Lily Fu

Summary of the findings

The main aim of the present thesis was to examine the influence of individual differences in executive function (EF) on approaches that could alleviate episodic memory ageing. To achieve this goal, three empirical studies were conducted to validate the effectiveness of the approach of embedding aids within ongoing encoding tasks (study 1), to evaluate the role of individual differences in EF on the effectiveness of this approach (study 2), and to identify the neural activities associated with the effect of individual differences in this effectiveness (study 3). Lastly, to confirm and test a more applied implication of the effect of individual differences, an empirical study on cognitive training was conducted (study 4). In this concluding chapter, the main findings from these four studies will be first summarized and then discussed.

Chapter 2 validated the effectiveness of embedding factors serving as ‘memory aids’ within the ongoing task to improve episodic memory performance of older adults. More specifically, we implemented suggestions provided by the level-of-processing theory that deep/semantic processing may lead to more durable memory traces. Moreover, another important factor was added to this approach, that is, cognitive effort. Results confirmed the beneficial effect of higher levels of processing and cognitive effort on episodic memory performance in both older and younger participants. Moreover, older adults took additional advantage of involving more cognitive effort if the encoding was deep/semantic. Interestingly, following the environmental-compensation account, older adults’ memory performance reached the level of younger adults’ when environmental support was provided by effortful semantic encoding as embedded aids.

Chapter 3 extended the study of chapter 2 and introduced an important element of my thesis – the importance of individual differences. To be more specific, this study addressed the question whose memory performance benefits most from effortful encoding, focusing on elderly with either low or high EF. Results demonstrated that high EF older adults, putatively possessing optimal cognitive reserve, are able to make use of embedded aids in the form of promoting cognitive effort. In contrast, with the appearance of a CRUNCH-like pattern. The CRUNCH model states that the extent to which the brain is able to compensate is restricted by its own cognitive capacity. Here, we observed that the benefit of promoting cognitive effort is limited for older individuals with low EF. These findings provided empirical evidence for the CRUNCH model and underline the impact of individual differences on ageing research.

Chapter 4 further extended the first two studies in chapters 2 and 3. Equipped with fMRI, this study examined the neural basis of the association between individual differences in EF and episodic memory performance. Results from multi-regression analyses confirmed that stronger EF abilities in late adulthood, putatively indicating a higher efficiency in using limited cognitive resources, were significantly and positively associated with E/R flip-associated neural activity. The E/R flip-associated activity is regarded as a potential neural mechanism for successful memory performance.

Chapter 5 took the results given by chapter 3 as starting point and tested predictions derived from the CRUNCH model in the framework of cognitive training. Specifically, chapter 5 addressed the question which older adults would benefit most from a cognitive training program: those with relatively strong or those with relatively low executive functioning. As hypothesized, low EF older adults showed a CRUNCH pattern during training of the alpha-span task, which might explain that no beneficial effects of training on performing a transfer task was found in this group. Similar to what was found in Chapter 3, results from this study were consistent with the magnification account, which states that high EF older adults should show the strongest training and transfer benefits. Again, this study underscores the importance of taking individual differences into account in ageing studies. Especially in intervention studies, it is crucial to match the difficulty of the program to the cognitive ability of the trained individual.

Discussion

Due to the vast improvements in health care over the last decades, normal adults can now live up to 100 years. At the same time, it has become clear that it is not only the physical body that declines with ageing, but that our brains are also susceptible to the effects of ageing, resulting in cognitive decline. Although in general, it is not challenging for older adults to have a physical hip or shoulder replacement, it is not possible at this moment to have their memory function replaced. For the requirement of improving the quality of life at old ages and to reduce the social burden to our society, there is an urge to understand cognitive ageing and to develop interventions for it.

Over the past several decades, psychologists, neurologists and other neuroscientists aim to understand cognitive ageing by comparing older adults with younger adults. However, when it comes to developing possible interventions, this framework can be insufficient. One simple analogy is that when people are sick, their states are compared with normal healthy people (e.g., elevated body temperature) and those people are tagged with “being sick”. However, it is not enough for a doctor to give prescriptions only by seeing the label “sick” because, obviously, different diseases have different causes and treatment has to be prescript with caution depending on each patient’s medical history. Regarding the development of an effective “treatment” for cognitive ageing, the same steps have to be taken, because there can be a variety of causes for the same type of cognitive decline. For example, ‘normal’ ageing-related episodic memory decline – that is, in the absence of neurodegenerative disease – is associated with decline in resources (i.e., reduced processing capacity and efficiency) and memory binding (i.e., difficulty in making associations spontaneously) (Daselaar & Cabeza, 2013; Old & Naveh-Benjamin, 2008; Tromp, Dufour, Lithfous, Pebayle, & Després, 2015; van Geldorp, Parra, & Kessels, 2015). More importantly, the time course of cognitive deterioration resulting from ageing is heterogeneous across individuals: some older adults are able to maintain a high level of functioning while others exhibit severe cognitive impairments, for example, memory loss (Christensen et al., 1999; Wilson et al., 2002).

As illustrated in Figure 6.1, this thesis focused on the resource-deficit origin of ageing-related episodic memory decline which causes the efficiency and capacity issues. Targeting the efficiency issue, we provided embedded aids (environmental support) in the memory encoding phase. By applying a cognitive training approach, the aim was to solve both of the issues. Most importantly, the impact of individual differences in EF on these approaches was studied. As indicated by the CRUNCH model, it is essential to keep individual’s cognitive capacity in mind.

Figure 1: The framework of effective interventions for age-related episodic memory decline.

Methodological and Theoretical Considerations and Future Directions

The plasticity of brain capacity and efficiency
Cognitive ageing is more often than not in companion with negative forms of plasticity, such as the adverse effects of neural and functional decline (Cramer et al., 2011). On the other hand, an increasing number of studies have shown a form of “positive” plasticity. For example, compared to young adults, it has been well documented that older adults display greater activation or additional recruitment of prefrontal and/ or parietal brain regions (Angel et al., 2011; Gutchess et al., 2005). Moreover, overactivation has been reported in the form of bilateral rather than unilateral recruitment of these brain regions in older adults (Cabeza, 2002; Cabeza & Dennis, 2012). These different brain activation patterns are usually linked to better or ‘youth-like’ behavioural performance and are summarized as compensatory processes (Reuter-Lorenz & Park, 2014).

As shown in Figure 6.1, cognitive ageing is suggested to be associated with functional decline in both cognitive capacity and efficiency. However, the question how exactly the different types of compensatory process (activation, recruitment) are related to these aspects of functional decline remains to be answered. One possibility is that the amplitude of activation simply represents efficiency and additional recruitment of brain regions reflects a compensation for decreased capacity. In Chapter 5, we found that older adults with higher EF have greater neural activities associated with successful memory events. Linking to the efficiency hypothesis in Chapter 3 that older adults with relatively high EF could target their attentional resources towards more useful information processing, we suggest that greater activation might indeed represent higher efficiency. Although sometimes it is claimed that decreased neural activity (at comparable performance levels) means greater efficiency (e.g. Heinzel et al., 2014), we argue that in our design, the contrast between remembered and forgotten items revealed neural processes that are only associated with successful memory performance, which means these are all “effective” neural activities. Therefore, greater “effective” neural activity represents higher efficiency. However, the present thesis does not directly speak to the issue of the relation between overactivation and capacity, which requires future studies.

Another form of “positive” plasticity might originate from cognitive training. Many studies have examined structural and functional neural changes in older adults before and after participating in a cognitive training program (Bäckman & Nyberg, 2013; Heinzel et al., 2014; Lustig et al., 2009;). Though it has been debated whether cognitive training has an effect at all, it is still questionable in those studies which reported beneficial effects of training (Anguera et al, 2013; Dahlin et al, 2008; Shah et al., 2017) and that through which aspect of cognitive ageing does training help. Is it by adding more cognitive capacity? Or by increasing efficiency of the brain? The answer to this question cannot be given within the scope of this thesis, but many criticisms towards cognitive training regard the effectiveness of training programs as the increase of subject’s strategy use or the overall speed, and put the equivalent of cognitive capacity and cognitive functions to a ‘muscle’ that can be enhanced by training into question. Also the transfer of improved performance on tasks to domains that are not trained is often limited. In the future, more empirical studies are needed to verify the effectiveness of the cognitive training approach and the mechanism underlining the effectiveness. These studies should also include a control condition, preferably consisting of participants performing some non-adaptive version of the task used in the target, training group.

Beyond ‘CRUNCH’
The Compensation-Related Utilization of Neural Circuits Hypothesis (CRUNCH) frames the different brain activation patterns of older adults in the context of cross-sectional studies as compensatory processes. A similar perspective is taken in the ‘Scaffolding Theory of Ageing and Cognition’ model (STAC, Park & Reuter-Lorenz, 2009) and later the revised STAC model (STAC-r, Reuter-Lorenz & Park, 2014). CRUNCH has advantages in explaining both under- and over-recruitment of brain activations during task performance, since it proposes that the extent to which an ageing brain can compensate by involving more neural activation is dependent on the cognitive capacity of the individual involved. It is especially beneficial in explaining results in the current thesis. However, in terms of number of factors assumed to affect compensation mechanisms, it is rather limited in comparison with the STAC or STAC-r model. The STAC model makes distinct and confirmable predictions regarding the influence of different structural and functional brain variables on the compensatory processes, and the STAC-r model incorporates other life course factors that can have direct effects on the development of compensatory scaffolding, such as intellectual engagement, fitness, multilingualism, stress, toxin exposure, etc., whereas the CRUNCH model only summarizes the general influence of the individual’s cognitive capacity and cognitive load of the ongoing task. One limitation of the current thesis in this framework is that we only took EF as a general indicator of the individual’s cognitive capacity, and did not investigate other attributes of life course factors. Moreover, even within the scope of EF, one could argue that the neuropsychological tests applied in our studies do not cover all aspects of executive functioning (e.g. task switching, strategy use). Future studies should incorporate a larger dimension of individual differences.

Besides the theoretical limitation of the CRUNCH model, I would like to note the somewhat limited neural evidence of this model that could be provided by this thesis. One of the temptations of the fMRI study described in Chapter 4 was to localize (in the brain) the onset of the CRUNCH-predicted pattern when cognitive effort increases. Hence, a similar experimental design as in Chapter 2 and 3 was used in this study. However, we failed to investigate this issue because the CRUNCH-like pattern was not replicated in the behavioural results. The reason is that we changed the trial duration of the encoding paradigm from self-paced to a fixed duration of 3500 ms. This adaptation was implemented to prevent the confounding of trial duration on the BOLD signal. However, due to this alteration, we could not capture the cognitive effort devoted to each trial adequately since previously we used the decision-making time as an index of effort. With a fixed duration, it is hard to say what participants actually did after the semantic decision was made. The remaining time gave them chance to reflect their decisions or simply to mind wandering.

Individualized ageing model and personalized intervention
A primary aim of this thesis is to draw attention to the necessity of considering individual differences in ageing studies. Although for reasons of limited resources the current studies adopted a cross-sectional approach, the most powerful tool in studying the cognitive ageing path is through longitudinal studies. Luckily, the blossoming of digitalized methods has provided us with more feasibility in tracking longitudinal data (Mahrt & Scharkow, 2013), which is especially beneficial considering the large amount of life course factors that could affect ageing progress and the existence of a large heterogeneity in the ageing population. Equipped with data-mining techniques and big data analyses, more aspects of individual factors can be analyzed and clustered, and ample opportunities are provided for understanding causes of cognitive ageing.

As mentioned earlier, different types of cognitive decline are presented differently within the ageing population, and even the same type of cognitive deficit can have different origins. A limitation of this thesis is that most of the elderly participants that we examined were well educated and had relatively intact cognitive functions. To study the diversity of cognitive ageing, I suggest future studies should take a broader range of the ageing population, tracking longitudinal data of many life course factors at the same time, and forming an individualized ageing model of each individual. For example, recording information from multiple dimensions, such as cortical thickness, white matter integrity, physical exercise, social interactions, intellectual engagement, and stress levels, of the individual subject, and at the same time, monitoring changes in different cognitive functions over time. After identifying a significant influence of certain factors, personalized interventions can be developed, aimed at counteracting these influences.

Conclusion

The current thesis shows that episodic memory performance of older adults can be improved by using embedded aids in the ongoing task. Older adults with relatively high executive functioning take the most advantage of these aids, suggesting that they are able to direct limited resources to a more efficient information processing. Evidence for this suggestion was found by the predictive effect of EF on neural activities associated with successful memory events (the E/R flip), and by the fact that older adults with relatively low EF displayed a CRUNCH-like pattern when task demands were beyond their cognitive capacity. This thesis underlines the significant influence of individual differences in ageing studies and provides more insights for interventions that aim to eliminate ageing-related episodic memory declines. More specifically, caution needs to be taken when providing older adults interventions. An essential step is to first evaluate the individual’s cognitive capacity and provide matched intervention methods. In addition, close performance monitoring is essential to assess the point at which the individual reaches the CRUNCH threshold. Future studies should include a broader range of the elderly population and simultaneously investigate more life course factors.