Working memory plasticity: Brain imaging studies

To the best of our knowledge, the first functional brain imaging study of the neural correlates of cognitive training in adulthood and aging targeted the method of

LOCI mental imagery mnemonic (Nyberg et al., 2003). In keeping with previous behavioral studies (Bakes & Kliegl, 1992), age differences in memory performance were found to be magnified after as compared to before the intervention. Analyses of functional brain activity patterns, based on positron emission tomography (PET) recordings, revealed that one basis for the age-related difference implicated putative task-specific processes such as creating visual images and binding of cue-target information.These processes were related to occipito-temporal cortex and medial- temporal lobe regions, and significant age differences were observed in these regions (Nyberg et al., 2003; see also Jones et al.,2006).A second basis for the observed age differences and their magnification following training was found to be age-related processing deficits (e.g. executive), which affected the older adults’ ability to effectively make use of the LOCI mnemonic (Nyberg et al., 2003; see also Jones et al., 2006).The apparent age-related processing deficit was linked to diminished frontal brain activity.

Thus, age-related differences in fronto-parietal brain activity, and associated executive processes, seem to underlie constraints in benefitting from mnemonic support. In a subsequent project, we therefore targeted training of executive processes (cf. section above, Dahlin et al., 2008a). The basic idea was that, if we could strengthen executive processes and related fronto-parietal circuits, then older adults might show similar training-related gains to those of younger adults. A second motivation behind targeting general executive processes was to test the hypothesis that training of executive functions might lead to stronger and potentially broader transfer effects than more domain-specific interventions such as the method of LOCI. Numerous brain-imaging studies converge to show that a common characteristic of many cognitive tasks is that they engage fronto-parietal networks (for reviews, see Cabeza & Nyberg, 2000; Naghavi & Nyberg, 2005). At least in part, this commonality may reflect shared executive processes (Collette et al, 2006; Marklund et al., 2007). Hence, by strengthening executive processes and associated fronto-parietal networks by a certain type of cognitive training program, transfer to non-trained tasks that also rely on fronto-parietal regions might be expected (cf. Klingberg et al., 2005). However, as was reviewed above, cognitive transfer effects are typically weak or non-existent (e.g. Owen et al., 2010; Lee et al., 2012), which apparently conflicts with the notion that training a specific executive task and associated fronto-parietal recruitment will benefit other executive (fronto-parietal) tasks. An alternative possibility, then, is that training affects much more specific processes and brain systems, which accounts for the general finding of limited transfer.

Figure 15.1 A schematically outlines our experimental protocol for testing whether executive functions training might broadly enhance the performance of older adults and lead to transfer via fronto-parietal networks. We targeted updating of information in working memory, a process related to fronto-parietal cortical regions as well as the striatum (Collette et al., 2006; Marklund et al., 2007; O’Reilly, 2006). Before and after five weeks of updating training (Dahlin et al., 2008b), functional magnetic resonance imaging (fMRI) was used to assess training-related changes in functional brain activity. Three different tasks were scanned: a letter- memory updating task which served as the criterion task and two transfer tasks: an n-back working memory task and a Stroop inhibition task. All three tasks were expected to engage executive control processes and fronto-parietal circuits (Collette et al., 2006; Miyake et al., 2000), and our fMRI findings supported this prediction (Figure 15.IB). The letter-memory task and the n-back task were expected to involve updating and engage the striatum (Miyake et al., 2000; O’Reilly, 2006), whereas the Stroop inhibition task was not, and the fMRI results confirmed this prediction (Figure 15.IB). Thus, to the degree that transfer was based on a shared fronto-parietal network, we would expect to find transfer from letter memory to both n-back and Stroop, whereas if transfer instead was based on the striatal updating network we would expect transfer to n-back only. The findings showed a highly selective behavioral transfer effect to n-back along with a training-related modulation of the fMRI signal in the striatum.These results confirm the hypothesis that a prerequisite for transfer is that the transfer task taxes the same basic process as supported by the intervention and related brain areas ()onides, 2004; Thorndike & Woodworth, 1901)—in the present case updating and basal ganglia circuits.

An additional goal of the experiment was to examine whether executive processing training would lead to similar training-related gains for older as for younger adults. The results were positive by showing parallel gains for both age groups, i.e. there was no magnification of age differences. However, the older adults’ level of performance on the letter-memory criterion task after five weeks of training was similar to the level reached by younger adults after two weeks. Moreover, the older adults did not show a significant transfer effect to the n-back task. The age-related constraints on learning and transfer after updating training were related to age- related changes in the striatum, thereby providing further support for a critical role of this brain region for updating.

Theoretical and computational models of the role of striatum in updating of information in working memory indicate that dopaminergic neurotransmission is a key factor (Drustewitz, 2006; O’Reilly, 2006). A recent PET study provided empirical support for this prediction (Backman et al., 2011). PET was used to assess dopamine D2 binding potential before and after five weeks of updating training, and an extensive neuropsychological batter)' was administered before and after the training period (cf. the fMRI protocol above).The behavioral results replicated the findings from the fMRI study by showing highly selective transfer to an n-back task only. The PET recordings were done during a baseline task as well as during a period when information was updated in working memory. Updating processing was found to affect bilateral striatal dopamine binding. Critically, a training-related influence on dopaminergic activity was observed in the left striatum, closely overlapping the region where a training-related effect was seen in the fMRI study. These findings extend related observations for the dopamine D1 system (McNab et al., 2009), and show that the dopamine system is modifiable by directed training. Physical activity (exercise) has been demonstrated to have beneficial effects on learning and cognition (Hillman et al., 2009), and the exercise-induced cognitive

(A) An outline of the experimental protocol

FIGURE 15.1 (A) An outline of the experimental protocol. (B) To the left, brain maps showing activation of common fronto-parietal circuits for all three tasks. The bar graph shows overlapping activation at pretest for both updating tasks (letter memory and 3-back) but not for the Stroop task

Source: Reproduced with permission from Dahlin et al. (2008b)

facilitation may in part be mediated by stimulation of dopaminergic neurotransmission by physical activity (Winter et al., 2007).

Taken together, the brain-imaging findings converge with behavioral findings of weak transfer effects, and indicate that a critical underlying factor is that the criterion and transfer tasks must engage overlapping specific processes and brain regions (cf. Boot et al., 2010), such as updating and the striatum as focused on here. Quite likely, other forms of training will critically depend on other processes and related brain circuits, but together with the rich set of behavioral studies showing limited transfer effects the imaging data reviewed here predict that attempts at developing training programs that lead to broad transfer effects are likely to fail.

< Prev   CONTENTS   Source   Next >