Cognitive abilities

Throughout this narrative review, the term cognitive ageing will be used to refer to changes in thinking, memory and reasoning skills ‘as a process of gradual, ongoing, yet highly variable changes in cognitive functions that occur as people get older’ (Blazer Reference Blazer, Yaffe and Liverman2015, p. 2). As Reference Livingston, Sommerlad and OrgetaBlazer et al continue, ‘Cognitive aging is a lifelong process. It is not a disease or a quantifiable level of function’. Discussion of cognitive abilities, and the age-related changes in them, encompasses a range of skills, from verbal and numerical abilities, attention and processing speed to reasoning and problem-solving, and the various aspects of memory, including working, semantic and episodic memory. The specifics of those are beyond the scope of this narrative review but good overviews are available (Gutchess Reference Gutchess and Thomas2020).

In terms of the range of thinking skills, however, a consistent finding replicated across hundreds of studies is that performance on measures of these are positively correlated, referred to as the positive manifold (Tucker-Drob Reference Tucker-Drob2019a). That is, someone who generally performs well at one type of task will likely perform well at others; note though, that this is ‘generally’ and ‘likely’, as different abilities have different intercorrelations, which allows the structure of broad cognitive domains to be defined. Although the structure of cognitive abilities has been debated, there is broad consensus of there being shared variance across diverse abilities, perhaps accounting for about 40%, as well as domain-specific variance (Carroll Reference Carroll1993).

What goes when?

Although being defined broadly for current purposes, it is important to note that the trajectory of cognitive ageing varies by ability (Hedden Reference Hedden and Gabrieli2004; Hertzog Reference Hertzog, Kramer and Wilson2008). The question of when cognitive decline might become apparent is of central importance but is not necessarily simple to answer. Processing speed, for example, is suggested to peak in the early 20s and decline gradually across the remainder of the lifespan. That relatively early peak and decline is atypical, as many skills continue to develop through mid-life (Hedden Reference Hedden and Gabrieli2004; Blazer Reference Blazer, Yaffe and Liverman2015). A useful distinction is often made between fluid and crystallised abilities. Fluid abilities are often defined as being more age sensitive and therefore decline earlier, perhaps indicating a biologically limited aspect of brain function (Hedden Reference Hedden and Gabrieli2004; Tucker-Drob Reference Tucker-Drob2019a). Crystallised abilities, so named to engender a sense of their fixedness, are more resistant to age-related changes and continue to develop through mid-life and into old age, with vocabulary or semantic knowledge often cited as exemplars of such age-resistant abilities (Hedden Reference Hedden and Gabrieli2004; Blazer Reference Blazer, Yaffe and Liverman2015).

In addition, it is often discussed how much of the age-related change in cognitive abilities is due to shared variance across domains and how much is specific. A recent meta-analysis suggested that about 60% of the variance in cognitive change was shared, directing attention towards the mechanisms that might underly this ‘general factor of cognitive aging’ (Tucker-Drob Reference Tucker-Drob, Brandmaier and Lindenberger2019b), while simultaneously highlighting that some changes are specific to cognitive domains.

It is important to clarify that, when discussing cognitive changes with age, these are ‘average’ changes observed in studies of many hundreds, or indeed thousands, of participants. While we might all experience these changes, their extent and timing are variable. Indeed, it is the existence of this person-to-person variation that supports attempts to identify the potential determinants of cognitive ability level and change. Figure 1 illustrates that variation with four hypothetical people and ‘cognition’ as a broad term. The shading suggests a general area of opportunity; that is, across the lifespan we might have the potential to be performing better or worse at any given age depending on the influence of a range of factors. In the illustration, all four people start at the same level, but their patterns of cognitive development vary from early adulthood through mid-life and into old age. Person A, for example, appears to show continued growth through mid-life, perhaps as the result of the opportunities presented by pursuing higher education followed by a mentally challenging occupation. Person C, meanwhile, appears to be declining slightly more than average from their 20s onwards, before a stabilisation in their 60s, perhaps as a result of retirement from a less challenging occupation followed bytaking up some new post-retirement opportunities (Hertzog Reference Hertzog, Kramer and Wilson2008). A consideration of some of the lifestyle and behavioural factors that might determine better or poorer trajectories forms the basis of the second half of this review.

FIG 1 Depiction of the zone of possible cognitive development for a given individual, along with four developmental curves (A, B, C, & D) indicating specific possible outcomes. Each possible curve starts from the same functional level at age 20, with different trajectories resulting as a function of interactions among behavioral, environmental, and genetic factors that permit vertical movement within the zone at different points in the life span” (Hertzog Reference Hertzog, Kramer and Wilson2008). Reproduced with permission from SAGE Publications.

Genes versus lifestyle

Before exploring the determinants of cognitive ageing, and specifically potentially malleable determinants, it is useful to summarise the extent to which genetic factors might account for individual variation in cognitive ability level and change. Studies have estimated the heritability of cognitive ability to increase from around 20% in early childhood to 60% in adulthood (perhaps being as high as 80% in old age) (Plomin Reference Plomin and Deary2015). Although there is clearly a genetic component, research over the past decade using genome-wide association techniques has suggested that cognitive ability, as with many other traits, is ‘highly polygenic’ (Davies Reference Davies, Tenesa and Payton2011). Or rather, that many thousands of genetic variants each make very small contributions: the ‘largest effect sizes of associations are very small indeed […] If the largest effect sizes are so small, the smallest effect sizes must be infinitesimal’ (Plomin Reference Plomin and Deary2015). Even consortia combining results from over 50 000 participants have accounted for just under 30% of the variance in general cognitive ability, suggesting ever larger samples will be required to identify ever smaller contributions (Davies Reference Davies, Armstrong and Bis2015).

In discussing genetic contributions, it is important to distinguish between predicting individual variability in cognitive ability level versus change over time. In one study with measures of cognitive ability across the lifespan, for example, it was suggested that genetic factors might account for about 25% of the variability in cognitive change from childhood (age 11) to later life (age 70) (Deary Reference Deary, Yang and Davies2012); the counterpoint, therefore, was that environmental factors might account for the larger proportion of the variability in cognitive change across the lifespan. In terms of changes during old age, research with the same sample did not find associations between polygenic scores and cognitive change from age 70 to 79 (Ritchie Reference Ritchie, Hill and Marioni2020). More recently in another longitudinal sample, researchers suggested that a polygenic risk score for late-onset Alzheimer's disease was associated with 25-year change in cognitive ability, although partly confounding mid-life and later-life changes, given the nature of the sample (Kauppi Reference Kauppi, Rönnlund and Nordin Adolfsson2020).

Understanding the extent of the genetic contribution provides a useful focus: how much of cognitive ability level or change might we be able to influence, or rather, what is our target? In summarising briefly, then:

• • Our cognitive abilities develop and change across the lifespan, including age-related changes in old age;

• • The timing and extent of these changes vary by the type of ability;

• • People also vary in the extent of age-related change they experience (and indeed the peak they might reach earlier in the lifespan);

• • Person-to-person variation directs attention towards identifying the genetic and environmental determinants, with some evidence suggesting a stronger environmental contribution to lifetime cognitive change;

• • And that directs attention towards ‘what keeps us sharp?’.

Education

Education is associated with higher cognitive performance, but the effect of education as a determinant of cognitive ageing trajectories has been less clear (Seblova Reference Seblova, Berggren and Lövdén2020). A recent review suggested that education might increase cognitive ability, reported as an increase of 1–5 IQ points per additional year of education (Ritchie Reference Ritchie and Tucker-Drob2018). Interestingly, the benefit of education was suggested as persisting across the lifespan, supported by a review by Chapko et al (Reference Chapko, McCormack and Black2018) in which education was one of the few consistently reported determinants of cognitive reserve.

There is less certainty about whether education modifies the rate of cognitive change in old age. The most recent meta-analysis described education as having a negligible effect on cognitive ageing, such that ‘the association of education with change is at least 12 times smaller than the corresponding association with level’ (Seblova Reference Seblova, Berggren and Lövdén2020). That is not to say that education is not relevant to cognitive health in old age: as it would be considered one of the factors that support development across the lifespan, education's role may be in providing a higher starting point from which later change might occur (Chapko Reference Chapko, McCormack and Black2018; Ritchie Reference Ritchie and Tucker-Drob2018; Seblova Reference Seblova, Berggren and Lövdén2020).

Occupational complexity

The complexity of one's occupation is often considered as a potential determinant of later cognitive ability level and change. It is, however, important to reiterate that occupational complexity is likely to be an outcome of earlier ability, education and a range of other social determinants. That said, the complexity of work with people and data might be associated with cognitive ability level after controlling for those factors (Smart Reference Smart, Gow and Deary2014). However, studies that have followed participants through the retirement transition have noted that any benefit of occupational complexity may be lost post-retirement (Finkel Reference Finkel, Andel and Gatz2009). That is, continued engagement might be important in terms of maintaining any cognitive benefits observed.

Chapko et al (Reference Chapko, McCormack and Black2018) suggested that the impact of occupation might best be described as inconsistent, perhaps as a result of the differential effects before and after retirement. For example, some studies have provided support for occupational complexity being protective against decline (Andel Reference Andel, Silverstein and Kåreholt2015), whereas others have not (Finkel Reference Finkel, Andel and Gatz2009). That is not necessarily suggestive that higher complexity is a predictor of decline per se, but rather that those who have achieved a higher level of ability as a result of their occupational engagement may appear to experience faster change once that stimulation is withdrawn (Hyun Reference Hyun, Katz and Lipton2019).

Smoking

Among the lifestyle factors consistently cited as being harmful to cognitive abilities is smoking. That said, a meta-analysis reported that the effect on cognitive decline was not significant (Peters Reference Peters, Poulter and Warner2008), whereas an earlier review reported greater decline in smokers (Anstey Reference Anstey, Von Sanden and Salim2007). Although smoking status may not be associated with the rate of decline, more recently reported in Deal et al (Reference Deal, Power and Palta2020), that is not to say that smoking is not harmful to cognitive health. If, for example, there is a consistent association between smoking and a lower level of cognitive ability in later life, those who smoke will be declining from that lower starting point.

Diet

In terms of diet, it is common to see studies reporting specific nutrients as being beneficial for brain health. However, there are often inconsistencies in the findings reported, specifically between observational versus randomised controlled trials, and it would be difficult to produce a detailed list of recommendations from hundreds of individually considered nutrients. It is therefore useful, of course, that individual nutrients exist within the context of more easily understandable dietary patterns, and it is increasingly common to consider dietary patterns versus the ‘reductionist approach’ (Jennings Reference Jennings, Cunnane and Minihane2020). Similar to the ‘marginal gains’ argument for lifestyle factors more generally, individual nutrients may make only small, and not necessarily independent, contributions; any potential benefit likely derives from their cumulative influence within particular dietary patterns (and then again, diet is one potentially protective factor among many).

Whole-diet approaches have often considered the Mediterranean diet or combinations of that with other diets specifically designed to reduce hypertension, for example. A review suggested that those most closely adhering to a Mediterranean diet were less likely to develop an incident cognitive disorder (Wu Reference Wu and Sun2017), with one study estimating that each extra point adherence to the Mediterranean diet was associated with about 1.7 years fewer years of cognitive ageing (Shannon Reference Shannon, Stephan and Granic2019). Jennings et al (Reference Jennings, Cunnane and Minihane2020) also highlighted that dietary patterns provide better targets for intervention than specific components.

Dietary supplements have also come under scrutiny as potentially beneficial for brain health. However, a report from the Global Council on Brain Health (Box 3) suggested, among other statements, that it was not possible to ‘endorse any ingredient, product or supplement formulation specifically for brain health, unless your health care provider has identified that you have a specific nutrient deficiency’.

Mental, social and physical leisure activities

In considering lifestyle and behavioural factors associated with cognitive health, engagement in mental, social and physical leisure activities is commonly proposed as beneficial (Hertzog Reference Hertzog, Kramer and Wilson2008; Bielak Reference Bielak2010). Mental engagement is sometimes considered a key factor, even being framed in the adage ‘use it or lose it’. Briefly, that encapsulates the sense that the deployment of cognitive abilities in completing mentally stimulating activities (or indeed educational or occupational pursuits) is what builds cognitive capacity or reduces decline. That said, the specificallymentally challenging aspects may be more prone to confounding by reverse causation (Gow Reference Gow, Corley and Starr2012). In general, however, it is often reported that more active individuals have a higher level of cognitive ability and a reduced rate of change (Hertzog Reference Hertzog, Kramer and Wilson2008), although the latter aspect is less consistently supported.

Intervention studies have been employed to address some of those inconsistencies, but also to explore the potential causal pathways between a change in some aspect of activity and changes in cognitive ability. In reviewing ‘real-world’ interventions, it was suggested that benefits were clearest from physical activity, and more so for the cognitive abilities sensitive to age-related decline (Vaportzis Reference Vaportzis, Niechcial and Gow2019). In a more recent review combining leisure activity and other psychosocial aspects, it was suggested that ‘an evidence-based intervention strategy to improve global cognition, memory and executive functioning would be a group therapy carried out for 4 months or more, promoting regular (at least weekly) activity, involving aerobic or resistance exercise, with a cognitively demanding and/or creative component’ (Whitty Reference Whitty, Mansour and Aguirre2020).

There are, however, specific difficulties in providing summaries, given the diverse nature of activities and interventions considered. Direct comparisons, and therefore meta-analyses, are often challenging.

Studies of lifestyle or activity-based interventions are few, owing to the complexity of running them, but there is a large literature on cognitive training paradigms (often commonly referred to as brain training). Targeted cognitive training has been reported as beneficial; that is, people show improvement on the specific tasks trained. However, there remains debate, and controversy, in the field regarding what that means, both theoretically and indeed practically (Simons Reference Simons, Boot and Charness2016). In conducting their review, Simons et al (Reference Simons, Boot and Charness2016) noted that methodological limitations restrict the utility of many studies. Furthermore, the Global Council on Brain Health (Box 3) offer the following careful statement:

Brain games may be fun for many, but, as suggested by the Global Council on Brain Health report, the weak or non-existent evidence for their benefits limits broader recommendations for them.