The pattern and loci of training-induced brain changes in healthy older adults are predicted by the nature of the intervention

Abstract

There is enormous interest in designing training methods for reducing cognitive decline in healthy older adults. Because it is impaired with aging, multitasking has often been targeted and has been shown to be malleable with appropriate training. Investigating the effects of cognitive training on functional brain activation might provide critical indication regarding the mechanisms that underlie those positive effects, as well as provide models for selecting appropriate training methods. The few studies that have looked at brain correlates of cognitive training indicate a variable pattern and location of brain changes--a result that might relate to differences in training formats. The goal of this study was to measure the neural substrates as a function of whether divided attentional training programs induced the use of alternative processes or whether it relied on repeated practice. Forty-eight older adults were randomly allocated to one of three training programs. In the single repeated training, participants practiced an alphanumeric equation and a visual detection task, each under focused attention. In the divided fixed training, participants practiced combining verification and detection by divided attention, with equal attention allocated to both tasks. In the divided variable training, participants completed the task by divided attention, but were taught to vary the attentional priority allocated to each task. Brain activation was measured with fMRI pre- and post-training while completing each task individually and the two tasks combined. The three training programs resulted in markedly different brain changes. Practice on individual tasks in the single repeated training resulted in reduced brain activation whereas divided variable training resulted in a larger recruitment of the right superior and middle frontal gyrus, a region that has been involved in multitasking. The type of training is a critical factor in determining the pattern of brain activation.

Figure 1. Flow chart according to the Consort reporting instructions.

Figure 2. Experimental paradigm.

Schematic representation of the task conditions order in fMRI (A), as well as the alphanumeric equation and visual detection tasks in both single-task (B) and dual-task (C).

Figure 3. Feedback during variable training.

Four examples of the histograms that provided visual feedback to the participants in the DIVIDED VARIABLE training group. The dark column represents the performance of alphabetical equation that was obtained under single-task baseline. The light column shows performance that was reached in the dual-task condition. The line represents the level of performance that was expected. (a; b) Examples of an 80% Equation trial where participants were asked to allocate 80% of their attention to the alphanumeric equation task: a) shows a trial where performance was below the expected threshold; (b) shows a trial where participants succeeded to obtain the expected level of performance. (c; d) Example of a 50% Equation trial where participants were asked to allocate 50% of their attention to the alphanumeric equation task: c) shows a trial where performance was below the expected threshold; (d) shows a trial where participants succeeded to obtain the expected level of performance.

Figure 4. Activations related to dual-tasking prior to training (pre-training session).

Network of prefrontal activation in dual-task, with more emphasis on equation (80% Equation (80/20)) in A and B; equal division of attention (50% Equation (50/50)) in C and D; and more emphasis on detection (20% Equation (20/80)) in E and F. The threshold for display is P<0.001, uncorrected, 10 voxels. Coloured bar is representative of t scores mentioned in Table 5. “L” denotes the left side of the brain, while “R” denotes the right side.

Figure 5. Activation-related to modulation of attention prior to training (pre-training session).

Subtracting dual-task 80% Equation (80/20) from dual-task 20% Equation (20/80) involves activation in the left superior and medial frontal gyrus (A and B), and left superior temporal and left cingulate gyrus (B). The threshold for display is P<0.001, uncorrected, 10 voxels. Coloured bar is representative of t scores mentioned in table 5. “L” denotes the left side of the brain, while “R” denotes the right side.

Figure 6. SINGLE REPEATED training effect.

Decreased (Pre>Post) activation in single-task with alphanumeric equation is found in the right inferior and middle frontal gyrus (A and B) and left middle frontal (B). Histogram in (C) indicates the Beta value (activity estimates ± SE) in right inferior and middle frontal gyrus. The threshold for display is P<0.001, uncorrected, 10 voxels. Coloured bar is representative of t scores mentioned in Table 5. “L” denotes the left side of the brain, while “R” denotes the right side.

Figure 7. DIVIDED VARIABLE training effect.

Increased (Post>Pre) activation in dual-task with more emphasis on detection (20% Equation (20/80)) is found in the right superior and middle frontal gyrus (10). Histogram in (C) indicates the Beta value (activity estimates ± SE) in the region showing increase activity in right superior and middle frontal gyrus during pre and post training session. The threshold for display is P<0.001, uncorrected, 10 voxels. Coloured bar is representative of t scores mentioned in Table 5. “L” denotes the left side of the brain, while “R” denotes the right side.

Figure 8. BOLD signal pre and post-intervention for the DIVIDED VARIABLE training.

Beta value (activity estimates ± SE) in pre- and post-training sessions for single-task (visual detection > rest; alphanumeric equation > rest) and for each dual-task condition (80% Equation (80/20) > rest, 50% Equation (50/50) > rest and 20% Equation (20/80) > rest) in Right area 10 - the region showing post-training effect.