Strategies to Promote Cognitive Health in Aging: Recent Evidence and Innovations

Abstract

Purpose of Review

We review recent work on applications of non-pharmacologic strategies to promote cognitive health in older adulthood and discuss potential network mechanisms, limitations, and considerations for improving intervention uptake and efficacy.

Recent Findings

In healthy older adults and patients with mild cognitive impairment, cognitive training produces global and domain-specific cognitive gains, though effect sizes tend to be modest and transfer is variable. Non-invasive brain stimulation has shown moderate success in enhancing cognitive function, though the optimum approach, parameters, and cortical targets require further investigation. Physical activity improves cognitive functions in late life, with emerging trials highlighting key intervention components that may maximize treatment outcomes. Multimodal interventions may be superior to single-component interventions in conferring cognitive gains, although interpretation is limited by modest sample sizes and variability in training components and parameters. Across modalities, individual differences in patient characteristics predict therapeutic response. These interventions may advance cognitive health by modulating functional networks that support core cognitive abilities including the default mode, executive control, and salience networks.

Summary

Effectiveness of cognitive enhancement strategies may be increased with clinician-led coaching, booster sessions, gamification, integration of multiple intervention modalities, and concrete applications to everyday functioning. Future trials involving rigorous comparisons of training components, parameters, and delivery formats will be essential in establishing the precise approaches needed to maximize cognitive outcomes. Novel studies using patient-level clinical and neuroimaging features to predict individual differences in training gains may inform the development of personalized intervention prescriptions to optimize cognitive health in late life.

Introduction

Advancing age is accompanied by declines in several core cognitive domains including processing speed, executive functions, and episodic memory [1]. Advances in non-pharmacologic treatment approaches highlight multiple intervention strategies—cognitive training, neuromodulation, and physical activity—that may improve cognitive health in older adulthood by rescuing brain networks that are particularly susceptible to aging, and/or augmenting the functioning of brain networks that are relatively resilient to aging. These approaches may either independently or synergistically enhance cognitive and brain health in late life. Recent studies have begun to identify specific approaches within these domains that may optimize cognitive outcomes.

These interventions may advance cognitive health in older adulthood, in part, by modulating connectivity in functional networks that underlie core cognitive abilities. Among these networks are (1) the executive control network (ECN), which underlies working memory maintenance and updating, task shifting, and cognitive control processes; (2) the default mode network (DMN), which is often considered a “task negative” network that is involved in self-referential thinking and deactivates during goal-directed behavior; and (3) the salience network (SN), which integrates sensory inputs to identify signals of salience and guide attention and goal-directed activity. These networks supporting core cognitive processes are also among those most susceptible to age-related changes in functional connectivity [2, 3]. This review summarizes the recent work on cognitive enhancement interventions, their potential network mechanisms, current limitations, and considerations for improving the uptake and efficacy of these promising interventions.

Cognitive Training

Cognitive training encompasses both cognitive stimulation and strategy-based interventions. Cognitive stimulation approaches, often administered via a computer or other electronic medium, aim to restore or augment specific cognitive functions via challenging cognitive tasks that ideally adapt to an individual’s performance and become progressively more difficult. Cognitive strategy training relies on teaching strategies that help individuals to compensate for and self-manage cognitive errors.

Recent meta-analyses of randomized controlled trials (RCTs) of cognitive training in healthy older adults and patients with mild cognitive impairment (MCI) have demonstrated positive effects on primary cognitive outcomes. In healthy aging, pooled effect size estimates indicate small to medium effects of cognitive training on attention (Cohen’s d = 0.651), processing speed (Cohen’s d = 0.294), executive functions (Hedges’ g = 0.420), visuospatial function (Hedges’ g = 0.183), and memory (Hedges’ g = 0.354) [4–6]. In MCI, pooled effect sizes for cognitive training interventions are similarly modest, with the strongest effects observed on memory and weaker effects on executive functions [5, 7••, 8]. Among these meta-analyses, substantial variability exists in the effect sizes of individual studies resulting in pooled effect size estimates with large confidence intervals. This variability is likely driven by differences in the cognitive training paradigm, choice of control group, and variability in sample characteristics.

Brain Network Mechanisms of Cognitive Training

Both cognitive stimulation and cognitive strategy training modulate functional connectivity within the ECN and DMN [9, 10••, 11, 12], two networks that subserve core cognitive processes and are susceptible to age-related change. Though the ECN and DMN appear to be consistently impacted by cognitive training, the direction of effects varies substantially from study to study [13]. Both increases and decreases in functional connectivity—interpreted as increased neural compensation and increased neural efficiency, respectively—are observed. One hypothesis for this variability in directionality is that changes in functional connectivity may depend on where an individual is in the aging continuum. In younger adults, cognitive training may drive a decrease in functional connectivity and optimize neural efficiency [14, 15]. As the brain ages, there is a loss of white matter integrity and gray matter volume, as well as a loss of healthy segregation of brain networks [16]. Cognitive training-driven increases in networks such as the ECN and DMN may represent compensation for structural damage. This may be especially relevant to individuals with amnestic MCI, considered a prodromal stage to Alzheimer’s disease, in which increased connectivity and activation is seen in the ECN and DMN following training [12, 17, 18], but not changes in neuronal integrity assessed via Positron Emission Tomography [19]. Interestingly, a recent meta-analysis of neuroimaging studies of cognitive training in healthy older adults using activation likelihood estimation found post-training increases in activity in the left ECN and decreases in activity in the right ECN [10••], suggesting that both increased compensation and efficiency may occur.

Intervention Considerations: Transfer and Durability of Cognitive Gains

A longstanding area of controversy in the cognitive training field is whether, and to what extent, training gains transfer to improvements in everyday functioning. This is sometimes referred to as far transfer or environmental transfer. Recent research suggests that transfer remains highly variable with modest effect sizes in older adults [5]. A 2021 meta-analysis of commercially-available computerized cognitive training (CCT) for older adults found a small and non-clinically meaningful effect size for far transfer to tasks of everyday function (Hedges’ g = 0.16) [20]. Another recent meta-analysis of trials that involved training of multiple cognitive domains found small but significant effect sizes for transfer to everyday functioning (Hedges’ g = 0.21–0.25) [5].

Given overall modest effect sizes, there is considerable room for the refinement of cognitive training paradigms to improve their effects both on the trained cognitive domain and on everyday functions. There are several ways in which cognitive training may be enhanced. First, efficacy and transfer may be enhanced by pairing cognitive training with clinician-led coaching. A recent study that combined adaptive CCT for MCI with personally tailored coaching (e.g., goal-setting, problem-solving, self-reflection, and feedback) demonstrated improved cognitive functioning and decreased reliance on caregivers [21]. Such multimodal interventions may enhance transfer by helping patients more clearly connect cognitive skills trained in CCT to everyday life. The patients that may benefit most from adjunctive coaching, and how the cost of coaching (e.g., by a specialist such as a neuropsychologist) is weighed against the potential benefit, requires further evaluation. It may be that aging adults with relatively low motivation or who lack metacognitive awareness of the functional impacts of cognitive abilities may benefit most [22]. More broadly, how CCT and cognitive strategy training may be combined synergistically to potentially bolster efficacy and transfer is an active area of research.

Another recent and promising avenue is the use of gamification. Gamification refers to the incorporation of videogame-based elements into CCT to increase motivation and engagement. Such elements may include the use of avatars, badge or point systems for reinforcement, immersive graphics, levels that increase in difficulty, sound effects, and a story or theme [23]. Gamified cognitive training is found by participants to be more motivating, engaging, and challenging compared to typical cognitive training [24]. This can be crucial for training efficacy as naturalistic studies of CCT suggest low frequency of engagement [25]. In our studies using gamified multitasking training in middle-age and older adults with depression and executive dysfunction, we have found high adherence and improvements in executive functions following training [26, 27]. A gamified iPad-based cognitive training program for episodic memory similarly demonstrated strong enjoyment and motivation in patients with amnestic MCI [28]. Similar to gamification, recent paradigms have been developed that embed cognitive training within simulations of real-world tasks and everyday environments by use of virtual reality [29, 30]. Combining gamified training features with ecologically valid training environments is a promising avenue for further research.

Booster sessions that harness the advantages of telemedicine may bolster long-term efficacy and prolong transfer effects. For example, the addition of telehealth-based cognitive training booster sessions following face-to-face cognitive training is associated with greater maintenance of transfer effects [31]. Booster sessions 3 years following cognitive training are also associated with sustained treatment effects [32]. Finally, efficacy and transfer of cognitive training may be enhanced by pairing cognitive training with neuromodulation and exercise, as we describe in the “Neuromodulation” and “Physical Activity” sections.

Neuromodulation

Non-invasive brain stimulation is another therapeutic tool that has shown moderate success in enhancing cognitive function in aging. Transcranial direct current stimulation (tDCS) is a safe, inexpensive neuromodulatory method which stimulates cortical targets using a weak, direct electrical current delivered using electrodes placed on the scalp. A meta-analysis of 24 articles comprising 566 participants synthesized studies comparing sham to active tDCS in adults aged 60 and over. Compared to sham tDCS, those receiving active tDCS demonstrated significant improvements in episodic memory immediately poststimulation (Hedges’ g = 0.625) and at longitudinal follow-up (Hedges’ g = 0.404), though the cognitive benefits were observed to weaken over time [33]. Moderation analyses indicated that greater benefits on episodic memory were achieved with a tDCS duration of 20 min or less, a larger stimulation target area, and bilateral rather than unilateral stimulation [33]. Along with episodic memory, tDCS has been shown to have a positive impact on global cognition [34], language [35], and visual working memory [36, 37].

Another neuromodulation approach, repetitive Transcranial Magnetic Stimulation (rTMS), has demonstrated efficacy in alleviating cognitive symptoms in patients with MCI. rTMS uses an electromagnetic coil to deliver a magnetic pulse that can be targeted at specific cortical regions to modulate neuronal activity and cortical plasticity. In a meta-analysis of 13 trials, active rTMS compared to sham significantly improved cognitive function, producing a moderate-to-large effect size (Standardized Mean Difference (SMD) = 0.77). Subgroup analyses revealed that low-frequency right DLPFC and high-frequency left DLPFC rTMS enhanced memory function, while high-frequency right inferior frontal gyrus rTMS improved executive function performance [38••]. Consistent with this, another recent meta-analysis of 9 trials found that rTMS improved global cognitive function (SMD = 2.09) in older adults with MCI [39]. Intervention effects were moderated by the frequency and laterality of rTMS stimulation, as well as the number of treatments, with a higher number of treatments (≥ 20) associated with improvements in memory performance. Though subgroup models should be interpreted with caution given the relatively small number of studies [38••, 39], these findings align with other reports that the nature and specificity of cognitive gains may vary depending on the frequency and cortical targets of rTMS stimulation, as well as the number of treatments.

Brain Mechanisms of Neuromodulation

Neuromodulation approaches stimulate the brain in vivo and may impact network connectivity by modulating neuronal excitability and altering cortical function. A study in 21 older adults used functional near-infrared spectroscopy (fNIRS) to examine the mechanisms of left prefrontal anodal tDCS. Results demonstrated a significant impact of tDCS on both working memory performance and bilateral prefrontal hemodynamic activity, providing evidence of tDCS-dependent changes in prefrontal activity in healthy older adults during the execution of a working memory task [40]. Recently, researchers have explored using high definition tDCS (HD-tDCS) as another alternative to tDCS. While conventional tDCS stimulation is relatively diffuse, HD-tDCS uses compact scalp electrodes to generate a more precise and concentrated current. A recent study examined the neural mechanisms of HD-tDCS and found that ten HD-tDCS sessions to the left DLPFC altered the intensity and synchrony of neural activity and increased the coherence (regional homogeneity) of neural activity in multiple brain regions, which may underlie its clinical effects [41].

rTMS may also modulate cortical circuits impacted in aging and associated with cognitive decline. In an RCT, 21 subjects diagnosed with amnestic MCI received either rTMS targeting the right DLPFC or sham stimulation. The subjects also completed resting state functional MRI and neuropsychological assessments before and after the intervention. Relative to the sham group, the rTMS group exhibited improved verbal memory performance and decreased functional connectivity among structures of the DMN. These preliminary data suggest that rTMS-induced hypoconnectivity within the DMN may be associated with cognitive improvements in patients with associated with cognitive improvements in patients with amnestic MCI [42].

Intervention Considerations: Stimulation Parameters and Individual Differences

While studies have demonstrated the potential for neuromodulation to advance cognitive health in late life, translational applications in clinical settings are limited. Before neuromodulation can become a truly effective tool for cognitive enhancement, more research is needed to establish the stimulation approaches, targets, and parameters that may optimize cognitive outcomes in aging. Key factors for further investigation include the optimum intensity, duration and electrode configuration, how these elements may vary depending on the targeted cognitive domain(s), and a better understanding of the relevant neurophysiological parameters and underlying neural mechanisms. Many extant trials are also limited by the use of small samples sizes, and large-scale adequately-powered RCTs are needed to better characterize the cognitive benefits of neurostimulation.

Future studies examining moderators of treatment response will also help to uncover the patient characteristics that may predict or accentuate intervention effects. Individual differences in responsiveness to neuromodulation may depend on age [36], baseline cognitive status [36, 43], and genetic polymorphisms, including the COMT val/val genotype [44]. Inter-individual variability in cortical structure and function may also impact treatment response. One study found that tDCS modulations of network connectivity were greater in magnitude in more compromised circuits, suggesting that individual differences in age- and disease-related network abnormalities may moderate tDCS effects [43]. Given this individual variability, recent work has examined the potential for neuronavigational tools paired with structural and functional imaging to increase the personalization and accuracy of stimulation targets. For example, Beynel et al. (2019) used subjects’ functional imaging maps to identify the left DLPFC target site at the individual-patient level . Administration of rTMS to the target site was associated with improved working memory performance relative to sham rTMS [45••]. Though this work is in the early stages and there is no current consensus on the optimal approach, individualizing target sites based on patient-specific anatomical markers and functional maps may help to refine and personalize neuromodulatory interventions.

Combination Training: Neuromodulation and Cognitive Training

Neuromodulatory techniques have been paired with cognitive training with the aim of enhancing plasticity, training gains, and transfer. Several recent studies have integrated tDCS into cognitive training protocols, which may potentiate the effects of cognitive training in older adults by enhancing functional connectivity in the ECN [46]. Behavioral findings, however, are equivocal and evidence of transfer is limited. Two recent trials demonstrated that augmenting computer-based working memory training with tDCS resulted in moderate effects on working memory performance [36, 37]. Moreover, tDCS of the left DLPFC led to increased functional connectivity among core hubs of the ECN [37]. Another trial evaluating tDCS paired with CCT demonstrated transfer to untrained but related cognitive functions [35]. In contrast, several recent studies involving a combination of tDCS and cognitive training have failed to show separation between active tDCS and sham stimulation for either near or far transfer tasks [47, 48].

Transcranial alternating current stimulation (tACS) has also been studied for its possible cognitive enhancing effects in healthy aging, with mixed results. A recent trial in older adults found that CCT paired with tACS delivered to the left DLPFC had no benefit relative to CCT with sham stimulation in the full sample [49]. However, among the subset of participants with the most pronounced cognitive difficulties at baseline, tACS generated significant improvements in global cognition compared to sham, with gains in performance persisting up to 12 months posttreatment. In a separate study, three sessions of tACS paired with targeted memory training produced a small positive effect on associative memory performance relative to sham stimulation [50]. Though promising, additional research is needed to disambiguate the stimulation targets and parameters that accentuate the favorable effects of cognitive training in older adulthood. Future studies are also needed to establish the ideal pairing of cognitive training and neuromodulation (simultaneous, sequential), the modality and duration of cognitive training, and how this may vary as a function of targeted cognitive domains and clinical presentation.

Physical Activity

Physical activity (PA) is another highly promising approach to preserve and enhance cognitive functions in older adulthood. Structured PA training improves global cognitive function in cognitively unimpaired older adults, and favorably affects several cognitive domains susceptible to age-related decline [51••, 52]. In a recent meta-analysis of 39 RCTs, PA led to modest improvements in global cognition and generated domain-specific gains in attention (SMD = 0.27), memory (SMD = 0.36), and executive function (SMD = 0.34) [53•]. A series of recent trials have also demonstrated the therapeutic potential of PA to preserve cognitive abilities or forestall further decline in individuals with MCI [54–57]. PA significantly improves processing speed, memory, and global cognition (SMD = 0.30) in individuals with MCI [56, 58], and has also been associated with reduced behavioral disturbances [59] and improvements in health-related quality of life [55].

Identifying the training parameters most likely to produce cognitive gains will help to guide the development of tailored exercise programs to optimize cognitive health in aging. The cognitive effects of PA may be determined by several components including intervention duration (number of weeks), session duration (length of each session), frequency (sessions per week), and intensity [60]. Advancements have been made in recent years to establish the ideal PA training parameters and moderators of treatment effects. For instance, an analysis of 98 RCTs reported that studies with a total exercise duration of 52 h or more generated significantly greater cognitive gains than trials of shorter duration [61]. Another meta-analysis found that high frequency PA interventions (5–7 days/week) generated significantly greater cognitive gains (SMD = 0.69) than low frequency interventions (≤ 2 days/week; SMD = 0.32) [53•]. Treatment outcomes may also be moderated by the intensity and type of PA [62]. Interventions that modulate cardiovascular fitness (e.g., aerobic exercise such as running, cycling, or walking) confer greater cognitive benefits than activities focused on strength alone (e.g., resistance training) [51••]. Multi-modal interventions (e.g., aerobic exercise and resistance training) may produce the largest cognitive effects, though the current evidence is mixed [61], and recent trials suggest that the relative superiority of multimodal treatments may be domain-specific (for review, see Barha et al. 2017) [63]. Thus, this emerging work has collectively characterized several key intervention components that may maximize treatment outcomes. However, further studies comparing dose-response parameters are needed to establish those most important for the modulation of cognitive health in late life.

Brain Mechanisms of Physical Activity

By operating through multiple cellular pathways that stimulate neurogenesis and synaptogenesis, studies have shown that PA impacts key structures and networks associated with age-related cognitive decline [52, 64–66]. Abundant evidence from RCTs shows that PA training preserves and even increases hippocampal volume in late life [67, 68], and may mediate exercise-induced improvements in memory performance [69]. Relative to controls, 6 months of aerobic exercise training increased functional network connectivity of the SN in community-dwelling older adults. In a separate trial, PA training altered functional connectivity among key nodes of the ECN and DMN when compared to controls [70]. Consistent with this, exercise-induced changes in aerobic capacity are associated with functional alterations between and within structures of the DMN [71]. Along with modulating brain morphology and function, PA targets peripheral health factors including dyslipidemia, hypertension, and arterial stiffness, that predispose older adults to cognitive decline and contribute to age-related neurodegeneration [72]. Recent work has identified additional circulating factors that mediate the cognitive benefits of PA [73•], which may facilitate more efficient use of PA to maximize cognitive effects.

Intervention Considerations: Motivation and Engagement

Despite the pronounced cognitive benefits of PA, older adults remain the most sedentary subset of the United States population. The majority of adults aged 65 and over (56%) fall well below national public health guidelines for participation in PA [74]. To boost PA among older adults, researchers and clinicians may consider patient-centered modification of current PA programs, as well as alternative approaches to develop accessible yet effective and engaging interventions. For instance, a study in older adults with MCI paired aerobic exercise training with coordinative movements that directly support activities of daily living, and observed benefits in cognition, health-related quality of life, sleep quality, and depressive symptoms [55]. By integrating into exercise programs training components with concrete and direct applications to daily functioning, this may enhance engagement or motivation among healthy and cognitively compromised older adults. Augmenting PA with psychotherapeutic strategies like Motivational Interviewing may also help to identify patient-specific barriers to participation and guide the formulation of individualized PA prescriptions to facilitate engagement.

Finally, pairing novel technology with PA training may produce more stimulating and engaging interventions, and help to motivate regular exercise in late life. For instance, preliminary studies in older adults have demonstrated the feasibility and efficacy of “exergaming,” which integrates PA and interactive video games (e.g., virtual-reality bike rides; pedaling and steering controls in a virtual world with the aim of attaining goals). Exergaming involves video games controlled by bodily movements in a safe surrounding, and can detect and reflect the participants movements in real-time, provide feedback, visual and auditory stimuli, and reinforcement in the form of goals and virtual rewards. Emerging studies demonstrate high rates of usability and adherence to exergames among older adults, and beneficial effects on depressive symptoms and cognitive functioning [75, 76]. Thus, personalizing interventions based on the needs, abilities, and preferences of each individual, including PA components that translate to daily functioning, and integrating PA with novel technology may increase the uptake and efficacy of this promising therapeutic treatment.

Combination Training: Physical Activity and Cognitive Training

Multimodal training may be superior to single modality interventions in conferring cognitive gains, with accumulating evidence supporting the additive and potentially synergistic effects of combined training. For example, in an 8-week RCT of 124 older adults, adding aerobic exercise to a cognitive training program produced broader and more pronounced benefits in processing speed and executive function relative to cognitive training alone [77•]. Similarly, a randomized trial in community-dwelling older adults found that augmenting a PA regimen with cognitive training of executive processes generated greater improvements on untrained measures of executive functioning than PA in isolation [78]. Along with cognitively unimpaired older adults, preliminary studies in individuals with MCI suggest that pairing PA with cognitive training may lead to greater cognitive gains than single modality interventions [79, 80]. Thus, combining PA and cognitive remediation may potentiate or maximize their impact on cognitive functioning in older adulthood [77•, 78, 79, 81].

Cognitive training may augment the cognitive benefits of PA by accentuating the neuromodulatory effects of PA. Specifically, PA facilitates neurogenesis and synaptogenesis, while cognitive remediation promotes cell survival and regulates and strengthens newly formed synaptic connections [82, 83]. Thus, neuroplasticity may be facilitated by PA and guided and functionally integrated into distinct, task-relevant brain networks by cognitive training [81]. Blending these reciprocal neurobiological processes through combined training protocols may maximize circuit modification and enhance cognitive outcomes in older adulthood [84]. However, investigations of combined training interventions remain relatively scarce and, as with the single-modality interventions, interpretation is limited by variability in training components, parameters, and delivery formats [80, 85]. Future trials leveraging multimodal interventions and involving rigorous comparisons to active controls and unimodal therapies will be essential in establishing the precise approaches needed to maximize cognitive health in older adulthood.

Recommendations and Future Directions

Individual Differences and Training Personalization

Greater personalization of interventions, using patient subtyping on the basis of clinical and neuroimaging data in order to predict individual differences in training gains, may amplify the efficacy and transfer of cognitive enhancement strategies. For example, it may be that existing cognitive training paradigms achieve greater far transfer than appears in pooled meta-analyses, but only for select subgroups of individuals. Age, sex, and cognitive status are among the factors that may impact the nature and magnitude of cognitive gains [51••]. For instance, recent reports suggest potential sex differences in the efficacy of PA interventions, with meta-analytic data supporting greater cognitive improvement following exercise training in older adult females relative to males [86]. Cognitive status may also moderate treatment effects. Cognitively healthy older adults and those with mild cognitive deficits tend to show greater cognitive benefits following PA training than those with more pronounced impairments [87]. In contrast, several recent trials combining cognitive training with neurostimulation have found that those with greater baseline impairment show the most improvement in cognitive performance [49]. Thus, there may be an optimal time window to initiate specific treatments to mitigate cognitive decline, and this window may vary across intervention strategies.

Further characterization of patient-level features will facilitate treatment matching and may ultimately lead to more effective interventions and more consistent transfer of gains for the individual patient. Emerging studies have leveraged functional and structural neuroimaging to identify novel neural predictors of therapeutic response. For example, volumetric differences in prefrontal, parietal, and subcortical structures predict individual variability in adherence to a PA training regimen in older adulthood [88]. In addition, greater baseline functional connectivity of the ECN is associated with more persistent training gains after adaptive working memory training [89]. Baseline whole-brain modularity—a measure of network integration and segregation—is also associated with greater cognitive training gains [90]. Future work identifying the clinical, behavioral, and brain network predictors of treatment response will enable the development of individualized intervention prescriptions most likely to optimize cognitive health in older adulthood.