Cognitive resilience/reserve: Myth or reality? A review of definitions and measurement methods

Abstract

INTRODUCTION

This review examines the concept of cognitive reserve (CR) in relation to brain aging, particularly in the context of dementia and its early stages. CR refers to an individual's ability to maintain or regain cognitive function despite brain aging, damage, or disease. Various factors, including education, occupation complexity, leisure activities, and genetics are believed to influence CR.

METHODS

We revised the literature in the context of CR. A total of 842 articles were identified, then we rigorously assessed the relevance of articles based on titles and abstracts, employing a systematic approach to eliminate studies that did not align with our research objectives.

RESULTS

We evaluate—also in a critical way—the methods commonly used to define and measure CR, including sociobehavioral proxies, neuroimaging, and electrophysiological and genetic measures. The challenges and limitations of these measures are discussed, emphasizing the need for more targeted research to improve the understanding, definition, and measurement of CR.

CONCLUSIONS

The review underscores the significance of comprehending CR in the context of both normal and pathological brain aging and emphasizes the importance of further research to identify and enhance this protective factor for cognitive preservation in both healthy and neurologically impaired older individuals.

Highlights

• This review examines the concept of cognitive reserve in brain aging, in the context of dementia and its early stages.

• We have evaluated the methods commonly used to define and measure cognitive reserve.

• Sociobehavioral proxies, neuroimaging, and electrophysiological and genetic measures are discussed.

• The review emphasizes the importance of further research to identify and enhance this protective factor for cognitive preservation.

1. INTRODUCTION

The term “reserve” in biology and medicine indicates the ability of an individual to maintain a given function (at the level of cellular assemblies, organ, or body systems) despite the presence of damage or disease; the same concept applies to the brain and its cognitive functions. ¹ Indeed, some people have a greater cognitive reserve (CR) than others, which allows them to maintain their cognitive function in the face of physiological (ie, age‐related) and pathological brain aging. ¹ Although the concept of reserve is far from being clear, even more difficult is the way of measuring it. The importance of such a neuroscientific topic can be appreciated by considering brain aging as the continuous balance between “protective” and “risk” factors for neurodegeneration: physiological aging being the result of the former, while progressive and pathological brain aging reflects the prevalence of the latter. Along this line of reasoning, it is evident that in recent years most of the pharmacological and nonpharmacological research to treat dementias aimed to identify/modify risk factors, while much less effort has been devoted to identifying and potentiating protective factors.

In this review, we will summarize and critically evaluate the current state of research on CR in physiological and pathological brain aging, particularly Alzheimer's disease (AD) and its prodromal stages. We will highlight the challenges and limitations of the various methods adopted so far to define and measure CR, as well as the need for more focused research to better understand such complex concepts. Additionally, we will explore the neural implementation of CR, the relationship between brain aging and CR, and the relationship between CR and pathological brain aging, such as dementia.

RESEARCH IN CONTEXT

1. Systematic review: This review provides an overview of cognitive reserve (CR) in relation to brain aging, focusing on dementia and its early stages.

2. Interpretation: CR refers to the ability to maintain or regain cognitive function despite brain aging, damage, or disease. Factors such as education, occupation complexity, leisure activities, and genetics influence CR. We critically evaluate commonly used methods to define and measure CR, including sociobehavioral proxies, neuroimaging, electrophysiological, and genetic measures. The challenges and limitations of these measures are discussed, highlighting the need for targeted research to improve understanding, definition, and measurement of CR.

3. Future directions: Further research is crucial to identify and enhance this protective factor for cognitive preservation in both healthy and neurologically impaired older individuals, considering both normal and pathological brain aging. Exploring the mechanisms behind cognitive resilience and reserve can potentially lead to intervention strategies aimed at slowing down or preventing cognitive decline.

One problem with measuring CR is that there are no uniform methods or guidelines. Different studies have used different definitions and measurements to estimate CR, such as educational level, occupational complexity, and leisure activities. However, these measures are not always directly related to CR, and do not provide an accurate assessment on an individual basis, making it difficult to disentangle the mutual contributions of brain pathology and CR to cognitive impairment. In particular, in the following paragraph, we will summarize the concept of CR (also referred to alternatively as “reserve” in the text below) and the most commonly used metrics (usually named proxies) for measuring it. The second part of this review will be divided with respect to the proxies, namely in the neuropsychological, sociobehavioral, neuroimaging, electrophysiological, and genetic domains for physiological and pathological brain aging (Figure 1).

FIGURE 1.

Proxies of cognitive reserve (CR). Comprehensive representation of factors contributing to CR, quantifiable through various elements including genetic and sociobehavioral factors, proxies derived from brain imaging measurements, and electrophysiological measures from EEG/MEG. A novel perspective, detailed in the manuscript, introduces the concept of network analysis and graphs as a new framework for understanding CR. EEG, electroencephalography; MEG, magnetoencephalography; MRI, magnetic resonance imaging; PET, positron emission tomography.

1.1. Physiological and pathological brain aging

Normal brain aging represents a physiological and gradual process during which some cognitive changes, mainly including difficulties in short‐term memory, processing speed, and executive functions, may commonly occur. ² Biologically, such cognitive decline seems to be due to structural (ie, cerebral atrophy in grey and white matters) and functional (ie, alterations in brain connectivity and neurotransmission) age‐related brain modifications. ³ However, in physiological aging—mainly because of neuroplasticity and neural reserve (NR)—global functioning is well‐maintained.

By contrast, dementia is a pathological age‐related brain degenerative syndrome characterized by a progressive decline in cognitive functioning, severe enough to negatively impact personal functional abilities and daily living. ⁴ Nowadays, AD represents the most common form of dementia. ⁵ Indeed, dementia can be considered as the final and irreversible syndrome stage where several neurodegenerative “killers” have progressively acted together and converged. Different and heterogenous mechanisms, including protein aggregations (eg, amyloid deposition, neurofibrillary tangles, alpha‐synuclein accumulation), synaptic traffic failures, inflammation, and neuronal death cumulatively producing cerebral network disconnections, have been supposed to be the leading causes of dementia. ⁶ It is now evident that prolonged (years or decades) preclinical and prodromal phases can precede the onset of dementia. The prodromal stage is better defined as mild cognitive impairment (MCI). By definition, MCI refers to a transitional state between normal aging and mild dementia, where cognitive performances are impaired more than expected for individual age or education while global functioning is fully preserved. ⁵ Due to the cognitive domain affected, MCI can be categorized into single‐domain or multiple‐domain and amnestic MCI (aMCI) or nonamnestic MCI (naMCI). ⁷ , ⁸ , ⁹ Individuals with MCI have been recognized as a high‐risk population for dementia onset and, therefore, accurate diagnostic evaluations and therapeutic strategies are required.

In this paper, the term “physiological aging” will refer to the concept of normal aging, whereas “pathological brain aging” will refer to neurodegenerative disorders.

1.2. Literature search procedure

1.2.1. Database selection

In undertaking our literature review, we utilized reputable academic databases, namely PubMed, Scopus, and Google Scholar. This selection aimed to ensure a comprehensive exploration of the literature pertaining to our review topics.

1.2.2. Keyword selection

A set of relevant keywords was carefully chosen to refine our search scope. These keywords included terms such as “Cognitive Reserve,” “Brain Reserve,” “fMRI” (functional magnetic resonance imaging), “MRI,” “PET” (positron emission tomography), “Neuroimaging,” “EEG” (electroencephalography), “MEG” (magnetoencephalography), “Proxies,” “Alzheimer,” “Dementia,” “Aging,” and “Cognitive Decline.” Boolean operators (AND, OR) were employed to enhance the precision of our search queries. We have used this combination of keywords and Boolean operators: (“Cognitive Reserve” OR “Brain Reserve”) AND (“fMRI” OR “MRI” OR “PET” OR “Neuroimaging” OR “EEG” OR “MEG”) AND (“Proxies”) AND (“Alzheimer” OR “Dementia” OR “Aging” OR “Cognitive Decline”).

1.2.3. Inclusion and exclusion criteria

To focus our search and maintain relevance, we included studies published within the last 15 years and restricted our search to articles written in English. We found 842 articles aligning with this research.

1.2.4. Search strategy

Our search process involved an initial broad exploration using the identified keywords. We rigorously assessed the relevance of articles based on titles and abstracts, employing a systematic approach to eliminate studies that did not align with our research objectives. The remaining articles underwent a more thorough examination during which we assessed the full text to ensure alignment with our research question. This phase allowed us to prioritize studies with methodologies and findings that directly contributed to the objectives of our research.

2. BRAIN AND CR/RESILIENCE

According to the Collaboratory on Research Definitions for Reserve and Resilience in Cognitive Aging and Dementia, the term “resilience” can be identified as a versatile term including every definition in the context of cognitive health and well‐being. It encompasses the brain's remarkable ability to sustain cognition and functionality, despite the challenges posed by aging and disease. The mechanisms responsible for resilience can exhibit significant diversity. These concepts shed light on the multifaceted nature of resilience and offer insights into how the brain adapts and maintains its optimal reserve under different circumstances. The current operational definition of resilience relies on neuropathology findings after death, categorizing individuals as resilient or not in a dichotomous manner. However, this approach may not capture the continuum of resilience that occurs throughout the aging process. It may be more clinically useful to adopt a continuous scale of resilience or consider multiple levels of resilience. For example, individuals who live to advanced ages without any signs of cognitive decline could be considered resilient regardless of neuropathology outcomes. Incorporating additional genetic, imaging, and other data into the definition would provide a better understanding of the phenotype of resilience before death. Within the larger and more general definition of resilience, over the years, numerous researchers have identified the concept of reserve and two related important subconcepts as represented by “brain reserve” (BR) and CR ¹⁰ , ¹¹ , ¹² (as shown in Figure 2). Colloquially, in engineering terms, BR might be considered the “hardware” while CR the “software,” namely the former refers to the structural differences in the brain that allow for greater resistance to neuronal/synaptic loss and brain damage, while the latter refers to functional differences and to the ability of an individual to use cognitive capabilities. BR is determined by factors such as brain size, measured as head circumference or intracranial brain volume, cortical thickness, the number of neurons and synapses, and the strength and architecture of neural connections. ¹¹ , ¹³ If we consider that neurodegenerative mechanisms are characterized by a progressive and slow loss of neural and network substrates, which is counteracted by BR providing support from a deposit of “functionally silent” or “vicarious” (thanks to plasticity) neurons and synapses, we understand the main reason for a prolonged preclinical and presymptomatic stage during which neurodegeneration in pathological brain aging (ie, AD, Parkinson's disease) is already present, but symptoms are absent or extremely scanty. Individuals with larger brains tend to tolerate a greater extent of degenerative and progressive pathology before they reach the functional threshold at which symptoms become clinically manifest; in other words, their preclinical and presymptomatic stage is significantly longer. A major limitation of the “large brain” concept of reserve is that it is a static and passive viewpoint and not valid in a generalized way. ¹⁰

FIGURE 2.

Implementation of cognitive reserve. Schematic illustration depicting the subdivision of reserve into cognitive reserve (CR) and brain reserve. CR encompasses neural processes, manifesting as neural reserve and neural compensation. Neural reserve refers to the brain's inherent capacity to optimize performance and withstand pathology, while neural compensation involves adaptive mechanisms that counteract cognitive decline.

On the other hand, CR refers to the ability of an individual to make flexible and efficient use of available brain reserve when performing tasks and—at the same time—to compensate, at least in part, for brain damage. ¹⁴ This can include problem‐solving skills, various types of memory, and the ability to adapt to new situations. Indeed, individuals with a higher CR are able to maintain cognitive function despite brain damage due to neurodegeneration, whereas those with a lower CR are more prone to cognitive decline. ¹⁰ , ¹⁵ Proxies of CR refer to measures that are currently used but may not directly reflect CR. Commonly employed proxies of CR include educational level, occupational complexity, and leisure activities. ¹⁶ , ¹⁷ Education is one of the most significant factors, as individuals with higher levels of education tend to have a greater CR. This is probably due to the fact that education trains individuals to a wide range of information processing and skills, which can help develop and strengthen neural connections. ¹⁸ Individuals with higher levels of education and professional occupation tend to experience cognitive symptoms at a more advanced stage of brain neurodegeneration compared to those with a lower level. However, among subjects with a similar clinical severity of cognitive impairment, high‐educated individuals seem to show a greater burden of brain pathology and a more rapid progression of disease. ¹⁹

Professional activity can also play a role in CR, as individuals chronically engaged in mentally stimulating activities, such as those in professions that require problem‐solving and decision‐making, tend to have a greater CR. ²⁰ Leisure activities such as reading, playing musical instruments, and engaging in social activities can also contribute to CR. ²¹ These measures are often used in research studies as a way to infer CR, but they may not provide an accurate assessment of CR as they are not always directly related to CR. Additionally, these proxies may not take into account the different ways an individual can compensate for brain damage and maintain cognitive function, making it difficult to estimate CR accurately.

A further concept, that  it  has spread lately, is brain maintenance (BM) which refers to the state of minimal changes in neural resources and neuropathology over time, contributing to preserved cognitive function in older age. ¹⁰ It is influenced by various genetic and environmental factors throughout the lifespan. BM can be operationally defined as minimal changes in brain markers of aging or disease, coupled with the preservation of cognitive function. ¹⁰ Research on BM requires measuring age‐related brain changes, injuries, or diseases that impact cognition and monitoring cognitive changes over time. Longitudinal studies are essential to examine the relationship between individual differences in brain anomalies and cognitive decline. BM and CR are complementary concepts, with BM focusing on cognitive trajectories related to brain change and CR addressing cognitive trajectories while controlling for changes in neural resources or neuropathology. ¹⁰

TABLE 1.

Definitions related to the concept of reserve.

Concept	Definition
Brain reserve (BR)	BR is determined by factors such as brain size, measured as head circumference or intracranial brain volume, cortical thickness, the number of neurons and synapses, and the strength and architecture of neural connections.
Cognitive reserve (CR)	CR refers to the ability of an individual to cope with brain damage by using preexisting cognitive processes or by using compensatory processes.
Brain maintenance (BM)	BM refers to the state of minimal changes in neural resources and neuropathology over time, contributing to preserved cognitive function in older age.
Neural reserve (NR)	NR refers to the idea that there is interindividual variability in the primary brain networks that underlie task‐related performances.
Neural compensation (NC)	NC refers to the process by which individuals with brain damage use alternate/vicarious brain structures or networks to compensate.

Note: This table provides concise and clear definitions for key concepts associated with the reserve framework, elucidating the nuanced distinctions among brain reserve, cognitive reserve, brain maintenance, neural reserve, and neural compensation.

FIGURE 3.

The concepts of efficiency and capacity. Neural activity increases with increasing task demands with two different efficiencies and capacities. In particular, neural activity increases with increasing task demands at two different rates representing low and high efficiency. Additionally, neural activity increases with increasing task demands to two different maximum levels. Once neural activity reaches this high or low‐capacity limit, even greater task demands do not cause further changes.

3. NEURAL IMPLEMENTATION OF CR

Since the CR proxies are not direct measures of CR and do not reflect the underlying brain processes through which CR operates, in 2009, Stern introduced the concept that CR might take two forms: NR and neural compensation (NC) ¹ (as shown in Figure 2). Distinguishing between these possible neural implementations of CR can be an important starting point for designing, analyzing, and interpreting functional imaging and neurophysiological studies in this area. ²²

The first concept of NR refers to the idea that there is interindividual (partly genetically determined) variability in the primary brain networks that underlie task‐related performances. ²³ The second concept of NC refers to the process by which individuals with brain damage use alternate/vicarious brain structures or networks to compensate. This alternate network may not be used until ongoing activity‐related demands exceed a threshold level or until the neural function within a primary network fails to maintain sufficient capacity to cope with ongoing activity. ²⁴ Definitions related to the concept of reserve are shown in Table 1.

Several models have been proposed to explain the neural implementation of CR. One popular model is the efficiency‐capacity model, in which CR is related to the efficiency and capacity of neural networks ²⁵ , ²⁶ , ²⁷ (Figure 3). These are two important aspects that refer to the functional differences in the brain. Efficiency implies the ability of the brain to perform a task with a progressively lower level of activation—after training—of dedicated (task‐related) neural networks. Just to give an example, when a complex motor action is performed for the first time, a complex network is needed with several nodes activated in series and in parallel, including contralateral sensorimotor primary, premotor, and supplementary motor, as well as ipsilateral primary sensorimotor cortices. After training, when the complex movement has been fully learned (automated/embodied), despite being carried out with optimal performance, only the contralateral sensorimotor cortices remain active. In other words, it is an intrinsic ability of the primate brain to perform a task using the lowest possible amount of network complexity (and probably energy consumption). Within the realm of cognitive functions, an individual with a higher neural efficiency—for instance also due to innate talent—may be able to perform a cognitive task with less activation of neural networks than another individual with lower neural efficiency after the same level of training/learning. ²⁶ Capacity, on the other hand, refers to the ability of the brain to activate neural networks to a progressively higher degree as the task difficulty increases. Also in this case, an individual with a higher neural capacity may be able to perform a cognitive task with greater activation of neural networks. The situation can be different in various cognitive domains; in other words, the same individual can be highly efficient in a given domain and its related tasks while being less efficient in another domain, and vice versa. ²⁷ Altogether, efficiency and capacity reflect the “software” of the brain, which allows for greater flexibility and adaptability not only in everyday life, but also in reacting to a brain damage. Individuals with higher neural reserve, characterized by both efficiency and capacity, may therefore have more efficient cognitive networks and higher capacity networks. Individuals with a greater neural reserve may have cognitive networks that are more efficient, meaning they require less activation to perform tasks at a similar level of performance compared to less efficient networks. Additionally, they may also have cognitive networks with a higher capacity, allowing them to activate such networks to a greater degree as the task becomes more difficult. Essentially, neural reserve may explain interindividual differences in cognitive tasks, suggesting that these differences may explain why some people are more vulnerable to the same type of brain damage or disease stage than others. ²³

It has been found that as age‐related brain modifications occur, a neural network that was used with high efficiency in the younger age becomes less efficient. ¹⁰ However, older people with high CR are still able to use networks efficiently compared to older people with low CR. It has been demonstrated that, in older age, people can additionally use a second compensatory network.

4. GENETIC FACTORS AND CR

Genetic factors play a crucial role in CR since they may influence the brain's structural and functional characteristics, including the number and distribution of neurons and synapses, as well as the efficiency of neural networks and signaling pathways. It is known that some genetic variants are associated with increased risk for AD and reduced CR ²⁸ ; in particular, the apolipoprotein E (APOE)‐ε4 allele is associated with an increased risk for AD and a faster rate of cognitive decline. Several studies have found that the presence of at least one APOE‐ε4 allele was associated with (and possibly contributing to) impaired cognition and the presence of multiple proteinopathies, while APOE‐ε2 was protective. ²⁹ These results indicate that maintenance of normal cognition may depend on resistance to the development of multiple concurrent proteinopathies. On the other hand, other genetic variants, such as those involved in brain‐derived neurotrophic factor (BDNF) signaling, have been linked to enhanced CR and better cognitive performance. BDNF is a protein that promotes the growth and survival of neurons in the brain ³⁰ ; it is involved in a variety of processes related to learning, memory, and cognitive function. Variations in the BDNF genes can influence BDNF production and signaling in the brain and may be associated with differences in CR and cognitive performance. In fact, the BDNF gene plays a crucial role in both the development of the brain and its ability to adapt to environmental changes in adulthood, which suggests that these genetic variations could influence the creation of neural networks that underlie CR. BDNF Met carriers (which is also known as the Val66Met polymorphism, a substitution of the amino acid valine [Val] with methionine [Met] at position 66 of the BDNF protein), may have a lower capacity for neuroplasticity and cognitive flexibility, which are important components of CR. ³¹ These individuals might compensate less for age‐related changes or neurological damage by reorganizing or rewiring their neural networks to maintain cognitive function. In a study conducted by Ward et al., the impact of BDNF and APOE on functional connectivity and CR was examined in 76 cognitively normal participants (52 females, 24 males) aged between 53 and 81 years. The study found that the BDNF Val66Met polymorphism influenced the relationship between executive network connectivity and CR, assessed by Wechsler Test of Adult Reading, the Lifetime of Experience Questionnaire, and the number of years of prior formal education. ³² Specifically, exposure to a more cognitively enriched environment was associated with increased functional connectivity in certain brain regions, including the dorsal attention network (DAN), hippocampal and amygdala regions, subcallosal cortex, white matter callosal cortex, and the postcentral gyrus in individuals who were Val homozygotes, but not in those who were Met carriers, suggesting that CR might have varying cognitive effects depending on the BDNF Val66Met polymorphism. However, there is a contrasting finding where functional connectivity within the Default Mode Network decreases with increasing CR. In a recent study, the polygenic risk score for β‐amyloid (PRSAβ42) was utilized as a proxy for AD pathology. ³³ The study aimed to explore the relationship between PRSAβ42 and the incidence of amnestic mild cognitive impairment (aMCI)/AD and examine the impact of CR (measured as education) on this association. Educational years were used as a proxy for CR. This study found that higher PRSAβ42 levels were linked to a 33.9% increase in the risk of developing aMCI/AD, while greater CR was associated with an 8.3% reduction in this risk. Additionally, an additive interaction was observed between PRSAβ42 and CR, whereby individuals with high levels of CR and PRSAβ42 showed a 62.6% reduction in the risk of developing aMCI/AD compared to those with high levels of PRSAβ42 alone. ³³ These findings provide support for the CR hypothesis, indicating that early‐life education may lessen the impact of genetically mediated amyloid accumulation on the development of aMCI/AD. These results have significant clinical implications, as they suggest that interventions designed to enhance CR, whether implemented early or later in life, could potentially delay cognitive decline and prevent AD, even in individuals at high genetic risk for amyloid brain pathology.

5. PHYSIOLOGICAL BRAIN AGING AND CR

Physiological brain aging refers to the physiological age‐related changes that occur in the elderly brain and which can impact cognitive function, including memory, processing speed, and attention. These changes do not necessarily indicate a progressive neurodegenerative disease. It is widely believed that CR can help to mitigate the effects of physiological brain aging by allowing the brain to utilize alternate neural pathways and recruit more complex networks and additional neural resources to maintain task performance at a level similar to younger adults.

5.1. Sociobehavioral and neuropsychological proxies of CR

Several studies have examined the influence of aging‐related brain changes on cognition and how CR can modulate these changes. ³⁴ Higher education has been linked to a slower rate of cognitive decline with age in some studies, although not all researchers support this view. ³⁵ In a previous study, a large‐scale dataset with over 2000 individuals provided no evidence supporting the hypothesis that higher education leads to slower rates of brain aging, as measured by volumetric changes in cortical and hippocampal structures using MRI. ³⁶ However, the premise underlying CR posits that it represents the capacity to sustain cognitive functioning in the presence of brain damage. In light of this, it is important to note that while education may not directly impact the brain as demonstrated by Nyberg et al., ³⁶ it could exert an influence on cognitive functioning when confronted with brain damage.

Other factors, such as the occupation level and leisure activities, have been associated with better aging outcomes and are thought to impact CR. A study involving 1347 older adults supported the importance of having an abundant early CR (measured by educational, occupational attainment) and late‐life leisure activity (measured by mental, physical, and social activities) in preventing cognitive impairment. ³⁷ , ³⁸ In a paper of Nelson et al., higher education and occupational position were used as CR measures. ³⁹ The authors considered higher education as an explanation for the associations between hippocampal volume and performance in executive control, total gray matter volume, and language, as well as memory. Similarly, the authors found that higher occupational position magnified the association between total gray matter volume and attention/working memory, language, and memory. These findings indicate that the association between brain volume and cognitive performance varies based on CR, with greater CR (higher education and occupational levels) associated with a stronger link between brain volume and cognition. General cognitive ability (GCA), considered a measure of “intelligence,” was positively associated with brain volume and cortical characteristics at various life stages, including young adulthood and older age. Higher GCA is related to lower risk of neurodegenerative diseases and plays a protective role in brain health. ⁴⁰

Musical and multilingual experiences have also been linked to healthy brain aging—especially when combined—suggesting that a broader spectrum of lifetime experiences contributes to CR. ⁴¹

Regarding neuropsychological assessments of CR, several questionnaires have been proposed ⁴² ; they measure heterogeneous components, reaching an approximate estimate of individual CR. Among all neuropsychological measures, the Cognitive Reserve Index questionnaire (CRIq), proposed by Nucci et al., ⁴³ assesses demographic data (date and place of birth, gender, nationality, marital status), followed by 20 items grouped into three different categories, such as education, working activity, and leisure time, in accordance with the hypothetical concept of CR. ⁴³ More precisely, the education section investigates years of education, plus the achievement of any other type of additional training. The occupational level section investigates five different levels of working activities (ie, unskilled or manual work, skilled manual work, skilled nonmanual or technical work, professional occupation, highly intellectual occupation); work activity is collected as the number of years in each profession over the lifespan. Ultimately, the leisure time section investigates intellectual, social, and physical activities carried out during leisure time; these data are also collected in terms of frequency, as well as the number of years of each activity. The questionnaire returns subscores for each domain, together with a total score.

In conclusion, all potential sociobehavioral proxies, influenced by environmental exposures, act together for developing CR. Nevertheless, these CR components should be considered dynamically and their involvement in CR might specifically take account of individual brain pathological changes. Further investigations should evaluate in more detail whether sociobehavioral proxies effectively represent a causative factor for dementia onset or not.

5.2. Neuroimaging and CR

According to recent literature, differences in neural recruitment and efficiency may be considered as possible mechanisms by which CR provides protection against pathological brain aging. Therefore, structural (ie, MRI) and functional neuroimaging techniques (ie, fMRI and PET) have been largely employed to investigate anatomical regions and neuronal networks activity (blood flow, energy consumption) that have been hypothesized to represent and sustain CR over time (see Table 2).

TABLE 2.

Summary of cited articles and outcomes on neuroimaging and cognitive reserve in physiological aging.

Author	Participants and study type	CR proxies	Neuroimaging	Outcomes
Valenzuela et al. (2008)	37 healthy controls Longitudinal study	Complex mental activity levels estimated by the Lifetime of Experiences Questionnaire (LEQ)	MRI	Hippocampal volume, whole brain volume (WBV) and white matter hyperintensities (WMHs) Higher LEQ scores experienced less hippocampal atrophy over the follow‐up period
Rovio et al. (2010)	31 healthy controls, 23 MCI and 21 demented Longitudinal study	Physical activity	MRI	Grey matter (GM) density and white matter lesions (WML) Persons who actively participated in physical activity at midlife tended to have larger total brain volume in late‐life than sedentary persons.
Bastin et al. (2012)	74 healthy older participants Cross‐sectional study	Education and verbal intelligence	FDG‐PET	Glucose metabolism Higher degree of education and verbal intelligence was associated with less metabolic activity in the right posterior temporoparietal cortex and the left anterior intraparietal sulcus.
Arenaza‐Urquijo et al. (2013)	36 healthy elders Cross‐sectional study	Years of education	MRI; FDG‐PET; rsfMRI	Gray matter volume; gray matter metabolism; functional connectivity Higher years of education were related to greater volume in the superior temporal gyrus, insula and anterior cingulate cortex and to greater metabolism in the anterior cingulate cortex. Education was positively related to the functional connectivity between the anterior cingulate cortex and the hippocampus.
Yoshizawa et al. (2014)	123 normal adults Cross‐sectional study	Years of education, age, and gender	18F‐FDG PET	Glucose metabolism Highly educated subjects revealed focal hypermetabolism in the right hemisphere and lower recruitment of glucose metabolism in memory tasks.
Vonk et al. (2022)	989 middle‐aged participants Longitudinal study	Education and Dutch National Adult Reading Test (DART)	MRI	Intracranial volume (ICV) and baseline brain parenchymal fraction (BPF) Higher DART, education was related to a slower rate of memory decline, particularly in late life.

Note: This table provides a comprehensive overview of relevant articles, encompassing participant demographics, study types, utilized cognitive reserve proxies, neuroimaging techniques employed, features investigated, and reported outcomes.

Abbreviations: CR, cognitive reserve; FDG, fluorodeoxyglucose; MCI, mild cognitive impairment; MRI, magnetic resonance imaging; rsfMRI, resting‐state functional MRI; PET, positron emission tomography.

Quantitative neuroimaging studies have shown that, as a whole, the human brain volume tends to shrink with age, especially in terms of gray matter rather than white matter measures. ⁴⁴ , ⁴⁵ Structural MRI techniques have been commonly been used to reveal whether CR proxies (ie, occupation, education, leisure activities, and head circumference as a BR proxy) may correlate to anatomical brain parameters (ie, whole brain size, white matter damage, and cortical atrophy). ⁴⁴ , ⁴⁶ , ⁴⁷ However, these findings in this field remain controversial, particularly due to the lack of follow‐up studies with repeated MRI and combined CR proxies in the same individuals.

In healthy elderly individuals, measures of CR parameters, such as education and intelligence, seem to be associated with greater brain volumes and have protective effects against progressive brain atrophy or white matter damage. However, these studies have limitations related to the lack of longitudinal data and potential additional features, such as genetic and cerebrovascular risk factors, which may remarkably impact the final measures. ⁴⁴

Instead of focusing solely on overall brain measures, other studies investigated a possible association between CR parameters and specific brain region structure. For instance, physical activity has been found to correlate with gray matter volume in the middle frontal gyrus, while high level of engagement in daily activities has been associated with reduced hippocampal atrophy. ⁴⁸ , ⁴⁹

A further study investigated the associations of CR and BR with memory decline in midlife and late life, following patients for 12 years. This study utilized proxies such as education and the Dutch National Adult Reading Test (NART) for CR, and intracranial volume and baseline brain parenchymal fraction for BR. The results showed that higher CR and BR were associated with a slower rate of memory decline, particularly in late life. ⁵⁰

Among all neuroimaging methods, fMRI remains the primary technique for investigating the neural implementation of CR in normal aging and thus exploring both NR and NC mechanisms. Task‐related fMRI has been utilized to identify possible correlations between specific cognitive demands and CR, whereas resting‐state fMRI (rsfMRI) is considered effective in evaluating intrinsic neural activity within brain networks related to CR. In elderly populations, task‐related fMRI and rsfMRI have demonstrated an inverse association between regional blood flow and CR components, suggesting that individuals with a greater CR may exhibit an increased neural efficiency. ⁵¹

Compared to task‐related fMRI, which directly relies on flow/metabolisms of specific brain regions involved in the tasks, rsfMRI is considered a more accurate and valid technique for studying neural networks correlated with CR. ¹⁰ Notably, rsfMRI has been used in functional connectivity studies, showing increased connectivity in individuals with higher CR. Additionally, years of education have been positively correlated with functional connectivity linking specific brain regions, such as the anterior cingulate cortex, hippocampus, inferior frontal lobe, posterior cingulate cortex, and angular gyrus. ⁵² However, it is important to note that while task‐related fMRI and rsfMRI are crucial, they only reflect brain activities that require changes in energy consumption, whereas neuronal assemblies utilize communication mechanisms that do not require such changes, including transient and rapid phase‐coherence synchronization of firing oscillations without frequency modifications, thus maintaining stable energy consumption. ⁵³ , ⁵⁴

Contrasting evidence has also been found also using PET imaging. In a previous study, possible differences in cerebral metabolism related to gender were evaluated in healthy subjects. Compared to male participants, females, as a whole, showed a greater cerebral metabolism, while males showed a glucose hypermetabolism in bilateral inferior temporal gyri and cerebellum. Interestingly, a positive correlation was found between higher educated participants and focal glucose hypermetabolism in right tempo‐parietal regions, suggesting that these areas might be involved in CR. ⁵⁵ Conversely, another study, involving healthy elderly individuals with a higher education level, observed lower glucose activity in posterior parietotemporal regions and left anterior intraparietal sulcus, which are considered potential substrates of CR. ⁵⁶

Previous studies have supported the hypothesis that neural processing identifies different patterns of brain network activation as a function of interindividual differences in CR, suggesting that variations in CR capacity might be an individual strategy to cope with physiological or pathological brain aging. ⁵⁷ , ⁵⁸

Despite several scientific efforts directed towards finding clearer and more valid approaches to better understand and anatomically identify CR using neuroimaging assessments, straightforward and definitive results have not yet been achieved.

5.3. Electromagnetic brain activity and CR

Neuroimaging methodologies cannot be systematically applied to large populations of healthy elderly individuals as part of the methods for measuring CR in a large sample due to logistic and economic challenges. Therefore, it might be important to study the neural mechanisms underlying CR using techniques that provide more direct measures of neural activity, such as EEG or MEG, which are widely available, low cost, and often wireless and portable for use in a more ecologically valid environment.

By evaluating the EEG recordings of 25 healthy young adults and 19 nondemented healthy older adults during a verbal recognition memory task, it was found that higher CR, as measured with the NART, the vocabulary subtest of Wechsler Adult Intelligence Scale (WAIS‐R), and years of education, was associated with smaller changes in event‐evoked brain potentials amplitude and a less prolongation in latency with increasing task difficulty compared to individuals with lower CR. These CR‐related differences might be viewed as markers of neural efficiency underlying reserve against neuropathology and age‐related burden. ⁵⁹

The study by Moussard et al. examined the relationship between music practice, considered as a potential factor influencing CR, and inhibitory control by assessing behavior and electrophysiological responses in a go/no‐go task in older musicians and an age‐matched group of nonmusicians. Differences were found, with musicians exhibiting larger amplitude responses and more efficient deployment of inhibitory control. Brain response amplitudes for no‐go trials were also correlated with task accuracy in musicians, and older musicians showed a more anterior distribution of the target‐response wave compared to nonmusicians. ⁶⁰ Another study examined interference control across the lifespan using the Stroop task and investigated the factors contributing to performance differences in older individuals, which may reflect CR. Older participants were divided into three groups based on performance, considering education, foreign language use, and intelligence quotient (IQ) as a proxy for CR. Results showed age‐related decline in performance, but high‐performing older individuals performed similarly to younger adults. These individuals exhibited increased cognitive resource engagement and processing efficiency, as indicated by brain potentials and responses, especially with more challenging tasks. The study suggests that proactive and reactive cognitive control modes are enhanced in high‐performing older individuals, and their preparatory efforts contribute to better performance by intensifying target‐related processing. ⁶¹

In a study by Fleck et al., ⁶² adults aged 45 to 64 were examined to assess the influence of CR level (measured by IQ and education) and age on resting‐state EEG coherence. The findings revealed differences in coherence between the left and right hemispheres and between eyes‐closed and eyes‐open conditions among CR groups. Younger participants with low CR exhibited higher EEG coherence compared to those with high CR, while older participants with high CR had higher EEG coherence compared to those with low CR. Moreover, older participants demonstrated greater right‐hemisphere coherence, while younger participants exhibited greater left‐hemisphere coherence. Increased right‐hemisphere connectivity was observed in elderly individuals with high CR, which could potentially indicate better cognitive performance associated with higher CR.

The influence of cognitive, social, and physical factors on functional brain connectivity and cognitive function in healthy adults and older adults has also been explored ⁶³ using resting‐state EEG recordings to measure cortical connectivity and differences in local and long‐range connectivity between cognitive, social, and physical groups. The results showed that participants with high CR exhibited higher long‐range connectivity between the occipital lobes and other cortical regions, which was associated with better performance on measures of spatial working memory and sustained attention.

In a study involving elderly participants, CR was assessed using education, occupation, leisure, and social activities. ⁶⁴ Resting‐state activity and the n‐back test were used to measure brain activity and working memory, respectively. High‐CR subjects demonstrated higher accuracy and faster reaction times compared to low‐CR subjects. Low‐CR subjects showed increased event‐related responses in the left occipital region, suggesting a greater utilization of available cognitive resources. High‐CR subjects exhibited a higher beta frequency peak in parietal and occipital regions during the task, while low‐CR subjects had a more prominent gamma frequency in the right temporal region during rest. Furthermore, high‐CR subjects had negative gamma frequency asymmetry between the occipital regions at rest, while low‐CR subjects displayed positive asymmetry. These findings were used to classify subjects as high‐ or low‐CR using a support‐vector machine classifier with an accuracy of 88.89%.

Further research aimed to evaluate the association between CR—measured in terms of verbal intelligence, the Cognitive Reserve Scale, Yale Physical Activity Survey, and cognition—and resting‐state EEG in healthy older adults. The study found that dynamic proxy measures of CR are positively associated with alpha power in resting‐state EEG and leisure activities are a significant positive predictor of alpha 2 rhythms. However, physical activities were not found to be predictors of cognitive performance or resting‐state EEG. ⁶⁵

The results by Buján et al. showed moderate negative correlations between the CRIq and occipital current source density in delta and beta 2 frequency bands, as well as positive associations between the CRIq and inter‐ and intrahemispheric lagged‐linear connectivity measures. ⁶⁶ The effect of age on cognitive performance was stronger for higher values of both measures, suggesting that lower values might protect or compensate for the effects of age on cognition. Table 3 summarizes the articles described in the present section.

TABLE 3.

Overview of cited articles and findings in the present paper on electromagnetic brain activity and cognitive reserve in physiological aging.

Author	Participants	CR proxies	EEG/MEG	Outcomes
Speer and Soldan (2015)	25 healthy young adults, 19 healthy older adults Cross‐sectional study	Composite of NART, vocabulary subtest of WAIS‐R and years of education	32‐channel EEG	ERP (P300) Individuals with higher CR showed smaller changes in P3b amplitude and less slowing in P3b latency (ie, smaller changes in the speed of neural processing) with increasing task difficulty
Moussard et al. (2016)	17 older musicians and 17 older non‐musicians Cross‐sectional study	Years of musical Practice	64‐channel EEG	ERP Older musicians showed larger N2 and P3 effects (“no‐go minus go” amplitude), with the N2 amplitude being correlated with behavioral accuracy for no‐go trials, and the topography of the P3 response was more anterior in musicians. Moreover, P3 amplitude was correlated with measures of musical experience in musicians.
Gajewski et al. (2020)	246 healthy participants grouped by age (young, middle‐aged, and elderly). Cross‐sectional study	IQ, education level, use of foreign language	32‐channel EEG	ERP High performance was associated with higher level of education, usage of foreign languages and higher IQ. The contingent negative variation (CNV) and the P2/N2 complex were larger in the old high than low performers and similar to middle‐aged or even young participants.
Fleck et al. (2017)	90 cognitively normal adults Cross‐sectional study	IQ and years of education	129‐channel EEG	Power spectral density, coherence Younger participants low in CR exhibited greater mean coherence than younger participants high in CR, whereas the opposite pattern was observed in older participants, with greater coherence in older participants high in CR.
Fleck et al. (2019)	104 healthy old adults Cross‐sectional study	CRIq	129‐channel EEG	Functional connectivity Participants high in social CR possessed greater local and long‐range connectivity in theta and low alpha for eyes‐open and eyes‐closed recording conditions. In contrast, participants high in cognitive CR exhibited greater eyes‐closed long‐range connectivity between the occipital lobe and other cortical regions in low alpha. Greater eyes‐closed local connectivity in delta was also present in men high in cognitive CR.
Yang et al. (2020)	41 healthy participants between the ages of 48 and 76 years old, Cross‐sectional study	CRIq	306‐channel MEG	Power spectral density, machine learning Subjects with a higher CR had a higher beta intensity in the parietal and occipital regions whereas subjects with a higher CR had a higher gamma intensity in the right temporal region in the resting state. Subjects with a higher CR had negative gamma asymmetry between the right and left occipital regions. These MEG results were used to classify subjects into high‐/low‐CR subjects and a mean accuracy of 88.89% was obtained.
Ferrari‐Diaz et al. (2022)	88 healthy older adults Cross‐sectional study	Verbal intelligence, Cognitive Reserve Scale, Yale Physical Activity Survey	19‐channel EEG	Power spectral density Leisure activities were a significant positive predictor of posterior alpha2 power.
Buján et al. (2022)	56 healthy older adults Cross‐sectional study	CRIq	64‐channel EEG	Source Reconstruction, Functional Connectivity Moderate negative correlations between the CRIq and occipital current source density in delta and beta 2 frequency bands, as well as positive associations between the CRIq and inter‐ and intrahemispheric lagged‐linear connectivity measures.

Note: This table offers a consolidated summary of referenced articles within this paper, providing key insights into participant characteristics, study outcomes, cognitive reserve proxies, and details on electromagnetic brain activity measurement, enhancing the understanding of their interrelation in the context of physiological aging.

Abbreviations: CR, cognitive reserve; CRIq, Cognitive Reserve Index questionnaire; EEG, electroencephalography; ERP, event‐related potential; IQ, intelligence quotient; MEG, magnetoencephalography; NART, National Adult Reading Test; WAIS‐R, Wechsler Adult Intelligence Scale.

6. PATHOLOGICAL BRAIN AGING AND CR

A wide range of factors, including genetics and lifestyle, may contribute to neurodegenerative mechanisms. According to the concept of CR, some individuals seem to be better equipped to deal with the effects of underlying neurodegenerative processes due to their lifelong experience/style, education, and other factors that enhance brain plasticity and resilience. Research has suggested that individuals with higher levels of CR tend to experience a slower decline in cognitive functions, maintaining their independence and quality of life for longer. This is because they present a greater capacity to compensate for changes in the brain, by relying on other cognitive and neural resources when necessary.

6.1. Sociobehavioral and neuropsychological proxies of CR

Back in 1994, Stern et al. ⁶⁷ indicated that individuals with higher level of education or occupational attainment showed reduced risk of developing dementia, such as AD. ⁶⁸ Further investigations confirmed these findings and added other possible proxies for CR, including IQ, late‐life leisure activities, engagement in social networks, and physical activity. ⁴⁴ , ⁶⁹

People with a socially active lifestyle, engaged in demanding and intellectual daily life activities, seem to effectively face cognitive impairment. ⁷⁰ Middle‐aged individuals with cortical and hippocampal atrophy, which are related to a higher AD risk, showed similar cognitive performances to those with lower AD risk when involved in complex activities and social dealings. ⁷¹ , ⁷² Additionally, living in rural versus urban areas has been described as a synergic feature with low education in the risk of developing AD. ⁷³ Furthermore, childhood intelligence, measured with a survey that used the Moray House Test (MHT) of mental ability, represents a reliable CR proxy since previous findings have highlighted its strong correlation with hippocampal volumes, as well as late‐onset dementia. ⁷⁴ A previous study reported that a high level of “general intelligence” in childhood exhibited a protective effect for late‐life cognitive decline, independent of the level of education, occupation, and healthy lifestyle. ⁷⁵ It is worth noting that all these sociobehavioral proxies tend to be associated with environmental factors contributing to the occurrence of cognitive decline. ²³

However, some evidence has shown that patients with dementia and a greater CR exhibit a faster progression of decline compared to those with lower CR. One possible explanation to this apparent paradox—as explained in a previous part of this review—is that individuals with cognitive impairment and higher CR tend to display symptoms only when the disease is in its advanced stages. A study by Serra et al. aimed to examine the effect of CR on the progression of AD in patients with aMCI over a 24‐month period. A modified questionnaire was used to evaluate different levels of CR, including education, occupation, and lifetime cognitive, social, and physical leisure activities. The study showed that CR did not affect cognitive performance in healthy older adults or those in advanced stages of AD neuropathology, but it can play a protective role during the early stages of dementia. The study found that CR was associated with a delay in the conversion of aMCI to AD, with patients with high CR converting to AD later than those with low CR. ⁷⁶ ,  

TABLE 4.

Overview of cited articles and findings on neuroimaging and cognitive reserve in pathological aging.

Author	Participants and Study type	CR proxies	Neuroimaging	Outcomes
Garibotto et al. (2008)	242 probable AD, 72 aMCI, and 144 healthy controls Longitudinal study	Education and occupation	FDG‐PET	Brain glucose metabolism (rCMRglc) Association between higher education/occupation and lower rCMRglc in posterior temporoparietal cortex and precuneus in pAD and aMCI converters
Solé‐Padullés et al. (2009)	16 healthy subjects, 12 aMCI, and 16 AD Cross‐sectional study	Premorbid IQ (WAIS vocabulary test), “education‐occupation,” and questionnaire of intellectual and social activities	Structural MRI; fMRI	Whole‐brain volumes; brain activity Higher CR was associated with reduced brain volumes in MCI and AD.
Bosch et al. (2010)	15 healthy elders, 15 aMCI and 15 AD Cross‐sectional study	WAIS‐III, “education‐occupation,” and other CR variables (ie, lifetime occupation and activities)	fMRI	Activation and deactivation brain areas Among aMCI and AD we observed positive correlations between CR measures and BOLD signals in task‐induced activation areas directly processing speech, as well as greater deactivations in regions of the DMN. These relationships were inverted in healthy elders.
Serra et al. (2011)	22 AD and 23 aMCI Longitudinal study	Education	3‐Tesla MRI	T1‐weighted volumes, using voxel‐based morphometry, for gray matter investigation GM volumes of Low Educational Level (LEL) compared with High Educational Level (HEL) patients were reduced in the supramarginal gyrus bilaterally and in the right posterior cingulate/precuneus and frontal opercular cortex. Conversely, HEL compared with LEL patients showed reduced GM volumes in the entorhinal cortices and temporal poles, regions typically affected by AD pathology.
Garibotto et al. (2012)	51 AD, 27 APOE‐ε4 carriers, and 24 noncarriers Cross‐sectional study	Education and occupation	FDG‐PET	Brain glucose metabolism Inverse correlation between educational/occupational level, and metabolism in the posterior cingulate cortex and precuneus in both ε4 carriers and noncarriers
Garibotto et al. (2013)	9 healthy controls, 7 early probable AD, and 9 MCI Longitudinal study	Education and occupation	PET imaging with [11C]‐MP4A	Acetyl‐cholinesterase (AChE) activity The analysis of prodromal and early AD showed positive correlations between education and AChE activity in the hippocampus, bilaterally, and between occupation and AChE activity in the right posterior cingulate gyrus.
Steffener et al. (2014)	39 healthy younger and 45 healthy older adults Cross‐sectional study	Cognitive scores (speed of processing, fluid reasoning, and memory) and lifetime exposures (education and verbal IQ measures)	MRI	Cortical thickness and subcortical volumes using T1‐weighted images Individuals with greater lifetime exposures correlate with MRI brain measures.
Bozzali et al. (2015)	11 AD, 18 MCI, and 16 healthy controls Cross‐sectional study	Composite score (years of education and type of school attended)	fMRI	Functional connectivity in default mode network (DMN) Education was found to modulate functional connectivity in the posterior cingulate cortex. This effect was highly significant in AD patients, less so in patients with MCI, and absent in healthy subjects.
Oh et al. (2016)	43 young and 62 cognitively normal older adults Cross‐sectional study	Neuropsychological assessment	fMRI; 18F‐florbetaben PET	Frontoparietal control (FPC) activity; amyloid positive or negative Age‐related increases in brain activity were found in FPC regions. For higher cognitive control load, however, Aβ+ elderly showed reduced task‐switching activation in the right inferior frontal cortex.
Premi et al. (2017)	108 presymptomatic MAPT, GRN, and C9orf72 mutation carriers and 123 non‐carriers Cross‐sectional study	Educational attainment and genetic pattern (TMEM106B polymorphism)	MRI	Cortical and subcortical grey matter volumes Education directly affected grey matter volume. TMEM106B genotype did not influence grey matter volume directly on its own but in mutation carriers, it modulated the slope of the correlation between education and grey matter volume
Serra et al. (2022)	110 AD, 104 aMCI due to AD, and 64 healthy subjects Cross‐sectional study	Years of education	3‐Tesla MRI	Cortical thickness (CTh) and fractal dimension (FD) aMCI with high CR compared to aMCI with low CR patients showed significant decrease of CTh in the right temporal and in the left prefrontal lobe. They showed increased FD in the right temporal and in the left temporo‐parietal lobes. Patients with AD‐high CR showed reduced CTh in several brain areas and reduced FD in the left temporal cortices when compared with AD‐low CR subjects.

Note: This table presents a summarized account of referenced articles, offering key insights into participant characteristics, study design, cognitive reserve proxies, and details on neuroimaging techniques and outcomes in the context of pathological aging.

Abbreviations: AD, Alzheimer's disease; aMCI, amnestic mild cognitive impairment; APOE‐ε4, apolipoprotein E gene, ε4 allele; CR, cognitive reserve; FDG, fluorodeoxyglucose; fMRI, functional magnetic resonance imaging; MCI, mild cognitive impairment; PET, positron emission tomography.

A recent method to measure CR, especially in disease, is the dynamic concept of reserve. These measures are sensitive to the cognitive changes (ie, variance) caused by disease and are typically conceptualized as the residual cognitive abilities (ie, memory, general cognitive efficiency, executive functions, etc.). These indexes are computed by assessing and combining data from patients at baseline and follow‐ups, tracking cognitive changes over time. ⁷⁷ Static and dynamic CR indexes are supposed to represent different aspects of reserve that do not operate in parallel but form a complex system in which crystallized cognitive abilities and actual cognitive efficiency interact with brain atrophy of memory‐related networks and relays. ⁷⁸

6.2. Neuroimaging and CR

Interindividual variability in CR may determine different responses to the occurrence of dementia, either by withstanding or compensating for it, leading to better cognitive performance and overall functioning. Therefore, gaining insight into the potential neural basis of CR can help to identify novel targeted therapies for preventing or slowing down cognitive decline. Neuroimaging techniques have extensively investigated brain sites and networks in individuals with neurodegenerative diseases, particularly AD (Table 4). As mentioned earlier, PET imaging studies with patients clinically matched for disease severity have reported a negative correlation between regional blood flow and CR measurements (based on education, occupation, or neuropsychological assessment) in highly educated patients with AD and aMCI (the stage prodromal to dementia), suggesting that higher CR may protect cognitive performance until advanced disease stages. ⁷⁹ , ⁸⁰ This correlation was also observed in both APOE‐ε4 allele carriers and noncarriers, where a negative association between CR indexes (including education and occupation levels) and metabolism in the posterior cingulate cortex and precuneus was documented. ⁷⁹ , ⁸¹ Moreover, a previous study reported a significant correlation between acetylcholine activity and CR proxies (education and occupation) in patients with MCI due to AD and early AD. Specifically, using carbon‐11‐labeled N‐methyl‐4‐piperidyl‐acetate ([11C]‐MP4A) PET imaging, a positive association was found in specific brain regions, such as the bilateral hippocampus and posterior cingulate gyrus, suggesting that cholinergic activity is typically preserved in the BR of individuals with AD. ⁸²

A structural MRI study reported that lower grey matter volumes in presymptomatic carriers of genetic mutations for frontotemporal dementia (FTD) were influenced by both genetic factors and CR proxies, including educational attainments. ⁸³ , ⁸⁴ Another study utilized voxel‐based morphometry to measure regional gray matter volume at a voxel scale in MRI to evaluate the effect of formal education on the brain tissue of patients with aMCI or fully developed AD. The groups were divided into high and low levels of formal education. The study found that the level of formal education did not have a significant impact on gross brain anatomy but was significantly related to local structural changes in brain tissue. Patients with high education had better preserved visuospatial abilities, including logical deductive reasoning, constructional praxis, and executive functions. At the same time, they have greater gray matter volumes in the bilateral supramarginal gyrus, right posterior cingulate gyrus, frontal operculum, and precuneus compared to low‐education patients. ⁸⁵ , ⁸⁶

Based on the current literature and PET imaging studies, fMRI evidence supports the hypothesis that temporal and default mode network (DMN) regions are related to NR mechanisms, while frontal areas and dorsal attentional networks are associated with NC. ⁸⁷ , ⁸⁸ , ⁸⁹ When comparing MCI and AD patients with healthy controls, an fMRI study found a specific correlation between higher CR (measured by WAIS‐III, education, lifetime occupation and activities) and greater activity in different frontal brain regions, such as the superior and medial frontal gyrus, left inferior gyrus, and precentral gyrus. This relation was revealed during comprehension tasks only in MCI and AD patients, whereas it was inverted in healthy elders. ⁹⁰ Another study examining visual tasks observed similar activity in these frontal areas, as well as the superior parietal lobule and superior temporal gyrus, which demonstrated greater involvement in AD and higher CR measured by premorbid IQ (WAIS vocabulary test), education, occupation, and a questionnaire on intellectual and social activities. ⁹¹ Regarding the DMN, higher CR scores, in terms of some proxies like WAIS‐III, education, lifetime occupation and activities, seemed to correlate with reduced DMN activity during a task‐related fMRI in MCI individuals compared to controls. ⁹⁰ However, an rsfMRI investigation showed a positive correlation between functional connectivity in the posterior cingulate cortex and education levels in AD patients, ⁸⁷ thus suggesting an initial process of NC.

Although longitudinal data are strongly needed, higher CR measures have consistently demonstrated a correlation with increased global and local functional connectivity, implying that CR may play a pivotal role in slowing down pathological processes in neurodegenerative diseases. Table 4 summarizes the articles described in the present section.

6.3. Electromagnetic brain activity and CR

Recently, there has been increasing attention given to the use of electrophysiological techniques for studying pathological brain aging. Several studies have analyzed dementia‐related diseases using EEG data analysis, including measures such as event‐related potentials (ERPs), power spectral density, and connectivity analysis. Some of the most studied clinical conditions in this context include states that are prodromal to dementia, namely subjective cognitive decline (SCD), MCI, and the pathological state of AD (Table 5). In a study by Gu et al., 39 individuals with aMCI and 46 controls underwent EEG recording during n‐back tasks. The study assessed CR using measures such as the CRIq and WAIS‐R, as well as ERP analysis. From the behavioral point of view, compared to controls, aMCI showed reduced accuracy and delayed response time. Higher CR improved neural efficiency, as measured by task performance and P300 amplitude/latency characteristics. Specifically, in aMCI participants with higher CR, attention processes were modulated, resulting in better task performance. ⁹²

TABLE 5.

Summary of cited articles and outcomes on neuroimaging and cognitive reserve in pathological aging.

Author and Year	Participants	CR proxies	EEG/MEG	Outcomes
Gu et al (2018)	85 subjects (39 aMCI patients and 46 their matched controls) Cross‐sectional study	Composite of CRIq subscores and VIQ as measured by WAIS‐RC	64‐channels EEG	ERP (P300) Higher CR reduced neural inefficiency, which might be associated with better task performance in HC. However, no correlation was detected between CR and neural inefficiency in aMCI patients
Babiloni et al. (2020)	118 elderlies with SMC and amyloid negative and 54 amyloid positive Cross‐sectional study	Education level	256‐channels EEG	Power Spectral Density Amyloid PET‐positive SMC (SMCpos) participants with high (over low‐moderate) education level showed higher temporal alpha 3 power density (possibly “neuroprotective”) and lower posterior alpha 2 power density (possibly “compensatory”).
Babiloni et al. (2021)	60 elderlies and 70 MCI‐AD Cross‐sectional study	Educational level	19‐channels EEG	Power Spectral Density As compared to the Nold‐Edu‐ subgroup, the Nold‐Edu+ subgroup showed greater alpha source activations topographically widespread. The ADMCI‐Edu+ subgroup displayed lower alpha source activations topographically widespread.
Griffa et al. (2021)	35 cognitively normal and 11 cognitively impaired Cross‐sectional study	Leisure and cognitively stimulating activities	306‐channels MEG	Power Spectral Density higher relative power in the alpha band was associated with higher CR
Devos et al. (2023)	29 older adults with cognitive impairment and 19 with normal cognition Cross‐sectional study	CRIq	256‐channels EEG	ERP (P300) A higher CRIq score was associated with larger event‐related potentials and a higher cognitive activity index, as measured by pupillary size

Note: This table provides a consolidated overview of referenced articles, offering crucial insights into participant characteristics, study outcomes, cognitive reserve proxies, neuroimaging techniques and outcomes obtained within the context of pathological aging.

Abbreviations: AD, Alzheimer's disease; aMCI, amnestic mild cognitive impairment; CR, cognitive reserve; CRIq, Cognitive Reserve Index questionnaire; EEG, electroencephalography; ERP, event‐related potential; MCI, mild cognitive impairment; MEG, magnetoencephalography; PET, positron emission tomography; SMC, subjective memory complaint; VIQ, verbal intelligence quotient; WAIS‐R, Wechsler Adult Intelligence Scale.

Another EEG study by Babiloni et al. examined elderly subjects with SCD and found a significant relationship with estimated CR. The study revealed that individuals with positive amyloid status and higher estimated CR (measured by education and occupation) exhibited higher temporal alpha amplitudes and lower posterior alpha amplitudes compared to individuals with lower estimated CR. The authors suggested that these findings indicate probable compensatory and neuroprotective mechanisms occurring in individuals with positive amyloid status and high CR, where additional neural resources are recruited. ⁹³ A research study investigated the effect of high CR on brain neurophysiology in normal elderly individuals and those with MCI, prodromal to AD. The study analyzed EEG data from 60 normal elderly and 70 MCI‐AD, stratified into higher and lower education attainment groups, like a proxy to measure CR. The results showed that the normal elderly individuals with high CR had greater alpha EEG activity compared to those with low CR. Conversely, MCI‐AD participants with high CR had lower alpha EEG activity compared to those with low CR. The authors suggested that high CR may be associated with changes in resting state EEG alpha rhythms and may offer neuroprotective benefits in normal elderly individuals and functional compensation in AD patients. ⁹⁴

In a MEG study by Griffa et al., the neural basis of cognitive functioning and CR was investigated in 35 cognitively normal and 11 cognitively impaired participants. CR was estimated through a self‐reported scale on engagement in cognitive activities. The study found that higher relative power in the alpha band was associated with higher CR. ⁹⁵ Another recent EEG study involved 29 older adults with MCI and 19 with normal cognition who performed an n‐back task at different difficulty levels. The results showed that the MCI group had lower accuracy and slower reaction times. Additionally, both groups showed that a higher CRIq score was associated with larger event‐related potentials and a higher cognitive activity index, as measured by pupillary size. The amplitude of the ERP was lower in the 0‐back and 1‐back tests compared to the 2‐back test, but only in the group with normal cognition. Furthermore, the impaired group exhibited greater ERP amplitude than the control group in the low‐demand task, suggesting lower efficiency in the impaired group. In the high‐demand task, the impaired group showed lower ERP amplitude, potentially indicating that they had reached their maximal neural capacity. ⁹⁶

One short‐term longitudinal study is investigating the effects of foreign language training, music training, and social training on cognitive functions and mental health in seniors with cognitive decline and depression. This ongoing study is a clinical trial and as of now, no specific results or hypotheses have been formulated, but it is important to highlight the need of longitudinal research in the future of CR studies. ⁴¹ Table 5 summarize the articles described in the present section.

7. NETWORK ANALYSIS: A POSSIBLE APPROACH TO EXPLORE CR?

Network analysis could emerge as a promising approach for delving into the intricate dynamics of CR. This type of analysis can allow the identification of key nodes within the CR network, quantifying the relationships and visualizing the complex interactions between various cognitive and neural components. Graph theory is a mathematical framework used to study the relationships between nodes and the characteristics of edges in a network system. ⁹⁷ , ⁹⁸ This framework has been widely applied to study the brain, where nodes correspond to brain regions or neuronal assemblies and edges represent the connections between them. ⁹⁹ Numerous studies have shown that graph metrics can effectively capture the characteristics of the complex brain network architecture, making them a valuable tool for investigating how the brain is organized during different cognitive processes, including those related to CR. ¹⁰⁰

The application of graph theory in analyzing brain networks using different sources of brain imaging, such as MRI/fMRI, EEG, and MEG, has greatly enriched theoretical findings in neuroscience, providing a better understanding of the complexity of cognitive processes. It holds promise as a potential avenue to study the underlying mechanisms of CR. Fischer et al. conducted an MRI study with a group of 43 cognitively healthy elderly individuals aged 60 to 85. They utilized diffusion‐tensor imaging to evaluate the structural brain networks, focusing on global network properties (global efficiency, mean shortest path length, and clustering coefficient), and employed the WAIS‐R to measure general intelligence. The WAIS‐R was used as a proxy of CR, although it is not a direct enriching lifestyle factor that builds up CR, but it is known to attenuate the degree of age‐associated cognitive decline and has thus been proposed to be an important component of CR. The results indicated an association between global structural brain network properties and general intelligence in older individuals, suggesting that age‐related network deterioration may impact general intelligence. ¹⁰¹ The compensation of age‐associated brain structural changes through network architecture may serve as a surrogate for CR. Another study examined the APOE‐genotypes in relation to graph theory analysis and reported that the APOE‐ε4 noncarriers with higher education exhibited reduced local efficiency, clustering coefficient, normalized clustering coefficient, and small‐world architecture based on MRI data. This finding supports the hypothesis of the protective effects of education attainment against cognitive decline, especially in APOE‐ε4 noncarriers. ¹⁰²

Further research conducted by Weiler et al. demonstrated that education, as proxy of CR, along with abnormal levels of cerebrospinal fluid biomarkers (ie, Aβ1‐42, total tau protein, and phosphorylated tau), positively correlated with higher network efficiency based on fMRI findings. This indicates that AD patients with higher CR are able to mitigate the effects of pathology. ¹⁰³ In one study, differences in local measures of functional connectivity using graph theory parameters were observed in aMCI patients. Low CR (in terms of educational level) patients showed a loss of nodal degree, whereas high CR patients showed a loss of nodal efficiency in several brain regions, including the inferior frontal gyrus, the anterior cingulate cortex bilaterally, and the left postcentral gyrus. These findings suggest that higher levels of CR may contribute to increased resilience to pathological damage in the early stages of dementia. ¹⁰⁴

One study utilized EEG to measure functional connectivity and investigated the role of CR in brain network topology and dynamics during a memory task. ¹⁰⁵ Participants were classified into high‐ and low‐CR groups based on their CRIq. The study described differences in network architecture, specifically at the microscale (node) level, by analyzing the “eigenvector centrality” metric, which measures the importance of a node in a network based on its connections to other important nodes. The low‐CR group exhibited higher eigenvector centrality in temporal and occipital areas, while the high‐CR group had higher eigenvector centrality values in central areas. The high‐CR group also showed higher participation coefficient values, particularly in frontal regions. The participation reflects the diversity of connections a node has within a network. It indicates how evenly connections are distributed among different communities of nodes within a network. The participation coefficient of a node is calculated by dividing the number of connections it has to other nodes in different communities, by the total number of connections it has. Nodes with high participation coefficients are considered to play a more integrative role in the network, connecting different communities and promoting communication between them. To summarize, this demonstrated that individuals with high CR levels, particularly in frontal regions, had higher participation coefficients, indicating a more evenly distributed network of connections across different brain communities. Conversely, the low‐CR group had higher eigenvector centralities in temporal and occipital areas, suggesting that individuals with low CR levels had brain regions that held more central positions in the network and exerted greater influence over neighboring nodes. ¹⁰⁵

In conclusion, graph metrics can effectively capture the characteristics of the complex brain network architecture, making them a valuable tool for investigating the balance between integration and segregation in brain networks for efficient cognitive processing.

7.1. Final discussion and conclusions

This review demonstrates that the concept of reserve has widely been used to explain the variability in outcomes related to brain physiological and pathological aging. Exploring the mechanisms underpinning CR can potentially lead to intervention strategies aimed at slowing down or preventing cognitive decline. In fact, research evidence strongly supports the hypothesis that individuals with greater CR have a better resilience and capacity to deal with age‐related neurodegenerative brain changes. This suggests a positive impact of this model on the clinical trajectory of the disease. ²⁶ , ³⁷ , ¹⁰³ However, the precise mechanisms underlying this phenomenon remain still unclear. Over the last decades, several studies have evaluated possible mechanisms involved in these processes without reaching generalizable results.

Up to date, it is postulated that individuals with substantial CR and BR when impacted by neurodegenerative pathological changes may encounter a delayed onset of dementia and experience a slower disease progression. Cognitive symptoms may emerge later in life, leading to a more gradual decline compared to individuals with a lower reserve. Our considerations are addressed to two other different possibilities. On the one hand, the question arises when considering individuals with elevated CR but low BR: Which factor takes precedence? In such cases, it is conceivable that they may face a delayed onset of neurodegenerative diseases due to the protective effects of CR. However, once symptoms manifest, the limited BR may hinder the structural capacity to compensate for damage. This scenario could result in a rapid cognitive decline despite the delayed onset, as the brain grapples with pathological changes. On the other hand, the debate is still open on what happens when both CR and BR are low. Individuals in this category may be more susceptible to an early onset of neurodegenerative diseases, because a low CR suggests a lack of capacity for compensatory mechanisms. Additionally, low BR implies limited structural resilience against pathological changes. Therefore, this dual deficiency may contribute to an early and rapidly progressing course of the disease. Adding a perspective on the static nature of BR and the dynamicity of CR, it is crucial to note that BR tends to remain relatively stable throughout an individual's life. In contrast, CR exhibits a dynamic quality, influenced by ongoing cognitive stimulation, education, and engagement in mentally stimulating activities. Static and dynamic CR indexes are proposed to denote distinct facets of reserve that do not function independently but, much like interconnected elements within a matrix, intricately contribute to the overall functioning of the reserve system. The static CR indexes could represent one facet, like the rows or columns of a matrix, while the dynamic CR indexes could represent another facet, forming a comprehensive and interrelated network. Together, these components create a complex system where each element, like cells in a matrix, plays a crucial role in the overall resilience and individual functioning of CR. Thus, within this system, crystallized cognitive abilities and current cognitive efficiency interact dynamically with brain atrophy in memory‐related networks and relays. This interplay emphasizes the multifaceted nature of CR, encompassing both stable, well‐established factors and dynamic, adaptive components. Together, they contribute synergistically to an individual's cognitive well‐being and resilience against neurodegenerative processes. This dynamic nature suggests that interventions targeting CR can be implemented at various stages of life, offering opportunities for enhancement even in later years.

However, demonstrating the hypotheses surrounding the interplay of CR and BR in the context of neurodegenerative diseases involves a complex and multidimensional approach, based on the application of different diagnostic tools (eg, sociobehavioral, neuropsychological, genetical, anatomical, and functional connectivity measurements). Through comprehensive neuropsychological assessments, the inclusion of advanced neuroimaging tools—such as structural and functional MRI—and accessible, noninvasive techniques with high temporal resolution, like EEG, enhances the depth of insights into the structural integrity and functional connectivity of the brain. Moreover, investigations into genetic factors that influence cognitive and brain reserve, combined with analyses of lifestyle factors, education, and cognitive engagement, can contribute to a more holistic understanding. It is undeniable that when considering each of these proxies individually, each of them presents inherent limitations. For instance, sociobehavioral proxies often rely on self‐reported data, introducing the potential for biases. Individuals may not accurately recall or report their cognitive activities, education, or engagement in intellectually stimulating pursuits. Moreover, the interpretation of sociobehavioral factors is subject to the influence of cultural and social contexts. The definition of what constitutes a mentally stimulating activity can vary across populations, thereby impacting the generalizability of findings. Additionally, sociobehavioral proxies may fall short in capturing changes in cognitive engagement over time, as the frequency and nature of activities may fluctuate, and a single measurement might not fully encapsulate an individual's lifelong cognitive experiences. Similarly, neuropsychological assessments, while valuable, may predominantly concentrate on specific cognitive domains, potentially overlooking the broader spectrum of cognitive abilities contributing to CR. This limitation implies that such assessments may not entirely capture an individual's overall cognitive engagement and complexity. Furthermore, inherent variability in performance, stemming from factors like fatigue, mood, or external influences, could compromise the precision of cognitive measurements.

By integrating these diverse methodologies with structural and functional imaging and electrophysiological techniques, researchers can empirically validate and refine the proposed hypotheses, shedding light on the intricate dynamics of CR and BR in shaping the course of neurodegenerative diseases. Undoubtedly, the integration of diverse techniques presents a significant challenge, especially considering the array of methodologies involved. However, notwithstanding these complexities, it is imperative to invest in this research paradigm. Through a complete examination of protective factors, we not only have the chance to enhance our understanding but also to potentially develop new therapeutic (pharmacological and rehabilitative) strategies. In addition, longitudinal cohort studies, which meticulously track individuals with varying CR and BR levels over extended periods, constitute a foundational element of this investigative strategy. However, they have been historically underexplored in research, largely due to the inherent complexities of following individuals for an extended period. Another debated point seems to be related to study cohorts, which are not large enough to reach reliable and generalizable considerations.

A further critical issue is the lack of consensus regarding the terms and methodology used to study cognitive resilience and reserve. While the concepts of resilience and reserve have been widely utilized, there is a notable variation in their definitions and operationalization across different studies and scientific communities. This lack of consensus creates challenges in comparing and synthesizing findings, hindering the advancement of knowledge in this area.

Moreover, this lack of homogeneity hampers effective communication and knowledge dissemination. Without clear and agreed‐upon definitions, there is a risk of misinterpretation and miscommunication between researchers and clinicians. In recent years, substantial progress has been achieved in tackling this challenge, thanks to the Collaboratory on Research Definitions for Reserve and Resilience in Cognitive Aging and Dementia. This unified framework tries to comprehensively encompass the various dimensions of cognitive resilience and reserve, representing a crucial step forward in advancing research and translating findings into practical applications.

Moreover, although numerous studies have directly linked sociobehavioral proxies to CR, their protective role against cognitive decline needs to be still fully demonstrated. The extent to which these proxies truly capture the multifaceted nature of cognitive resilience and reserve remains uncertain. Relying solely on sociobehavioral proxies limits our understanding of the underlying brain mechanisms that contribute to cognitive resilience. To gain a more comprehensive understanding, it is necessary to integrate sociobehavioral proxies with objective measures of brain structure and function provided by advanced technologies.

In conclusion, this review has demonstrated the complexity of adequately measuring CR and the elusive nature of the concept itself. Each scientific community working on this topic has developed dedicated methods that are often technology‐dependent, creating a fragmented landscape where findings may not be easily comparable or generalizable. Harmonizing the field through an integrated and multidisciplinary approach is essential in order to establish consistent methodologies and facilitate collaboration between scientific and clinical communities.

However, there is a growing and general consensus that CR is one of the most important protective factors against neurodegenerative processes. Any intervention targeting lifestyle and modifiable factors that influence CR can significantly enhance brain resilience in the elderly and serve as a means to evaluate the effectiveness of pharmacological and nonpharmacological treatments. Additionally, the introduction of the concepts of static and dynamic CR, coupled with the exploration of network analysis as a promising research approach, lays the groundwork for a transformative shift in the landscape of CR and resilience studies. The potential distinctions between static and dynamic CR introduce a nuanced perspective that acknowledges the interplay of stable cognitive factors and adaptive mechanisms. Simultaneously, the integration of network analysis holds promise as a sophisticated method to unravel the complex interactions within cognitive processes.

The time has come for an integrated and multidisciplinary approach to defining and measuring CR, involving collaboration between scientific and clinical communities to harmonize the numerous data in this field.

AUTHOR CONTRIBUTIONS

Chiara Pappalettera: Conceptualization; Methodology; Writing—original draft preparation. Claudia Carrarini: Conceptualization; Writing—original draft preparation. Francesca Miraglia: Writing—reviewing and editing. Fabrizio Vecchio: Conceptualization; Methodology; Writing—reviewing and editing. Paolo M. Rossinik: Conceptualization; Methodology; Writing—original draft preparation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the supporting information.

Supporting information

Supporting Information

Pappalettera C, Carrarini C, Miraglia F, Vecchio F, Rossini PM. Cognitive resilience/reserve: Myth or reality? A review of definitions and measurement methods. Alzheimer's Dement. 2024;20:3567–3586. 10.1002/alz.13744