global functional connectivity across scan sessions following Cognitive Training or Walking. A, The difference between scan 1 and scan 2 whole-brain correlation matrices from young (top) and aged (bottom) rats for the Cognitive Training (Young: N = 7; Aged: N = 11) versus Walking (Young: N = 5; Aged: N = 8) conditions. B, Line graphs representing the number of significant edges (z ≥ 0.3) across scan sessions by age for the Cognitive Training (left) or Walking (right) conditions. There was a significant effect of age (F_(1,16) = 5.23, p = 0.04, η² = 0.25) and a significant interaction effect of scan session and age (F_(1,16) = 7.25, p = 0.02, η² = 0.31), indicating that aged rats had fewer edges compared with young rats at scan 1, and that while the number of these edges increased at scan session 2 for aged rats, the number of edges decreased for young rats. In the Walking condition, there were no significant effects of age or scan session on edge number. Error bars are ± 1 SEM.

Node degree and modularity of high degree nodes. A, Scatter histograms representing node degree in young and aged rats across scan sessions relative to Cognitive Training (Young: N = 7; Aged: N = 11) or (B) Walking (Young: N = 5; Aged: N = 8). There was not a significant main effect of age (Cognitive Training: F_(1,16) = 0.70, p = 0.42; Walking: F_(1,11) = 0.03, p = 0.86) or scan session (Cognitive Training: F_(1,16) = 0.40, p = 0.54; Walking: F_(1,11) = 0.46, p = 0.51), nor a significant interaction of age and scan session (Cognitive Training: F_(1,16) = 0.72, p = 0.41; Walking: F_(1,11) = 0.32, p = 0.59) on the number of nodes with a degree >10. C, Line graphs depicting the modularity of the 20 highest degree nodes for the Cognitive Training and (D) Walking conditions. In both conditions, there was not a main effect of age (Cognitive Training: F_(1,16) = 1.40, p = 0.25; Walking: F_(1,11) = 0.003, p = 0.96) or scan session (Cognitive Training: F_(1,16) = 0.58, p = 0.50; Walking: F_(1,11) < 0.001, p = 0.99). There was a significant interaction with age and scan session in the Cognitive Training condition (F_(1,16) = 6.21, p = 0.02, η² = 0.28), which was not evident in the Walking condition (F_(1,11) = 0.15, p = 0.71). Error bars are ± 1 SEM.

Seed-based connectivity for the perirhinal cortex. A, Voxel analyses in young and aged rats across scan sessions in response to Cognitive Training (Young: N = 7; Aged: N = 11), or (B) Walking (Young: N = 5; Aged: N = 8). Panels C–F show the correlation coefficients of the seed-based analysis (y-axis) by scan session (x-axis). C, In the Cognitive Training condition, the correlation coefficient of the perirhinal cortex and infralimbic cortex did not significantly differ by age (F_(1,16) = 1.71, p = 0.21) or scan session (F_(1,16) = 0.03, p = 0.88). The interaction between age and scan session, however, trended toward significance (F_(1,16) = 3.22 p = 0.09). D, In the Walking condition, the correlation coefficient of the perirhinal cortex and infralimbic cortex did not significantly differ between age groups (F_(1,11) = 0.87, p = 0.37), scan session (F_(1,11) = 0.01, p = 0.91), nor was the interaction significant (F_(1,11) = 0.81, p = 0.39). E, In the Cognitive Training condition, the correlation coefficient of the perirhinal cortex and dorsal hippocampus did not differ by age group (F_(1,16) = 0.07, p = 0.79) or scan session (F_(1,16) = 0.75, p = 0.40). The interaction between age and scan session, however, reached significance (F_(1,16) = 7.33, p = 0.02, η² = 0.31). F, In the Walking condition, the correlation coefficient of the perirhinal cortex and dorsal hippocampus was not significantly different between age groups (F_(1,11) = 1.34, p = 0.27), scan sessions (F_(1,11) = 1.27, p = 0.28), nor did the interaction effect reach significance (F_(1,11) = 0.06, p = 0.81). Error bars are ± 1 SEM.

Seed-based connectivity for the prelimbic cortex. A, Whole-brain voxel analyses in young and aged rats across scan sessions for the Cognitive Training (Young: N = 7; Aged: N = 11), and (B) Walking conditions (Young: N = 5; Aged: N = 8). Panels C–F show the correlation coefficients (y-axis) by scan session (x-axis). C, In the Cognitive Training condition, the correlation coefficient of the prelimbic cortex and dorsal hippocampus did not significantly differ by age (F_(1,16) = 0.15, p = 0.70), scan session (F_(1,16) = 0.01, p = 0.93), nor was the interaction significant (F_(1,16) = 0.51, p = 0.49). D, In the Walking condition, the correlation coefficient of the prelimbic cortex and dorsal hippocampus also did not differ between age groups (F_(1,11) = 0.10, p = 0.76), scan sessions (F_(1,11) = 2.79, p = 0.12), nor was there and a significant interaction (F_(1,11) = 3.32, p = 0.10). E, The correlation coefficient of the prelimbic cortex and perirhinal cortex in young and aged rats in the Cognitive Training condition did not significantly differ between age groups (F_(1,16) = 0.54, p = 0.47), scan sessions (F_(1,16) = 1.13, p = 0.30), nor was there a significant interaction (F_(1,16) = 0.27, p = 0.61). F, In the Walking condition, the correlation coefficient of the prelimbic cortex and perirhinal cortex in young and aged rats did not differ between age groups (F_(1,11) = 4.86, p = 0.05, η² = 0.31), scan sessions (F_(1,11) = 0.14, p = 0.71), nor was the interaction significant (F_(1,11) = 0.27, p = 0.61). Error bars are ±1 SEM.

Age-related differences in seed-based connectivity for the rostral retrosplenial cortex. A, Rostral retrosplenial cortex to whole-brain voxel analyses in young and aged rats across scan sessions in response to Cognitive Training (Young: N = 7; Aged: N = 11), or (B) Walking (Young: N = 5; Aged: N = 8). Panels C, D show the correlation coefficients (y-axis) by scan session (x-axis). C, In the Cognitive Training condition, the correlation coefficient of the rostral and caudal retrosplenial cortex did not significantly differ between age groups (F_(1,16) = 0.59, p = 0.46), or scan sessions (F_(1,16) = 0.01, p = 0.93). There was, however, a significant interaction of age and scan session (F_(1,16) = 5.43, p = 0.03, η² = 0.25). D, In the Walking condition, the correlation coefficient of the rostral and caudal retrosplenial cortex was significantly different between scan sessions (F_(1,11) = 9.52, p = 0.01, η² = 0.46), but there was not a significant effect of age group (F_(1,11) = 2.40, p = 0.15). Unlike the rats that underwent Cognitive Training, there was not a significant interaction between age and scan session (F_(1,11) = 0.07, p = 0.80). Error bars are ±1 SEM.

Clustering coefficients. Panels A–E show the clustering coefficient (y-axis) by scan session (x-axis) for the Cognitive Training (left panels; Young: N = 7; Aged: N = 11) and Walking (right panels; Young: N = 5; Aged: N = 8) conditions. A, In the perirhinal cortex, there was not a significant effect of age (Cognitive Training: F_(1,16) = 0.02, p = 0.90; Walking: F_(1,11) = 0.40, p = 0.54), scan session (Cognitive Training: F_(1,16) = 0.72, p = 0.41; Walking: F_(1,11) = 2.06, p = 0.18), nor a significant interaction (Cognitive Training: F_(1,16) = 1.24, p = 0.28; Walking: F_(1,11) = 0.15, p = 0.71). B, In the dorsal hippocampus, there was not a significant main effect of age (Cognitive Training: F_(1,16) = 2.43, p = 0.14; Walking: F_(1,11) = 0.03, p = 0.86) nor a significant interaction (Cognitive Training: F_(1,16) = 0.68, p = 0.42; Walking: F_(1,11) = 2.24, p = 0.16). There was not a significant effect of scan session for Cognitive Training (F_(1,16) = 0.47, p = 0.50), but there was a significant effect of scan session in the Walking condition (F_(1,11) = 14.94, p = 0.003, η² = 0.58). C, In the infralimbic cortex, for the Cognitive Training condition, there was a significant effect of age (F_(1,16) = 4.47, p = 0.05, η² = 0.22) and an interaction of scan session and age (F_(1,16) = 4.43, p = 0.05, η² = 0.22). There was not a significant effect of scan session (F_(1,16) = 1.84, p = 0.19). In the Walking condition, there was a significant effect of age (F_(1,11) = 5.31, p = 0.04, η² = 0.33) and an effect of scan session (F_(1,11) = 14.22, p = 0.003, η² = 0.56). Unlike the Cognitive Training rats, there was not a significant interaction of age and scan session (F_(1,11) = 0.26, p = 0.62). D, In the prelimbic cortex, there was not a significant main effect of age (Cognitive Training: F_(1,16) = 1.73, p = 0.21; Walking: F_(1,11) = 3.54, p = 0.09) or an interaction of scan session and age (Cognitive Training: F_(1,16) = 0.01, p = 0.92; Walking: F_(1,11) = 0.02, p = 0.89). While there was not an effect of scan session for the Cognitive Training condition (F_(1,16) = 0.07, p = 0.80), there was a significant decrease in the Walking condition (F_(1,11) = 11.66, p = 0.01, η² = 0.51). E, In the rostral retrosplenial cortex, there was not a significant main effect of age (Cognitive Training: F_(1,16) = 0.50, p = 0.49; Walking: F_(1,11) = 2.39, p = 0.15), scan session (Cognitive Training: F_(1,16) = 1.91, p = 0.19; Walking: F_(1,11) = 2.75, p = 0.13), nor an interaction of scan session and age (Cognitive Training: F_(1,16) = 0.18, p = 0.67; Walking: F_(1,11) = 0.03, p = 0.87). Error bars are ±1 SEM.