Longitudinal white matter changes associated with cognitive training

Abstract

Improvements in behavior are known to be accompanied by both structural and functional changes in the brain. However, whether those changes lead to more general improvements, beyond the behavior being trained, remains a contentious issue. We investigated whether training on one of two cognitive tasks would lead to either near transfer (that is, improvements on a quantifiably similar task) or far transfer (that is, improvements on a quantifiably different task), and furthermore, if such changes did occur, what the underlying neural mechanisms might be. Healthy adults (n = 16, 15 females) trained on either a verbal inhibitory control task or a visuospatial working memory task for 4 weeks, over the course of which they received five diffusion tensor imaging scans. Two additional tasks served as measures of near and far transfer. Behaviorally, participants improved on the task that they trained on, but did not improve on cognitively similar tests (near transfer), nor cognitively dissimilar tests (far transfer). Extensive changes to white matter microstructure were observed, with verbal inhibitory control training leading to changes in a left-lateralized network of frontotemporal and occipitofrontal tracts, and visuospatial working memory training leading to changes in right-lateralized frontoparietal tracts. Very little overlap was observed in changes between the two training groups. On the basis of these results, we suggest that near and far transfer were not observed because the changes in white matter tracts associated with training on each task are almost entirely nonoverlapping with, and therefore afford no advantages for, the untrained tasks.

Keywords: cognitive training; diffusion tensor imaging; far transfer; longitudinal imaging; near transfer.

FIGURE 1

Study Procedure. Five scanning sessions were completed over the course of 4 weeks, with 1 week in between each session. Participants trained on the task at home between scans

FIGURE 2

Amount of improvement across the at‐home training sessions for (a) Self‐Ordered Search and (b) Double Trouble

FIGURE 3

Task improvement from beginning of cognitive training to the end, for the Self‐Ordered Search (SOS) and Double Trouble groups (DT). The first and third quartiles are marked by the lower and upper edges of the boxes, respectively. Lower and upper whiskers extend to the smallest and largest value, respectively, within 1.5 times the interquartile range. Outlying values beyond these ranges are plotted individually

FIGURE 4

Changes in FA across all scanning sessions relative to baseline. (a) In red, we show the areas in which the changes from baseline are significantly larger in the Double Trouble training group than in the Self‐Ordered Search training group. (b) In blue, we show the areas in which the changes from baseline are significantly larger in the Self‐Ordered Search training group than in the Double Trouble training group. Clusters have been thickened for visualization using tbss_fill, and results are overlaid on the FMRIB58_FA template and the mean skeletonized FA data of the current sample

FIGURE 5

Areas showing group differences in the change in FA from Baseline to Scan 5. In red, we show the areas in which the changes from Scan 1 to Scan 5 are significantly larger in the Double Trouble training group than in the Self‐Ordered Search training group. In blue, we show the areas in which the changes from Baseline to Scan 5 are significantly larger in the self‐ordered search training group than in the double trouble training group. Changes uniquely associated with Double Trouble were largely within the left inferior occipitofrontal and longitudinal fasciculi, while changes associated with Self‐Ordered Search were largely within the right superior longitudinal fasciculus. Clusters have been thickened for visualization using tbss_fill, and results are overlaid on the FMRIB58_FA template and the mean skeletonized FA data of the current sample

FIGURE 6

Areas showing group differences in the change in FA from Baseline to Scan 2. In red, we show the areas in which the changes from Baseline to Scan 2 are significantly larger in the Double Trouble training group than in the Self‐Ordered Search training group. In blue, we show the areas in which the changes from Baseline to Scan 2 are significantly larger in the Self‐Ordered Search training group than in the Double Trouble training group. Clusters have been thickened for visualization using tbss_fill, and results are overlaid on the FMRIB58_FA template and the mean skeletonized FA data of the current sample

FIGURE 7

Brain regions that showed significant changes from Baseline to Scan 5 for both Double Trouble and Self‐Ordered Search. Significant regions included the primary auditory area, the anterior thalamic radiation, the corticospinal tract, and the forceps major, as well as one region each within the inferior occipitofrontal fasciculus and the superior parietal lobe. Clusters have been thickened for visualization using tbss_fill, and results are overlaid on the FMRIB58_FA template and the mean skeletonized FA data of the current sample

FIGURE 8

FA values from Scan 5–Baseline difference maps, correlated with Scan 5–Baseline difference scores. Within the brain maps, red depicts areas in which Double Trouble scores correlated with FA changes within that training group; blue depicts areas in which the Self‐Ordered Search scores correlated with FA changes within that training group. Scatterplots show the relationship between difference score on the task and change in FA from Baseline to Scan 5. Clusters have been thickened for visualization using tbss_fill, and results are overlaid on the FMRIB58_FA template and the mean skeletonized FA data of the current sample. White circles indicate the region of interest displayed in the corresponding scatterplot