State of the art and future directions for measuring event-related potentials during cycling exercise: a systematic review

Selection criteria

We chose to include studies involving adults only, as the effect of exercise on ERP in children may be different and would introduce bias into the results. For the same reason, we excluded studies that included pathological conditions. In addition, we chose to include only studies in which the mode of exercise was pedaling (on a cycloergometer or bicycle), in order to reduce variability in methods between studies, thus rendering the results more comparable.

Search equation

A first search was performed on Google Scholar in June 2023 with the following equation: “event-related potentials” AND (cycling OR pedaling OR biking). To avoid excluding potentially relevant articles, no filters or time limits were used. The search for studies had to be carried out using a broad, unfiltered search equation on Google Scholar, which offers articles that are less relevant overall, but more numerous. With PubMed, this equation yielded only one result, not relevant to this review. Web of Science did not identify any new articles. For this review, articles were selected and read by two reviewers (RRG, DG). Disagreements during this phase were resolved by consulting a third reviewer.

Extracted data and assessment of bias

The data extracted were: ERP study objective, task type (or ERP paradigm), EEG system used, study population data, exercise characteristics (duration, intensity, pedaling cadence), ERP and behavioral results. We chose to record the behavioral results (response time and error rate on the attentional task) because they can help in the analysis of the impact of cycling on ERPs. The studies mainly focused on a comparison of ERP amplitude or latency between pedaling and non-pedaling conditions. However, some studies focused on other comparisons, e.g., several exercise intensities (Bullock, Cecotti & Giesbrecht, 2015; Olson et al., 2016; Dodwell et al., 2021) or pedaling cadences (Akaiwa et al., 2022). Finally, two others took into account the effect of exercise duration (Grego et al., 2004; Olson et al., 2016).

Aerobic exercise intensity is an essential parameter to explore because it can influence the amplitude and latency components of ERPs after acute exercise (Gusatovic et al., 2022). In this review, we attempted to assess the impact of intensity.

Search results

The equation produced around 3,500 results, according to the search engine. Only the first 1,000 results (first 100 pages) were accessible. However, all the articles in this study were found in the first 500 results. Of the 1,000 results available on Google Scholar, 72 were selected by title. The abstracts were then read and 15 were eliminated because the measurements had been taken after; and not during exercise. Others did not concern ERP measurement, or the exercise used was not cycling. Nineteen abstracts were selected. Finally, full-text articles were explored. E-mails were sent to the authors of two studies to find out whether ERPs were actually measured during cycling. In the end, a total of 17 studies were retained. The remaining three studies included in the review (Zink et al., 2016; Scanlon et al., 2017; Robles et al., 2022) were not found during the search, as they do not contain the exact phrase “event-related potentials” in their full text. They were identified through previous searches with other equations or from references in the selected articles. We used the “PRISMA 2020 flow diagram for new systematic review which included searches of databases, registers and other sources” model (Page et al., 2020) as shown shown in Fig. 1.

Study objectives

Study features

EEG signal measurement and analysis

Conditions of use

Study results

Results were reported for various types of ERPs (P3, P2, etc.) or involved comparisons (e.g., between cycling and non-cycling, or another condition with cycling). Table 6 presents all the studies’ results about ERPs and behaviours (response time and accuracy).

Results based on a cycling compared to a resting condition

The results presented in this sub-section address the effect of cycling on ERPs and ERP characteristics, compared to the “resting” condition, i.e., a condition in which the ERP paradigm was performed without cycling. Dodwell et al. (2021) is the only study that did not observe the P3 wave, but the authors reported a greater amplitude of the posterior PCN ERP in the presence of a distractor.

Results regarding the P3 wave:

Amplitude:

Four studies reported a greater amplitude of the P3 wave (Schmidt-Kassow et al., 2013; Olson et al., 2016; Schmidt-Kassow, Thöne & Kaiser, 2019) in the cycling condition compared to the resting condition.

Six studies reported the contrary, namely a lower amplitude of the P3 wave (Zink et al., 2016; Conradi et al., 2016; Scanlon et al., 2019; Olson, Cleveland & Materia, 2023; Akaiwa et al., 2022; Yagi et al., 1999). Specifically, for Zink et al. (2016) and Scanlon et al. (2019), this change was observed in a free cycling condition outdoors, on a bicycle. In Conradi et al. (2016) this result was obtained bearing in mind that in one cycling condition, the auditory stimuli were automatically synchronized to the subjects’ spontaneous pedaling cadence.

In six studies (Killane, Browett & Reilly, 2013; Vogt et al., 2015; Bullock, Cecotti & Giesbrecht, 2015; Torbeyns et al., 2016; Scanlon et al., 2017; Kuziek et al., 2018), there was no significant difference in the amplitude of P3.

P3 latency time:

Two studies reported reduced latency in the cycling condition compared to the resting condition (Yagi et al., 1999; Bullock, Cecotti & Giesbrecht, 2015). In Bullock, Cecotti & Giesbrecht (2015), the P3 wave was observed as a function of the P3a and P3b components, with P3a corresponding to the P3 wave for distractors (right-facing faces) and P3b corresponding to the P3 wave for targets (left-facing faces). A lower latency of P3a was observed.

One study reported longer latency (Pontifex & Hillman, 2007)

Five studies report no significant difference in latency (Killane, Browett & Reilly, 2013; Schmidt-Kassow et al., 2013; Zink et al., 2016; Olson, Cleveland & Materia, 2023; Akaiwa et al., 2022).

Other ERPs:

Some studies observed P1, N1, P2 or N2 waves, which are earlier than P3 and more closely linked to sensory processing. In Bullock, Cecotti & Giesbrecht (2015), a greater amplitude of the P1 wave was reported. In Zink et al. (2016), no change was reported for N1, while a larger amplitude was observed in Scanlon et al. (2019), and a smaller amplitude in Pontifex & Hillman (2007). In Olson et al. (2016), a greater amplitude of the N2 wave was reported (for both intensities compared to rest). Pontifex & Hillman (2007) found a lower amplitude. In Vogt et al. (2015) and Torbeyns et al. (2016), there was no change in N2. In Scanlon et al. (2019) and Pontifex & Hillman (2007), a greater amplitude of P2 was reported.

The three studies (Scanlon et al., 2017, 2019; Kuziek et al., 2018) that evaluated the MMN (mismatch negativity) response found no significant effect of exercise.

Response time

Five studies reported a reduction in response time (Yagi et al., 1999; Bullock, Cecotti & Giesbrecht, 2015; Olson et al., 2016; Dodwell et al., 2021; Robles et al., 2022) of which two (Olson et al., 2016; Bullock, Cecotti & Giesbrecht, 2015) compared only the higher intensity condition to the resting condition.

Six studies (Pontifex & Hillman, 2007; Vogt et al., 2015; Olson et al., 2016; Scanlon et al., 2020; Olson, Cleveland & Materia, 2023; Bullock, Cecotti & Giesbrecht, 2015) reported no significant differences of which two (Olson et al., 2016; Bullock, Cecotti & Giesbrecht, 2015) compared only the lower intensity condition to the resting condition.

Five studies reported an increase in the error rate with low cycling intensity (Yagi et al., 1999; Pontifex & Hillman, 2007; Olson et al., 2016; Zink et al., 2016; Akaiwa et al., 2022). In particular, in Akaiwa et al. (2022), a loss of accuracy was observed for the slow pedaling cadence condition.

Six studies reported no significant difference (Vogt et al., 2015; Bullock, Cecotti & Giesbrecht, 2015; Torbeyns et al., 2016; Akaiwa et al., 2022; Dodwell et al., 2021; Olson, Cleveland & Materia, 2023).

Comparison between two intensity levels

Three studies tested the effect of several levels of cycling intensity, comparing not only cycling conditions with a resting condition, as mentioned above, but also two cycling conditions of different intensities (Bullock, Cecotti & Giesbrecht, 2015; Olson et al., 2016; Dodwell et al., 2021).

In the two studies that observed the P3 wave (Bullock, Cecotti & Giesbrecht, 2015; Olson, Cleveland & Materia, 2023), no differences in amplitude or latency were reported between the two intensity levels. On the other hand, Bullock, Cecotti & Giesbrecht (2015) reported higher latency of the N1 wave in the high intensity exercise condition compared to the low intensity condition, and a tendency towards a lower amplitude. There was also no difference for the N2 wave in Olson et al. (2016). In Dodwell et al. (2021), the presence of a distractor reduced the amplitude of the posterior PCN wave at high or resting intensity, but not at low intensity.

With regard to response times, a significant decrease from low to high intensity was observed for Bullock, Cecotti & Giesbrecht (2015), Olson et al. (2016) and only a trend for Dodwell et al. (2021). There was no significant difference in the error rate between the two intensities across the three studies.

Results for characteristics other than exercise intensity

We present below the results of comparisons other than vs the resting condition.

The effect of duration was tested in Grego et al. (2004) and Olson et al. (2016) by performing a series of oddball paradigms at several timepoints during moderately intense cycling (60% and 66% VO₂max respectively). In addition, the effect of different cycling frequencies was tested (Akaiwa et al., 2022). Finally, the studies by Scanlon et al. (2020), Robles et al. (2022) tested the effect of a more or less noisy environment on attention-related ERPs during cycling.

Effect of duration

The study by Grego et al. (2004) compared P3 ERPs before and after exercise as well as during exercise at different intervals (at 3, 36, 72, 108 and 144 min), but did not compare ERPs during cycling with a resting condition. The aim was to study the effects on athletes during a 3-h exercise session. An increase in P3 amplitude was observed between the first and third hour (measurements at 72 and 108 min). In addition, an increase in P3 latency was observed during the third hour (measurements at 108 and 144 min). The study by Olson et al. (2016) also tested a potential effect of exercise duration by performing series of ERPs at 5, 15 and 25 min of exercise. A decrease in the amplitude of the P3 wave was observed on each series.

Effect of cadence

In Akaiwa et al. (2022), results were presented for various cadences (optimal, 30% faster and 30% slower). On the Pz electrode, the amplitude of the P3 wave was lower in the slower and faster cycling conditions compared with the optimal cadence.

Effect of synchronizing the bicycle with periodic sounds

The studies by Schmidt-Kassow et al. (2013), Schmidt-Kassow, Thöne & Kaiser (2019) reported that synchronizing the cadence with the sound rhythms of the oddball task (at 60 Hz) resulted in a greater amplitude of the P3 than in the non-cycling condition. In Conradi et al. (2016), sounds were synchronized with the subject’s rhythm (passive synchronization, not active as in the other two studies) and this was linked to a lower amplitude of the P3 compared with the non-cycling condition.

The effect of a noisy or agitated environment during cycling

In the studies by Scanlon et al. (2020) and Robles et al. (2022), which compared several cycling conditions, in more or less calm environments and without a static condition, no difference in the amplitude of P3 was found. Conversely, they found an increase in the N1 wave when moving from a quiet environment (outdoors with little traffic) to a noisier one (close to road traffic). Modulation of the P2 wave was only observed in Robles et al. (2022) with a decrease in amplitude between low and intermediate traffic conditions.

Feasibility: it is possible to measure ERP while cycling

All the studies in this review show that it is possible to reliably measure ERPs during exercise. The diversity of exercise conditions in this review shows that it is possible to study ERPs under conditions that might reasonably be expected to be difficult, such as 3 h of moderately vigorous exercise (66% VO₂max) (Grego et al., 2004). The studies by Scanlon and Robles found that noise levels in the external environment had an effect on ERPs N1 and P2 during cycling. The lower amplitude of N1 in a noisier environment and the higher amplitude of P2 up to a noise threshold (not precisely determined) makes it possible to confirm hypotheses about the sensory processing functions of these ERPs and to show that it is possible to measure these changes reliably during outdoor cycling.

Half of the studies used portable EEGs with different electrode systems or types (active, wet, gel). It is also possible to use a portable EEG with a 64-electrode configuration, as in the study by Vogt et al. (2015). The use of many electrodes is not necessary to measure and analyze the P3 wave or other ERPs, but it could enable the use of source localization methods. Although specific sensors have often been used to detect ocular artifacts, EEG electrodes close to the eyes would enable ocular artifacts to be detected and managed well. Vogt et al. (2015) used the PO9 EEG electrode to detect lateral eye movements.

As far as the methods to combat artifacts are concerned, some studies used algorithms such as ICA, but their use is generally limited to basic EEG processing methods (frequency filtering, amplitude rejection threshold). Moreover, studies of more intense exercise conditions do not appear to have used more advanced methods. According to a recent review (Sadiya, Alhanai & Ghassemi, 2021), hybrid methods combining several methods such as ICA and deep learning have recently been created and could contribute to progress in this field.

Results analysis

The two studies that used a Flanker task reported an increase in P3 wave amplitude (Pontifex & Hillman, 2007; Olson et al., 2016). However, more specifically, this result was the same for the lower (40% VO₂max) and higher (60% VO₂max) intensity conditions in Olson et al. (2016). Perhaps the Flanker and oddball tasks are linked to a different effect of exercise on the P3 wave. However, a different effect of the Flanker task compared with the oddball task does not appear in the results of the review by Gusatovic et al. (2022). Indeed, in the review by Gusatovic et al. (2022), 7 out of 16 studies (44%) using the Flanker task reported an increase in amplitude compared with 5/10 studies using an oddball task (auditory or visual). There was no clear trend towards increased P3 amplitude in studies using the Flanker task.

The low amplitude of the P3 wave and the lack of a significant difference between exercise and rest may be linked to low-intensity exercise. Indeed, low exercise intensities may not be sufficient to induce neurophysiological adaptation. Exercise has to be sufficiently intense to have an effect on the P3 wave, as suggested by the theories of Dietrich (2006), Dietrich & Audiffren (2011) and Hebb (1955) discussed below.

The reduction in response time and the increase in error rate seem to be linked to higher intensity. Response time never increases (in terms of significant difference) with exercise, but is sometimes reduced. Error rates are never lower during exercise, but sometimes higher than at rest. These results show that aerobic exercise can lead to a reduction in response time and an increase in error rate in an attentional task such as the oddball or Flanker tasks.

Hypotheses or theories to explain results

All studies except (Vogt et al., 2015) used an attentional task. Among them, twelve used an auditory oddball task and only three used a visual oddball (see Table 2 for the total number using each paradigm). Attention is a limited resource that is important during cycling activity, particularly to manage external risks (e.g., obstacles, road conditions). Observing the effect of exercise on the P3 wave could enhance our understanding of the effects of exercise on attention. All studies except (Dodwell et al., 2021) investigated the amplitude of the P3 wave. This ERP wave is involved in the attentional and cognitive processing of stimuli, and its amplitude is sensitive to the unexpected nature or relevance of stimuli. Some studies reported a decrease in the amplitude of the P3 wave during exercise compared with rest, while others reported an increase in P3 amplitude.

When a decrease in amplitude was observed, one hypothesis was the sharing of attentional resources in the brain, in order to perform several tasks simultaneously. This hypothesis has been proposed in studies such as Yagi et al. (1999), Scanlon et al. (2019), Akaiwa et al., (2022). The hypofrontality transfer hypothesis (transient hypofrontality) of Dietrich (2006) goes in this direction. It is based on the idea that attention requires metabolic resources whose availability is limited, and thus, motor and attentional processes operating in parallel are in competition. According to this theory, sufficiently strenuous exercise leads to a concentration of metabolic resources towards motor or exercise-relevant regions, to the detriment of regions less useful for exercise, such as the prefrontal regions. This could be the case for the parietal region for which the P3 wave was analyzed most frequently. According to this theory, executive control processes would benefit from simultaneous aerobic exercise up to a certain intensity or duration.

An increase in the amplitude of P3 in the exercise condition compared to the resting condition was observed in Pontifex & Hillman (2007) and Olson et al. (2016). In both studies, the accuracy of subjects’ responses to the Flanker task was poorer in the exercise condition than in the resting condition. According to Pontifex & Hillman (2007), the concomitance of aerobic bicycle exercise and the attentional Flanker task results in increased recruitment of neural resources or decreased inhibition. Similarly, it was suggested in Olson et al. (2016) that there is a strengthening of attentional resources. However, this contradicts Dietrich’s hypothesis. Furthermore, according to Pontifex & Hillman (2007), the loss of response accuracy is linked to a loss of efficiency of neuroelectrical resources, which does not seem to fit well with the idea of “increased recruitment of neural resources”.

In Dodwell et al. (2021), other hypotheses were put forward. In the condition with a distractor in the attentional task, a greater amplitude of the ERP PCN was observed in the “moderate” cycling condition (40–50% HR reserve) compared with the resting condition, and a tendency towards a lower amplitude in the “intense” cycling condition (60–70% HR reserve) compared with the moderate condition. The hypothesis was that the facilitation of attentional or cognitive abilities follows an inverted U-shaped curve (Hebb, 1955), with optimal facilitation for low or moderate intensity exercise, and no facilitation for intense or non-existent exercise. However, these results were not replicated in the other two studies involving two conditions of different intensity and a rest condition (Bullock, Cecotti, Cecotti, Cecotti and Cecotti). Bullock, Cecotti & Giesbrecht (2015), Olson et al. (2016). This hypothesis therefore remains to be verified by further studies.

The ERPs P1, N1, P2 and N2 appear earlier than P3, and involve lower-level sensory processing. Considering the studies that compared results between exercise and resting, these ERPs were not treated systematically. For the lowest amplitude of N1 observed, the authors of Pontifex & Hillman (2007) suggested a deterioration in visual attention, while the increased amplitude of P2 suggested an increase in attentional selectivity, and the low amplitude of N2 suggested an attentional conflict.

However, these hypotheses are weak because they relate to results that have not been reproduced by other studies. Further studies are needed to assess the validity of each of these hypotheses or theories.

Discussion of methods and limits of evidence

The studies in this review did not use same criteria to define exercise intensity. The validity of the definition of intensity in these studies could therefore be called into question. In Bullock, Cecotti & Giesbrecht (2015), basing the determination of “high intensity” on the assessment of perceived effort (using the Borg scale) is unreliable, especially for subjects who are not used to assessing their effort. In fact, the “high intensity” value varied from 70 W (which almost corresponds to a warm-up intensity) to 120 W, depending on the subjects. In Torbeyns et al. (2016), the intensity was low and determined by an exercise test, but it is unclear whether this was a conventional exercise test, and the reliability is unknown. In addition, in several studies, target intensities were calculated only theoretically. As a result, it was not possible to classify the studies as low, moderate or vigorous intensity to assess the effect of intensity. Since exercise intensity and cadence were not always rigorously defined, there could be potential for bias due to different exercise conditions between subjects. According to MacIntosh et al. (2021), the distinction between low and moderate-to-vigorous intensity should be determined by measuring a blood lactate threshold or ventilatory threshold. These physiological parameters are well linked to homeostatic change which, in theory, makes it possible to distinguish between low-intensity exercise and moderate- or vigorous-intensity exercise. Descriptions from American and Canadian institutions also suggest heart rate percentage values and other values that are associated with moderate or vigorous intensity levels.

Testing several intensities under the same conditions would be a reliable way of assessing the effect of intensity. However, only three studies did so (Bullock, Cecotti & Giesbrecht, 2015; Olson et al., 2016; Dodwell et al., 2021) and each had its own method of measurement and analysis, as well as different intensity criteria, thus making it difficult to compare their results. In addition, studies in which the exercise takes place in an outdoor environment (park, road) are exposed to environmental variability, such as weather parameters and traffic conditions (Zink et al., 2016; Scanlon et al. 2019, 2020; Robles et al., 2022).

Exercise duration also varied across studies. In most cases, the exact duration of cycling was not mentioned. It consisted in performing several blocks of numerous short trials of the attentional task, sometimes with pauses between blocks. In some studies, the exercise condition lasted 20 min or more (Bullock, Cecotti & Giesbrecht, 2015; Olson et al., 2016; Olson, Cleveland & Materia, 2023; Dodwell et al., 2021), but this was probably not the case for most studies. However, the significance is not necessarily the same if ERP was measured at the start of cycling exercise, or at the end, or during exercise with recovery breaks.

Furthermore, the methods used to clean EEG data and quantify ERPs, such as amplitude, can affect the study results. For example, the amplitude of the P3 wave could be quantified as a peak value relative to baseline, or as the difference between the amplitude of rare tones in the auditory oddball and the amplitude of frequent tones comparing cycling and resting conditions as in Olson, Cleveland & Materia (2023). However, we could not be sure of the exact method used to quantify ERPs in all studies. For example, in the methods section of Torbeyns et al. (2016), it is stated that “The latency and amplitude of each ERP component were quantified using the mean amplitude and corresponding latency […]”, but exactly what this means is unclear. This type of difference between studies may have less effect thanks to within-group models, but may nevertheless influence results based on the comparison between a cycling condition and a resting condition.

The study populations were predominantly mixed, and mostly aged between 20 and 25. However, the EEG signal can be influenced by a person’s age (Remijn et al., 2014), which could have repercussions on measurement results. For example, one study found that older participants show a lower MMN amplitude than younger participants (Gaeta et al., 1998). Therefore, the results of the studies in this review cannot be generalized to all ages.

Finally, the laterality or handedness of the subjects was not systematically reported. Only seven studies out of 20 indicated that they had included right-handed subjects only. However, ERP components can be affected by handedness (Remijn et al., 2014), such as in motor preparation (Schmitz et al., 2019). It has also been shown that handedness can have an impact on the ERP components from a visual oddball task (Eskikurt, Yücesir & İsoglu-Alkac, 2013). For that reason, it may be problematic to mix right-handed, left-handed or ambidextrous subjects in the same study.

Limits of this study

One difficulty in the search process for articles to include is that we cannot know whether ERPs were measured during cycling with titles such as “Effect of cycling on ERPs” because it may be related to EEG measurements after acute exercise only. We did not find a search equation that made it possible to select ERP measurements performed only during cycling without resulting in some of the included studies being overlooked.

This review included some studies that did not have a resting condition to compare ERPs with a cycling condition. This complicates the comparison and organization of results due to heterogeneity.

The Cochrane Risk of Bias Assessment Tool 2 was used to assess the risk of bias in the included studies. The categories in this tool concern bias within studies. They differ from the problem of methodological heterogeneity between studies. The tool comprises five categories of possible bias. It was found that some categories do not correspond well to the type of studies in this review, notably “deviations from planned intervention”, “outcome measurement” and randomization bias. It seems that this tool is best suited to blinded studies of the efficacy of new treatments. It is possible that this tool was not applied appropriately, but there may be other instruments that are more suited to the type of studies included in this review. However, the “missing data” category found that the studies included did not consistently report a problem of imbalance in the amount of data analyzed between the exercise and resting conditions. We hypothesize that, due to movement, EEG data are more likely to be marred by artifacts in the cycling condition than in the resting condition. This could have an impact on the quantity and quality of data between the two conditions and bias the results. Overall, the Cochrane tool enabled us to conclude that there was a moderate risk for all studies due to the “missing data on results” category.

This review did not include studies that measured ERP during aerobic exercise other than cycling, such as rowing, walking or running on a treadmill. Including studies of other forms of exercise would broaden the range of results at the cost of greater heterogeneity in exercise conditions. The review by Schmidt-Kassow & Kaiser (2023) included the latter two types of exercise, but with a different objective from that of the present review.

Finally, this review did not provide detailed information on how to measure ERPs during cycling or how to deal effectively with artifacts.

Perspectives

A recent study (Schmidt-Kassow & Kaiser, 2023) addressed the subject of behavioral and EEG measures during cycling, walking or running, with studies assessing various cognitive processes such as attention, inhibition, memory, vigilance and cognitive flexibility. They reported that behavioral studies assessed more conditions of intensity, duration and cognitive paradigms than EEG studies. However, most EEG studies (which are partly common to this review) assessed attentional abilities via an oddball task. According to the authors, in EEG studies, researchers were more concerned with demonstrating the reliability of ERP results in a motion condition. We speculate that this might be one reason why they used mainly an oddball task. We mentioned in the feasibility part of this discussion that ERPs can be measured in different cycling conditions. This fact gives confidence in the assessment of other aspects of cognition (other than attention) during aerobic exercise. This would make it possible to compare behavioral and EEG studies and assess the effect of aerobic exercise on various cognitive processes during exercise.

The studies in this review used ERP measurement during cycling for a variety of objectives. Notably, the studies by Scanlon et al. (2019, 2020), Robles et al. (2022) reported an impact of urban cycling on attention via ERPs. Their research may be linked to knowledge of the potential risks of this type of situation, since attention is a key ability in managing risks on the road.

ERPs during cycling may also be of interest in sports research, since (Grego et al., 2004) found changes in ERP amplitudes during a 3-h long race for cyclists. ERPs during cycling can also be linked to concentration during sports performance, and (Schmidt-Kassow et al., 2013; Conradi et al., 2016; Schmidt-Kassow, Thöne & Kaiser, 2019) found that actively synchronizing one’s cycling cadence with a metronome had an impact on ERPs.

Some studies attempted to assess the impact of different exercise intensities on ERPs (Olson et al., 2016; Bullock, Cecotti & Giesbrecht, 2015; Dodwell et al., 2021), but further studies are needed to understand how intensity may impact components of ERPs such as amplitude and latency.

We could also imagine measuring ERP during cycling with objectives for overall health. Indeed, the ERP amplitude could be used as a marker of concentration or motivation, and could be applied in cardiac rehabilitation to assess motivation during a rehabilitation session.

Common definitions of intensity should be used to make it easier to compare results between studies. It might also be preferable to define exercise intensity using the same method. By taking a percentage of a theoretical heart rate reserve, as in Dodwell et al. (2021), only the resting heart rate of the participants is required, and this is less demanding than doing an exercise test for each individual. The theoretical heart rate reserve method is easy and would be sufficient if studies also referred to official ratios to decide on intensity levels, as presented in Table 1 of MacIntosh et al. (2021).

In addition, all exercise conditions should also be detailed in the articles. It is important to specify the cadence used during cycling sessions because it can influence intensity (even if the effect is small). Indeed, as shown in Akaiwa et al. (2022), different cadences can lead to different behavioral and ERP results. The time at which measurements are taken is also important, since results do not have the same significance if recorded at the beginning or end of an exercise session, even if it is of light intensity.

In order to be able to extrapolate the results to the general population, researchers should include a wider range of ages, or carry out specific studies on older populations, such as cardiac rehabilitation patients, who could benefit from such research in the future, with the development of new technologies. It is also important to systematically mention the characteristics of the study population (number, age, gender and laterality), as these may have an impact on the results and be useful for future meta-analysis.

The paucity of studies on vigorous intensity and the variability of methods lead to considerable uncertainty as to the effect of intensity on attentional ERPs, including the most widely studied parameter, namely P3 amplitude. The P3 latency and other ERPs are reported less frequently. Additional studies are needed to include two cycling conditions of different intensity in order to better assess the immediate effect of cycling exercise on ERPs and attentional abilities.