Dose–Response Matters! – A Perspective on the Exercise Prescription in Exercise–Cognition Research

1In the last decades, the average time that people are physically active has decreased dramatically in Western countries, and physical inactivity has been named as a big, if not even the biggest, health problem of the twenty-first century. Remarkably, physical inactivity is associated with impaired cognitive functions and higher risk of risk of neurodegenerative diseases (e.g., dementia) in the aging population. In order to counteract such negative effects of physical inactivity, an increase in the habitual physical activity level, which is typically engendered through a regular engagement in physical exercises, is empathically recommended. It has been well demonstrated in the literature that a single bout of physical exercises as well as repeated sessions of physical exercises in the form of a training intervention (e.g., endurance training) induce substantial neurocognitive changes. Based on such positive effects of physical exercises and/or physical training on brain plasticity and on cognition, it is not surprising that many scientific disciplines (e.g., medicine, psychology, neuroscience, and sport science) pay attention to this research field. Although these different scientific disciplines use different approaches to understand the relationship between physical exercises and/or physical training and the central nervous system, it is undoubted that all of them are based on an appropriate exercise prescription that specifies exercise (e.g., exercise intensity, exercise duration) and/or training variables (e.g., frequency of training sessions). Furthermore, exercise prescription is the key for dosing and for individualization of acute physical exercises and physical training. Individualization (personalizing) is an emerging approach aiming at maximizing the efficiency of an intervention by accounting for the interindividual heterogeneity in the response to acute physical exercises and/or physical training. Notably, what parameters are optimal to prescribe the best exercise for an individual is extensively discussed in the literature, but not all scientific disciplines investigating exercise–cognition are taking this issue into account sufficiently. Hence, the purpose of this article is to shed light on differences in exercise prescription and their relation to the dose and the interindividual heterogeneity in neurocognitive outcome measures.

2Prior to going more deeply into the topic of physical activity, physical exercise, and/or physical training, it is necessary to clarify these terms because they represent different concepts while it is, unfortunately, common behavior to use them interchangeably. “Physical activity” is defined as muscle-induced bodily movement that increases energy expenditure above ∼1.0/1.5 MET (metabolic equivalent of task; 1 MET = 1 kcal (4,184 kJ) × kg–1 × h–1). Hence, the term physical activity is a hypernym (i) that covers a wide range of physical activities that are conducted on a regular or unregular basis in a relatively unstructured and unplanned manner and (ii) that includes specific, planned, and structured forms of physical activities that are known as physical exercises. Physical exercises should be distinguished based on temporal characteristics into acute physical exercise (single bout) and chronic physical exercises (repeated bouts of acute exercises). A single bout of physical exercise is commonly referred to as an “acute (single) bout of physical exercise” or as “acute physical exercises”. Furthermore, chronic physical exercises can be denoted as “physical training” when they are conducted regularly in a planned, structured, and purposive manner with the objective to increase (or maintain) individual capabilities in one or multiple fitness dimensions. In essence, distinguishing and using these terms carefully allows a better classification and interpretation of observed effects and of the underlying (neuro)biological mechanisms.

3Since every human is unique, there is a considerable amount of within-individual (intra-individual) and between-individual (interindividual) heterogeneity in acute psychophysiological response(s) to the same acute physical exercises and/or long-term adaptions to the same physical training. Especially, the interindividual heterogeneity gained attention in the research of the recent years and is commonly observed in studies dealing with endurance (cardiovascular) training, resistance (strength) training, or combined training (consisting of endurance and resistance training). In order to account for this interindividual heterogeneity, the concept of (i) “responder” [also referred as “individuals with high sensitivity”] and (ii) “non-responder” [also referred as “individuals with low-sensitivity”, limited responders, or “individuals which did not respond”] was introduced, however, with varying definitions. While the definition and methods to classify responders and non-responders are currently discussed in the literature, it is relatively accepted that (i) not all outcome variables are affected equally by the responsiveness state (e.g., be a responder or non-responder), (ii) measurement errors are inevitable in repeated measurements and are caused, for instance, by random biological fluctuations that do not represent a meaningful change in the outcome variable, and (iii) some responses are likely to be transient, causing uncertainty regarding the time course of the responsiveness state. Hence, the following working definitions can be proposed (see Table 1). Regarding the response to acute physical exercises and/or physical training, (i) responders are individuals who exhibit, at a certain time point, changes in a variable of interest that are above (below) a distinct threshold, and (ii) non-responders are individuals who exhibit, at a certain time point, changes in a variable of interest that are below (above) a distinct threshold. There is ongoing vivid discussion on how to define these critical thresholds and whether further subgroups should be established. For instance, “adverse responders” or “negative responders”, have be defined as individuals who exhibit, at a certain time point, in response to acute physical exercise or physical training, unfavorable responses below (above) a distinct threshold. In addition, “above” and “below” need to be referenced relative to a specific outcome in the variable of interest. For instance, in a cognitive test, performance could be operationalized by “number of correct items” and “reaction time” (variables of interest). Regarding number of correct items, it is favorable to achieve a higher number of correct items (responder: above; non-responder: below). Regarding reaction time, on the other hand, it is favorable to react faster (responder: below; non-responder: above). Regardless of the ongoing discussion about how to classify the level of responsiveness, there is some evidence that the interindividual heterogeneity in response to acute physical exercise and/or physical training might contribute to the interindividual heterogeneity observed in neurocognitive outcomes. This evidence is outlined in the following section.

4In the following, we will refer to acute endurance exercises and endurance training because (i) from a neuroevolutionary view, endurance capacities (e.g., running during foraging) are important to ensure physical and/or neurocognitive well-functioning, (ii) acute endurance exercises and/or endurance training are currently in the focus of exercise–cognition research, (iii) endurance training induces substantial structural brain changes, and (iv) endurance training entails greater benefits in cognitive performance than resistance training.

5With regard to acute physical exercises, it was observed that individual baseline working memory function was linked with changes in working memory performance following acute very-light-to-moderate-intensity endurance exercises. Furthermore, responders, who showed improved cognitive performance after a single bout of very-light-to-moderate-intensity endurance exercise, exhibited a higher level of prefrontal activation during exercising. This finding suggests that not only peripheral systems are affected by the individual responsiveness state but also the central nervous system itself.

6Changes in cardiorespiratory fitness (CRF) measures in response to a 20 weeks endurance training program tended to vary tremendously among individuals. Exemplarily, observed, after a 21 weeks long combined exercise intervention in older adults, changes in CRF levels [assessed via maximal oxygen uptake (VO2 max.)] ranging from -8 to 42%. Such interindividual differences in response to endurance training also affect the expression of neurotrophic factors [e.g., brain-derived neurotrophic factor (BDNF)], which play an important role in neuroplasticity and cognition. For instance, after 6 weeks of high-intensity endurance training, responders who improved their CRF [assessed via peak oxygen uptake (VO2 peak)] to a greater extent than non-responders exhibited a significantly higher increase in serum BDNF. This finding suggests that the state of responsiveness influences important mechanisms involved in neuroplasticity and cognition. Notably, a higher level of CRF level [mostly operationalized by VO2 max. (highest value attainable by a subject) or VO2 peak (highest, “system-limited” value attained during the CRF test)] is associated with (i) better cognitive performance in younger adults, older adults, and older adults with mild cognitive impairments; (ii) favorable functional brain changes in younger adults and older adults; and (iii) favorable structural brain changes in older adults and individuals with Alzheimer’s disease. Furthermore, in response to endurance training, increases in VO2 max. (iv) mediate the improvement in cognitive functions in younger adults and (v) are associated with increases in hippocampal volumes in older adults.

7In sum, based on these associations between measures of CRF, measures of brain function and structure, and cognitive performance measures (albeit these correlations are not strictly causal in nature), it seems plausible to hypothesize that the large interindividual heterogeneity in measures of CRF may also contribute, among other factors, to the interindividual heterogeneity in measures of neurocognition in response to endurance training. However, to clarify the validity of these assumptions, further research is required.

8The interindividual responsiveness to physical exercises and/or physical training and, in turn, the interindividual heterogeneity in outcomes are caused by several moderators, including both non-modifiable factors (e.g., sex or genotypes) and modifiable factors (e.g., nutrition, social or cognitive activities, exercise prescription). Notably, these factors can also be classified as endogenous factors (factors attributable to the individual such as sex or genotypes) and exogenous factors (factors attributable to external inputs, e.g., generated by exercise prescription). Currently, the roles of non-modifiable (endogenous) factors such as sex and genotypes are investigated most. Among these factors, it has been shown that a considerable amount (approximately up to half of the variance) of the interindividual heterogeneity in physical outcomes, cognitive outcomes, and brain structure outcomes are explained by genetics. However, considering current evidence, lifestyle factors may equalize a “genetic handicap” since people with a high CRF level but “unfavorable” genetic polymorphisms do not need to perform significantly poorer than individuals with low CRF level but “favorable” genetic equipment. These findings suggest that a genetic handicap can be counteracted by other factors and that “overemphasizing” genetics for the individualization of exercise prescriptions is counter-productive. However, analysis of the genetics of participants is undoubtedly helpful in supporting individualization of acute physical exercise and/or physical training by aiding, for instance, the identification of potential responders and non-responders. Remarkably, it has also been highlighted that no “global non-responders” exist. Moreover, it is assumed that non-responsiveness can best be counteracted by modifying the dose of the physical exercise and/or physical training. The latter suggests that the dose of physical interventions per se contributes significantly to the observed interindividual heterogeneity in (neurocognitive) outcomes.

9The terminus dose is differently defined in the literature, but in exercise(-cognition) research, “dose” is commonly referred to as the product of exercise variables (e.g., exercise intensity, exercise duration, type of exercise; see Table 2) when considering an acute bout of physical exercises. In training studies, dose can be seen as the product of exercise variables (e.g., exercise intensity, exercise duration, type of exercise), training variables (e.g., frequency of training sessions), and the application of training principles. In reverse, dose could be modified in acute physical exercise studies by adjusting the exercise variables, while in physical training studies, exercise variables, training variables, and training principles must be taken into account (see Table 2). Such a purposeful modification is referred to as the adjustment of the exercise prescription.

10In the context of exercise prescription, it is also imperative to clarify the terms “external load” and “internal load.” While external load along with influencing factors (e.g., climatic conditions, equipment, ground condition) is defined as the work completed by the individual independent of internal characteristics, internal load is defined as individual and acute biomechanical, physiological, and/or psychological response(s) to the influencing factors and the work performed. According to the definition of internal load, which states that internal load is characterized by the individual and acute psychophysiological response(s) to the external load, it appears that internal load can be adjusted by modifying the external load. However, given that exercise variables such as exercise intensity can be operationalized using parameters of either external load (e.g., running with a speed of 10 km/h) or internal load (e.g., running with 60 of maximal heart rate), current definitions of dose are rather broad. Since dose is an essential factor for triggering neurobiological processes (e.g., release of neurotrophins such as BDNF; ), which in turn lead to neuroplastic and cognitive changes, it is crucial to agree on an appropriate concept of dose. Although markers of internal load could be more difficult to measure (compared to markers of external load), we suggest that dose should be operationalized by using a specific marker or specific markers of internal load as a proxy. The two reasons for this assumption are outlined in the following.

11Given (i) that internal load equals, per definition, the individual and acute psychophysiological response(s) to a given external load and (ii) that neurocognitive changes are triggered by such distinct psychophysiological responses, it seems reasonable to assume that internal load is a better proxy for dose than external load.

12There are several markers of internal load that can be used to prescribe the exercise intensity in acute endurance exercises and/or endurance training [e.g., oxygen uptake, heart rate, or heart rate variability (HRV)]. For instance, HRV, i.e., the beat-to-beat variation over a distinct time period, under rest conditions or while exercising is an interesting marker of internal load because the internal load quantification by HRV indices is progressive and takes the individual fitness level as well as daily readiness and actual health state into account. Furthermore, resting-state HRV is associated with cognitive performance.

13However, currently, several hypotheses exist in literature that explain the positive effects of acute physical exercises and physical training on brain plasticity and cognition. Among them, one of the most popular hypotheses is the “neurotrophic hypothesis,” which posits that in response to physical exercises, the organism releases several neurochemicals (e.g., neurotrophic factors such as BDNF), which in turn trigger neuroplasticity and facilitate cognitive enhancement. Hence, it seems more promising to use a marker or markers of internal load that are related to changes in neurotrophic molecules in order to individualize and adjust exercise prescription. In this regard, the peripheral level of blood lactate could be a promising marker of internal load because peripheral blood lactate (e.g., from muscles) can cross the blood–brain barrier and provides energy to the brain. Hence, it is not surprising that relative changes in peripheral levels of blood lactate are correlated significantly with cognitive performance levels after high-intensity interval endurance exercises. The crucial role of blood lactate for neuroplasticity is further emphasized by findings of peripheral blood lactate levels being associated with the peripheral serum BDNF levels. However, the exact molecular mechanisms of increased BDNF production in response to physical exercising are not fully understood (for review, see ). BDNF in the brain is involved in neuroplasticity, and serum levels of BDNF have been shown to be directly linked to cognitive performance after an acute bout of high-intensity endurance exercises. Moreover, (i) serum BDNF mediates improvements in cognitive functions following a 1-year aerobic endurance training, (ii) greater serum BDNF concentration changes in response to a 1-year-long aerobic endurance training are linked to hippocampal volume changes, and (iii) reduced levels of serum BDNF are related to a decline in hippocampal volume and poorer memory performance. In sum, a specific marker or specific markers of internal load such as the peripheral blood lactate level seems to constitute a promising proxy for dose. However, the optimal marker(s) that is (are), with regard to neuroplasticity and cognition, the most suitable proxy for the dose of physical exercise and/or physical training has yet to be discovered.

14Based on the large interindividual heterogeneity (i) in psychophysiological responses to acute physical exercises and (ii) in long-term adaptions to a physical training, it is assumed that tailoring of these to the characteristics and needs of a particular person is well suited to maximize their efficiency. Such an individualization of acute physical exercises and/or physical training could be achieved by adjusting the exercise prescription (e.g., exercise intensity), which influences, in turn, the dose (objectified by a specific marker or specific markers of internal load; see previous section and Figure 1A). In order to illustrate our thoughts in practical terms, we focus on exercise intensity because a full discussion of all exercise variables, training variables, and training principles is beyond the scope of this article. As outlined in the previous section, using markers of internal load to prescribe exercise intensity is preferable instead of using parameters of external load such as speed in running specific exercises. Therefore, traditional markers of internal load such as the fixed percentage of the maximally achievable value of oxygen uptake or heart rate are often used. Using a fixed percentage of a maximally achievable value of oxygen uptake or heart rate involves a considerable amount of interindividual heterogeneity in other markers of internal load (e.g., metabolic responses objectified by, for instance, peripheral blood lactate). Metabolic responses (e.g., peripheral blood lactate level) constitute specific markers of internal load that are likely to be proxies for the dose that triggers neuroplastic processes and cognitive changes (see “Why a specific marker of internal load is needed as a proxy for dose”). Hence, traditional exercise prescriptions lead to largely varying individual doses as revealed by the marker(s) of internal load. This may lead, among other factors, to the observed interindividual heterogeneity in neurocognitive outcomes (see Figure 1B). Consequently, approaches that ensure that a comparable dose is provided to each individual (e.g., adapted exercise prescriptions that ensure a comparable level of peripheral blood lactate) may lower the interindividual heterogeneity regarding neurocognitive outcomes. Hence, such approaches are favorable in exercise–cognition research (see Figure 1). In this context, individual threshold concepts (aerobic and anaerobic threshold) that are based on individual metabolic (or respiratory) responses could be used to determine an individual’s initial exercise intensity. However, while there is a strong theoretical basis for the application of a threshold-based exercise prescription for endurance exercises and endurance training, the challenges and pitfalls of determining such individual thresholds may explain why many researchers continue to favor exercise intensity prescriptions based on relative percentages of maximum values. Although our assumptions are well grounded on possible neurobiological mechanisms, they are mostly theoretical in nature, and thus, further research comparing, for instance, traditional versus adapted exercise prescriptions with regard to neuroplasticity and cognition is urgently needed.

15Since the level and detail of description required to extensively describe and discuss the influence of all exercise variables, training variables, training principles, and factors influencing exercise–cognition interaction go far beyond the scope and intent of this article, our assumptions still remain imperfect because other exercise-related factors such as movement frequency (e.g., cycling cadence) or psychological factors such as affective response (e.g., enjoyment or expectations) have not been considered. Nevertheless, given (i) that our knowledge of the dose–response relationship between acute physical exercises and/or physical training, neurobiological processes (e.g., neuroplasticity), and cognitive changes is still limited, (ii) that peripheral blood lactate levels constitute an established marker of internal load, and (iii) that peripheral blood lactate levels are easily quantifiable by portable devices, the use of peripheral blood lactate as a proxy for dose seems a reasonable starting point. Nevertheless, lactate monitoring suffers from the drawbacks that (i) it necessitates blood sampling, which could be impractical in daily practice, and (ii) it requires a graded exercise test to calculate an individual threshold to prescribe the exercise intensity. Regarding the first objection, new methods to non-invasively determine critical physiological thresholds (e.g., lactate threshold) by means of muscle near-infrared spectroscopy may constitute a more appropriate approach in daily practice, but this has yet to be investigated. With regard to the second objection, it is worth mentioning that graded exercise tests are relatively complex and time consuming and that exercise intensity could be more easily determined by using specific formulas (e.g., Karvonen formula to determine a target heart rate). However, a graded exercise test should be an integral part of the process of a proper exercise prescription because, currently, exercise intensity cannot be accurately predicted by specific formulas, and a fixed percentage of a maximally achievable value of heart rate leads to a considerable amount of interindividual heterogeneity in metabolic responses (e.g., blood lactate), which is deemed to contribute, at least partly, to the interindividual heterogeneity in neurocognitive outcomes (see previous sections).

16Still, even if peripheral blood lactate concentrations are associated with serum BDNF concentrations, further studies will be required to investigate the dose–response relationship between exercise prescription and (serum) BDNF levels. Since BDNF release is also influenced by several other non-modifiable (e.g., sex;; ) or non-exercise-related modifiable factors (e.g., sleep or nutrition;,;; ) that are known to influence neuroplasticity in general (e.g., sleep,;; or nutrition,;; ), these factors should be carefully monitored in further studies.

17In addition, with regard to the optimal dose, it could be useful to gather markers of internal load that are directly related to the state of the central nervous system itself (e.g., brain activity during exercise) because differences in brain activity (e.g., measured by functional near-infrared spectroscopy) (i) allow distinguishing between responders and non-responders, (ii) are sensitive to changes of exercise variables (e.g., exercise intensity), (iii) are sensitive to demands posed by the cognitive task or the motor task, (iv) and are associated with performance improvements in motor(–cognitive) tasks. Hence, markers of internal load assessing activation of the central nervous system may serve to quantify “complexity” (defined as neurocognitive demands posed by the exercise), which is an important variable with regard to neurocognitive changes in response to acute physical exercises and physical training, too. However, while measuring brain activation during exercise offers great potential to understand exercise–cognition interaction in general and interindividual variability in particular, future research in this area is strongly needed before measures of brain activity can be used to guide exercise prescription.

18Furthermore, we wish to stress that a traditional individualization of exercise prescription is perhaps necessary to answer basic research questions (e.g., Are the peripheral blood lactate release and changes in neurocognition a function of exercise intensity?) but that the individualization using an adapted exercise prescription may lead to further insights into exercise–cognition research (e.g., How to adapt exercise intensity to achieve a comparable change in the release of peripheral blood lactate across individuals and how this affects neurocognition?).

19In essence, this article aimed at providing a suggestion for a clearer definition of the dose in exercise–cognition research and presenting evidence in how interindividual variability in the dose might contribute to the interindividual heterogeneity in neurocognitive outcomes. We propose that the dose of an acute bout of physical exercises and/or physical training should be operationalized by a specific marker (or specific markers) of internal load. Modifying the exercise prescription by carefully adjusting the external load, a comparable dose can be achieved across individuals (see Figures 1A,B). Research is strongly encouraged to investigate in the future whether an exercise prescription inducing a comparable dose may lower the interindividual heterogeneity considering outcome variables on different levels of analysis and on different aspects of the brain (; see Figure 1C). Finally, understanding how a comparable dose affects neurocognitive outcomes is an important step toward identifying what dose is optimal for achieving the greatest benefits with regard to neurocognitive outcomes in an individual.

20FH wrote and edited the manuscript. PM, TG, and NM reviewed and edited the drafted versions.

21The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.