Habituation to pain: self-report, electroencephalography, and functional magnetic resonance imaging in healthy individuals. A scoping review and future recommendations

Supplemental Digital Content is Available in the Text.


Introduction
The concept of habituation has a long history, and it is one of the first and simplest forms of learning that has been studied. 13Ever since the first known research description of habituation in 1887, 70 the definition of habituation has been under construction and in development.In the context of exercise, habituation has been named "practice effect," 33 whereas in other fields, it was named "adaptation," "extinction," or "fatigue." 13,94,95The first list of behavioral characteristics of habituation was composed by Thompson and Spencer. 95They established a definition for habituation, which was updated in 2009 by Rankin et al. 75 : "a behavioral response decrement that results from repeated stimulation and that does not involve sensory adaptation/ sensory fatigue or motor fatigue." 75Changes in behavioral response may be linear or exponential and show recovery over time. 75Although the neurobiology of habituation is still not completely understood, it became clear that it involves some form of neural plasticity. 13The process that early on was observed on the behavioral level as "fatigue" of the body turned out to have an underlying neural process.When merely sensory or motor fatigue is present, it is nowadays excluded from the general definition of habituation.However, some scholars adhere to 2 definitions, where central habituation is taking place at the spinal cord and brain (excluding sensory effects) and peripheral habituation has been related to effects taking place at the sensory afferent level (also referred to as "fatigue" or "adaptation"). 32,76Throughout this review, we adhere to habituation as the central process and refer to peripheral habituation in the discussion of sensory effects.
Short-term and long-term effects of habituation differ with respect to their timescale and have been recognized as different concepts.They are defined as response decrements within one session (with a time course over minutes up to a few hours) and over several daily repetitions (with a time course of hours, days, or weeks), respectively. 75,89Similarly, qualitatively different mechanisms are involved in short-term and long-term habituation. 75esides habituation, repeated stimulation may instead also lead to sensitization: "an increase in behavioral response over time." 34,101Sensitization is, like habituation, considered a simple form of nonassociative learning. 75,95Yet, it is unclear how these processes might interact.
Several theories for habituation and sensitization have been proposed. 94The current prominent theories are the stimulus model comparator theory, the Wagner-Konorski Gnostic Unit Theory, and the dual process theory.The stimulus model comparator theory 90 describes that new stimuli are first amplified but inhibited when they resemble the model of stimuli (ie, a form of predictive coding).The same applies to the Wagner-Konorski Gnostic Unit Theory, 53,106 except that this theory also involves an associative network that is able to influence the model.The dual process theory states that habituation and sensitization are independent processes that interact and result in a final response. 34abituation and sensitization are central concepts for the understanding of pain.Pain is an important mechanism in signaling threat to the body, and learning is critical in handling pain.Habituation and sensitization to pain are the core part of this learning process and may result in (mal-)adaptive changes in pain processing.The underlying motivational-ethological framework poses that habituation to pain occurs when threat is relatively low and it is adaptive to pursue other goals. 15The environment and social context also seem to be of particular importance. 15The literature of habituation to pain dates back as far as the 1960s.Early studies used repeated immersions in cold water and focused mostly on physiological signals like blood pressure. 23,55owever, pain sensation in itself became more important in understanding habituation to pain and was reported to diminish after several days of the cold pressor test. 23Moreover, LeBlanc and Potvin 55 suggested the involvement of the central nervous system in habituation to pain.Subsequent studies used electrical dental stimulation, 14,26 measured firing of trigeminal peripheral nerves, 46 and evoked potentials 14 to address habituation to pain in various settings.Since these early studies about habituation to pain, the literature has expanded tremendously, including standardized methods for measuring pain sensations using the numeric rating scale (NRS) and visual analogue scale (VAS), development of paradigms as well as testing in different contexts, and advances in measuring such as electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) for the investigation of neural substrates of habituation.Behavioral and neural responses are now commonly seen as measures of habituation. 75It needs to be stressed that most literature does not distinguish between habituation of nociceptive processes or pain. 1 There is a conceptual difference between nociceptive processes and pain.Pain refers to an unpleasant sensory and emotional experience, whereas nociception describes the processing of noxious stimuli in the nervous system. 74Neural activity described in fMRI and EEG literature may or may not result in (self-report of) pain perception. 1 Although for this review, we adhere to descriptions as used in the original articles, this is an important point to consider for fMRI and EEG studies that do not include self-report.So far, the literature on habituation to pain including EEG and fMRI measures have not been summarized.This is of major importance to increase our understanding of the neural processes underlying habituation to pain.Furthermore, the link between neural and behavioral measures of habituation is not well understood.For example, standard measures, such as selfreport ratings, have only recently been explored in conjunction with potential neural correlates.Therefore, the objective of this scoping review is to summarize the current state of the field of habituation to pain using self-report ratings, EEG, and fMRI in healthy individuals.To limit the scope of the review, behavioral measures, such as reflexes, were not considered.Moreover, recommendations for further studies are provided, including the use of terminology, experimental design, and methods to analyze habituation to pain.

Methods
This scoping review follows the PRISMA Extension for Scoping Reviews (PRISMA-ScR) guidelines and reports accordingly. 98he review protocol is preregistered at the Open Science Framework: osf.io/nypbw.

Search strategy
PubMed, PsycINFO, and Web of Science databases were searched for eligible studies from database initiation until July 2020.Search terms consisted of elements related to pain AND habituation (the search strategy for PubMed can be found in the Supplementary Files, available at http://links.lww.com/PAIN/B920).Terms related to "repeated painful stimulation" were also added, to ensure a full inclusion of studies where habituation might be investigated.No filters were applied.Reference lists of articles fulfilling the inclusion criteria were manually screened for additional references.Before finalizing the article, a final search was performed including all articles published until 20 August 2022.

Study criteria
Exclusion criteria were preregistered and evaluated in order (see Table 1; inclusion criteria are added for clarity).This review focuses on self-report ratings and brain measures (EEG and fMRI) of habituation.Consequently, articles should include one of these 3 measures and describe a habituation process involving repeated painful stimulation at the same intensity.This latter point was evaluated in 2 ways: first, the concept of habituation should be discussed in the article and, second, a change in pain, measured through ratings or neural activity, should be quantified.The decision whether a change in pain, measured through ratings or neural activity, is referred to as habituation or not was left to the original authors.With respect to intervention studies, we only included intervention studies if they obtained measures of habituation separately before and after the intervention (ie, to see the effect of an intervention on habituation).Studies that merely reported a change in pain ratings (or amplitude) after an intervention were excluded.Habituation effects of an intervention that itself included repeated stimulation (eg, transcutaneous electrical nerve stimulation) were also excluded.Accordingly, we only included effects of repeated stimulation during control conditions of interventional studies if habituation was an outcome measure of the interventional study.Because of our broad search, many studies were identified with well-known paradigms (eg, cold pressor test, high-frequency stimulation), but either these studies did not discuss their results in the context of habituation or they did not quantify habituation before or after an intervention.Therefore, these studies were not included.Studies that included paradigms where the stimulation intensity was adapted as a measure of habituation were also not included.Studies on habituation of pain thresholds were also excluded because they do not require repeated stimulation above the pain threshold, which is our primary interest and involves a different evaluation process (eg, decision making whether something is painful and how much pain can be tolerated).Articles were screened during selection whether the stimulus was painful, through the use of calibration, description of the stimuli as painful (eg, stimuli elicited a sharp painful sensation), or the consistent use of habituation to pain throughout the article.

Selection procedure
References from PubMed, PsycINFO, and Web of Science databases were combined in EndNote and deduplicated.Titles and abstracts were screened for eligibility by 2 independent reviewers (M.vdM.and C.V.).Full-text screening was performed as a final selection round for all articles that were likely relevant or unclear based on the title and abstract only.This was performed by the same reviewers (M.vdM.and C.V.).Discrepancies for inclusion were discussed with a third independent reviewer (J.P.) to come to an agreement.This review was originally preregistered at the Open Science Framework to include both healthy participants and chronic pain patients, but it was decided to deviate from the protocol and only include healthy participants.Our decision was made based on the fact that there is abundant literature available to describe the current state and neural mechanisms in healthy participants.This review will serve as an important step and foundation for further review of habituation to pain in chronic pain patients. 102

Data extraction and synthesis
All relevant data from the texts were extracted by the first reviewer (M.vdM.).Extracted data contained study design, description of population, period (long-term or short-term habituation), habituation protocol, type of pain stimuli, site and number of pain stimuli, duration, interstimulus interval, statistical analysis, brain activity (eg, event-related potentials, activated areas), and variables that might affect habituation (eg, anxiety).Studies were categorized based on measurement type (ie, self-report, EEG and fMRI) and topic (eg, experimental settings).

Visualization
Extracted data were visualized using R Statistical Software 4.2.1 73 in RStudio version 2022.7.1.554 82using packages ggplot2 3.3.6, 108ggforce 0.3.4, 71and ggpubr 0.4.0. 48To visualize areas that show fMRI activity related to habituation, only those articles were included that reported significant habituation effects in healthy participants and were corrected for multiple comparisons.Peak MNI coordinates were imported into Brain-Voyager 22.2 (Brain Innovation, Maastricht, the Netherlands), 30 and a sphere (5-mm radius) was drawn around the peak coordinates followed by the export of these regions of interest.This approach facilitates the visualization of the individual study peak coordinates and consistency in the peak localization of habituation effects, rather than merely the overlap of clusters.Talairach coordinates were converted to MNI coordinates using BioImage Suite Web. 54,67The regions of interest were then visualized using MRICron. 80

Results
Database search of PubMed, PsycINFO, and Web of Science identified 5100 records (Fig. 1).Automatic and manual deduplication allowed to remove 1759 records, resulting in 3341 records for abstract and title screening of which 228 articles were retrieved for full-text screening.After screening by 2 independent reviewers (M.vdM.and C.V.) and discussing discrepancies with a third independent reviewer (J.P.), 54 articles were selected and included.Reference list tracking added another 5 articles, and a final repetition of the searches to ensure that no recent articles were missed allowed to add 4 more articles, resulting in a total of 63 articles.Two studies that did not meet the imaging criterion were included for the self-report results only.As described in the methods section, articles including a chronic pain population were excluded.Currently, there is a movement toward open science and data sharing.However, in our studies, only one study was preregistered, 93 and 2 studies shared data and/or code. 6,59he majority of the articles included in this review are more than 5 years old (Fig. 2A), and this likely contributed to the limited open science practices.Furthermore, the sample sizes are, in general, small, with a median sample of n 5 16 for within-subject and between-subject studies (Fig. 2B).Studies with larger sample sizes are needed to increase power and chance of replicability, but only a few of such studies were identified.Three different modalities were included: self-report (n 5 29), EEG (n 5 21), and fMRI (n 5 13).Most studies investigated short-term habituation with self-report and/or EEG (Fig. 2C).All long-term studies using EEG or fMRI (n 5 4) also included self-report, whereas for short-

Criterion
Included Excluded term studies, this was only included in 2 fMRI studies (22.2%) and in 13 EEG studies (72.2%) (see Figure 2D) (note that these percentages are in agreement with Fig. 2, where articles using the same data sets are omitted).

Methodological variations
In terms of measuring habituation, there have been considerable methodological variations (see Fig. 2).First, the duration of the study and measured habituation effect does considerably differ.Long-term studies (n 5 14) measured habituation over several days, whereas short-term studies (n 5 49) defined habituation based on a single session.Long-term paradigms included 8 to 11 days in which participants received several blocks of painful (heat) stimulation on each day (a procedure initially developed by Bingel et al. 9 ).Changes in self-report or brain activity were then measured over days and sometimes as well within a day.Short-term paradigms showed more variations in setup, time window, and the number of stimuli (given the short timescale), ranging from only 3 to 150 stimuli (median 5 20; Fig. 2E).In addition, a wide range of stimuli durations have been used from only 0.1 ms (electrical stimuli) or 1 ms (heat stimuli using a laser) to 40 seconds (heat stimuli using a thermode) or even minutes in the case of a cold pressor test.Second, studies differed in whether they used individual calibration to determine the strength of the stimulus (Fig. 2F).When calibration is not used, it may result in large differences in experienced pain and ceiling or floor effects (eg, the stimuli might not be that painful for some participants, whereas others might perceive them as too painful).This difference in calibration to determine the strength of the stimulus may affect the outcome (ie, habituation or sensitization patterns) because habituation tends to be more rapid and/or more pronounced with lower stimulus intensities. 75hird, changes in pain intensity were measured in various ways, for example, within a run (ie, series) of stimuli and/or across runs that could be spaced within or across days.Such variations in timescales and range of stimuli result in different operationalizations of habituation (ie, how to define and measure habituation), hampering comparisons across studies.Fourth, various methods were used to define and present habituation.For example, percentage change from a last to first block of stimulation has been reported, but direct comparison of trials or testing for linear trends in the data was also reported.With methodological variations to define habituation, utmost care should be taken for a correct interpretation and comparison.Averaging over trials of the dependent variable can lead to different results than the "raw" single trial or block data, which is preferred when the data quality allows this.The variation in "calculating" and defining habituation is closely linked to a fifth issue, which is the use and selection of the statistical model.All the methodological variations result in difficulties to correctly compare the outcome of the individual studies.In this context, the formulation of general conclusions is challenging.The next sections describe the results of studies on habituation to pain in healthy participants as related to the modality self-report (section 3.2), EEG (section 3.3), and fMRI (section 3.4) as well as associations among them (section 3.5).

Type of stimuli
The majority of self-report habituation studies in healthy participants have used heat stimuli using a thermode (n 5 21), (Fig. 3).
Several studies using heat stimulation in long-term paradigms showed habituation over days (Tables 2 and 4). 9,20,76,77,92Some of these long-term studies reported sensitization patterns within one day. 11,25When investigating short-term patterns, however, which should be equivalent to effects within one day, habituation is reported as well. 52,97Thus, there does not seem to be a reliable pattern for these short-term effects of heat stimuli.Habituation using repeated cold (water) stimuli is also investigated (Table 2), and it was shown that some participants showed habituation over series of the cold pressor test. 111The latter was confirmed and extended at the group level 91 and with the use of cold pain stimuli through an MRI-compatible thermode. 109

Population characteristics
Age and sex have been investigated with respect to population characteristics (see Tables 2 and 3).With age differences, habituation trajectories were shown to slightly differ over time, as in particular, older subjects showed reduced habituation and in general more pain. 21Conversely, no effect of age was shown in a recent study. 49Studies focusing on sex differences reported mixed results.Three studies reported no sex differences in habituation with the use of either a short-term 32,97 or a long-term paradigm. 11Conversely, 2 studies reported increased habituation in women as compared with men, which might have been related to use of lower stimulation intensities for women. 18,38hen using the same stimulus intensity, habituation was present for both sexes, and no differences in habituation strength were shown. 39Jepma et al. 47 also reported habituation to be present in both sexes, although it was more pronounced in women.Overall, sex effects for habituation to pain do occur but likely depend on the use of a specific paradigm, the calibration of stimuli, and/or used stimulation intensity.

Expectations and attention
Several psychological variables such as expectations and attention were reported to influence habituation (see Tables 2 and 4).In a study about expectations, participants were instructed that they would either habituate or sensitize over time, and this resulted in decreased or stable responses, respectively. 20Furthermore, a nocebo manipulation resulted again in stable, that is, not habituated, pain ratings over several days, whereas controls not exposed to this manipulation habituated as usual. 78A similar study showed longterm-habituation in both a control and a nocebo context groups over 21 days, although effects in the nocebo context group were less pronounced. 25Interestingly, when rating imagined heat pain, participants reported sensitization, whereas the actual rating of physical stimuli led to habituation. 27An unexpected change in electrical intensity was also reported to affect habituation. 3,4ttention and fear have also been linked to habituation for modulating expectations and information processing.Yet, results are still diverse, and large-scale investigations are lacking.For instance, when attention was modulated, distraction resulted in habituation, whereas attention resulted in the absence of habituation. 2Anxiety did not affect habituation in the same study. 2urthermore, reduced habituation to pain has been shown in participants with higher levels of pain-related catastrophizing 22 but not in participants with sensory overresponsiveness. 107ifferences between sensory and affective aspects of habituation were investigated in 2 long-term heat pain studies showing both habituation of pain intensity and pain unpleasantness over days. 11,28reimhorst et al. 11 reported an increase in both pain and unpleasantness within days, whereas, interestingly, the slope within these days diminished across days.Conversely, a preliminary shortterm study did not report a decrease in pain and unpleasantness. 58n conclusion, expectations and attention are shown to modulate habituation to pain on a short-term and long-term timescale.Both sensory and affective aspects of pain are subject to habituation, although, sometimes, conflicting evidence is presented in the literature with respect to effects of anxiety and attention.

Experimental settings
Studies with respect to experimental settings showed that habituation to pain depends on spatial summation (ie, small distance between probes reduced habituation) and (initial) stimulation intensity (stronger habituation for lower (initial) stimulus intensities). 18,39,47This latter point is in line with the properties listed by Rankin et al. 75 but is not always observed.In some studies, the rate of habituation appears to be the same, but only the amplitude (ie, level of the rating) increases with higher stimulus intensities, for example 28,109 .The site of stimulation might also play a role.For instance, stimulation on the forehead resulted in larger within-session sensitization than stimulation on the arm. 86Habituation over days did not show any differences with respect to the stimulation site.Site-specific effects were further investigated in a paradigm with 3 runs of heat stimulation, where within each run a different (adjacent) site was stimulated, which was repeated in the next runs. 47This study showed sensitization within the first run and habituation across the 3 runs, potentially reflecting site-nonspecific sensitization and sitespecific habituation, respectively. 47Greffrath et al. 32 also reported a strong initial habituation effect for stimulation on the same site and a steady, but slow, decrease when using a variable stimulation site.A central component of habituation to pain was subsequently tested by investigating transfer effects to the nonstimulated arm.In this paradigm, participants are stimulated several times on one arm ("stimulated arm") and then once more at the other arm ("nonstimulated arm"), thereby investigating whether the participant would report a similar or different intensity.Although stimulation at the nonstimulated arm resulted in an increase of pain (ie, no continued habituation or transfer effect) in one study, 28 another study showed habituation to pain for both participant's stimulated and nonstimulated arms. 76This suggests involvement of both spinal and supraspinal mechanisms. 76Therefore, the intensity and site of stimulation seem to affect habituation to pain, in particular with the use of repeated heat stimulation on the same site.2).DNIC involves the inhibition of nociceptive input at the spinal level when (simultaneous) noxious stimulation is taking place elsewhere on the body.It was shown that both habituation and DNIC resulted in decreased pain Habituation to pain: self-report.intensity, with a significant correlation between these effects, indicating a general endogenous analgesic effect.In an offset analgesia paradigm, which is a different endogenous pain modulation paradigm, the stimulus temperature is increased for a few trials and then returned to baseline, which usually results in a disproportionately large reduction in pain perception.In this paradigm, habituation effects were shown in the higher temperature trials, indicating that several processes might occur simultaneously. 93Furthermore, the development of habituation was shown to be unaffected by the opioid receptor antagonist naloxone.The authors concluded from this that the underlying mechanism does not directly involve the endogenous opioid system. 76In summary, inhibitory modulation of pain, as for instance, DNIC or offset analgesia, may show similar decreases in pain as noted by habituation.Hence, the investigation of potential common pathways involved in either DNIC or habituation to pain would be an interesting direction for future research.The endogenous opioid system does not seem to be involved.

Reliability of habituation effects over time
Use of different stimulation timescales may also effect habituation to pain.In a paradigm investigating effects over different timescales, it was shown that in this study, repetitive brief pulses (max 1.8 Hz) induced sensitization; however, over trials (within runs), this resulted in habituation to pain, and across runs, the ratings were stable. 51Therefore, it is important to distinguish and compare analyses over different timescales because they might show opposing effects.Analyses of intersession variability for pain intensity also revealed decreased habituation to pain over sessions, for different temperatures. 72Nevertheless, an earlier study with a similar design, but smaller sample, did not show any habituation effects over sessions. 81Furthermore, a recent study showed that pain ratings decreased within a block and remained stable over blocks. 16However, a comparison of pain ratings between 2 visits showed very low reliability. 16Overall, reliability of estimates of habituation to pain across timescales seems rather limited, and the extent to which habituation occurs across sessions and timescales needs to be determined.

Individual differences
Importantly, large interindividual variability has been shown in rating data, ranging from sensitization to habituation to pain within a run of heat stimuli. 47These individual differences were also noticed in a long-term heat study, where no habituation effects over time across the group of healthy volunteers (n 5 27) were shown as compared to controls without any stimulation. 92evertheless, behavioral data demonstrated that a subgroup of participants (n 5 14) sensitized, whereas the others (n 5 13) habituated over the stimulation days.Similar long-term paradigms observed habituation on the group level but sensitization in 25% of the participants. 9,77Individual differences have also been noted in a study by Slepian et al. 88 : 25% of participants reported habituation, but the group-level analysis showed a sensitization effect. 88These individual effects are thus very important to report and must be taken into account for a comprehensive understanding of habituation to pain.

Conclusions
Self-report studies have shown that psychological aspects, such as expectations, are able to influence habituation to pain.The effects of population characteristic on habituation, like age and sex, are not yet conclusive.Experimental settings, such as the number of stimuli and stimulation site, also influence habituation to pain and results vary depending on the investigated timescale.Habituation to pain might have common mechanisms with other pain modulation paradigms.Individual differences should be the focus of future research because this will provide detailed insight into the variability in habituation to pain.In contrast to self-report studies, which mainly used heat stimulation with a thermode (section 3.2.1),most EEG studies used heat stimulation using a laser and electrical stimulation (Fig. 3).Brief stimulation allows measuring evoked potentials, such as the N1, P1, N2, or P2 amplitude.Contact heat-evoked potentials (CHEPs) have been developed over recent years and consist of very brief heat stimuli with reduced rising times compared with standard thermode heat stimuli, which make them more compatible with EEG. 31 In this review, we adhere to the description of EEG components as used in the original articles (eg, N2-P2 amplitude, when it is described as such in the article).

Population characteristics
In view of habituation to pain based on EEG recordings, population characteristics, such as age and genetic differences, have been studied (Table 3).Habituation of the N2-P2 amplitude was shown to be affected by a genotype involved in monoaminergic activity and regulators of gene expression. 19,84Age effects on the N2-P2 amplitude showed an increase in habituation over several groups between 18 and 72, although there were no differences between the groups after a Bonferroni correction. 17Another study reported a decrease in the N2-P2 amplitude in participants between 50 and 70 compared with young subjects (mean age 5 24.9 years). 49ffects of sleep (or the lack thereof) have not been found to affect habituation for both pain ratings and the N2-P2 amplitude.This was observed regardless of whether a crossover within-subject design 61 or a randomized between-subjects design 66 was employed.Conversely, in another crossover within-subject design, Schuh-Hofer et al. 87 observed increased habituation of the P2 component after sleep deprivation, in both a neutral and focusing condition, but not during distraction.However, habituation was not related to individual sleepiness scores, and the sample was considerably smaller than the first 2 studies. 87In conclusion, sleep and age do not seem to affect habituation to pain in a consistent way.Furthermore, there are indications that several genotypes may affect habituation to pain.These population characteristics should be considered in future studies.

Expectations and attention
EEG research indicates that the expectation to feel more pain decreases habituation of the N2-P2 but not the N1 amplitude (Table 3), 69 although this has not been replicated in larger-scale studies.Habituation of the N1 and P2 component was also shown when investigating different attention and distraction tasks (such as mental arithmetic), indicating that attention or distraction did not affect habituation to pain. 10

Experimental settings
Paradigms using laser-evoked potentials (LEPs) have consistently shown a decrease in N2-P2 amplitude over blocks of stimuli (Table 3). 41,42In addition, LEP-evoked habituation to pain was investigated under the influence of capsaicin-induced peripheral and central sensitization. 43Capsaicin-induced central sensitization did not alter the LEP-evoked habituation, whereas at the same time, habituation was shown to be reduced in the primary stimulation area after applying capsaicin. 43When using a fixed location for CHEPs, a fast decay of the N2-P2 amplitude followed by a plateau is reported. 32he stimuli at the fixed location may have resulted in fatigue at the sensory afferents, that is, peripheral habituation.With the use of a variable location, a gradual decrease in N2-P2 was observed, and therefore, central habituation is likely to occur. 32In conclusion, EEG recordings and experimental settings can be used to determine and discriminate between peripheral and central sensitization as well as peripheral and central habituation to pain.

Investigating mechanisms of habituation
In research on conditioned pain modulation (CPM), a study compared habituation and CPM effects 24 : here, the CPM showed a stronger reduction in EEG amplitude as compared to the habituation protocol.It was concluded that the underlying mechanisms of CPM and habituation might partially overlap. 24

Reliability of habituation effects over time
A reproducibility study using CHEPs reported similar habituation to pain over 2 sessions (6 months interval), although the average amplitude was increased during the second session. 83A recent study reported no effects on habituation to pain within and across blocks for the N2-P2 amplitude, and starting values (first trials) were similar between the 2 visits. 16

Analysis methods and statistical approach
Use and implementation of state-of-the-art analysis methods and/ or correct statistical approaches are of utmost importance for correct and optimal interpretation of results.The current standard is looking at trends over time in evoked potentials, where effects are not always established (Table 3). 12The use of new analysis methods demonstrated the presence of habituation to pain on a finer timescale. 104Multilevel modelling showed the benefit of using single trial data and including decay functions such as quadratic effects to model habituation over time.Moreover, it is possible to take into account individual differences by including a random intercept and/or slope in the model.This statistical approach of multilevel modelling was extended by the event-related, fixedinterval area (ERFIA) technique, which partitions the data into small epochs, calculates the area under the curve, and uses this as dependent variable in a multilevel model.This enabled demonstration of habituation effects outside the peak amplitude, for example, between 100 ms and 160 ms after stimulus onset. 103A higher temporal resolution was also pursued in a paradigm with LEPs, to minimize the required stimuli. 49Another approach in characterizing the event-related potential (ERP) waves is the use of singular value decomposition. 59This approach revealed similar decay functions of N1 and N2-P2 waves and similar waveforms for habituated and nonhabituated trials. 59New EEG analysis methods and modeling combined with advanced statistical approaches allows improved and detailed analysis of habituation to pain.

Conclusions
EEG studies have significantly contributed to our understanding of habituation to pain.Habituation to pain seems to play an important role in the reduction of the N2-P2 amplitude, although this is less robust for the N1 amplitude.Although effects of genotype, sleep, and distraction are still inconclusive, different experimental settings allow to investigate both peripheral and central effects of habituation to pain.New analysis methods and state-of-the-art statistical approaches further allow for a more profound understanding of mechanisms underlying habituation to pain.Pain is associated with activity in several brain areas including the somatosensory cortices, insula, cingulate cortex, periaqueductal gray, and prefrontal cortex. 60,110A decrease in activity across (sequences of) pain stimuli has been interpreted as habituation, although this still is a matter of discussion (see also Section 4.1 Challenges in the field).In neuroimaging fields outside pain, this activation decrease has also been described as the repetition suppression effect, but the differences with habituation have not been systematically reviewed. 5Short-term habituation to pain with the use of fMRI has been investigated using analyses over different time windows (Table 4). 7,36,45,63,64,68Over the course of 4 heat stimuli, brain activity decreased in several areas, including the bilateral frontal gyrus, left insula, and bilateral anterior cingulate gyrus. 7A similar study using 2 runs of 4 electrical stimuli showed a decay in activity in the left anterior cingulate cortex (ACC) and left primary somatosensory cortex (S1) over the 4 stimuli but not over the 2 runs. 45Two subsequent studies used heat stimuli with a laser and contrasted stimuli of the first half with the second half of the experiment. 63,64Although the first study did not show any habituation, the second (more powered) study showed increased activity in the first half of the experiment in the contralateral S1, bilateral secondary somatosensory cortex (S2), bilateral insula, the posterior ACC/midcingulate cortex (MCC), bilateral supramarginal gyrus, and bilateral parietal operculum (see Supplemental Table 1 for a complete overview, available at http://links.lww.com/PAIN/B920). 63,64Furthermore, a study comparing several runs of electrical stimulation reported decreased activity in the MCC, left S2, and ventral anterior part of the right insula. 36Similarly, a recent work reported that the bilateral inferior frontal gyrus (IFG), including anterior parts of the insula, bilateral S1/S2, and the anterior MCC (aMCC), showed decreased activity over runs. 68Moreover, activity in the aMCC was specific to habituation of painful compared with nonpainful stimuli, thereby indicating a role in pain identification. 68In addition, exploratory analyses also indicated habituation within runs in the IFG and aMCC.The groundwork for fMRI long-term habituation to pain studies was conducted by Bingel et al. 9 with a study involving 8 daily stimulation sessions.Repeated painful heat stimulation resulted in long-term habituation because changes in brain activity were observed on day 8 compared with day 1, which then remained stable until day 22.Over the 8 days, brain activity decreased in the bilateral anterior insula, the medial thalamus, contralateral S2, and bilateral putamen, whereas increased activity over time was found in the rostral ACC (rACC), specifically the subgenual ACC (sgACC). 9Therefore, the sgACC was suggested to be involved in antinociceptive activity and mediation of (long-term) habituation to pain.The brain activity in the sgACC was shown to return to baseline (ie, similar to day 1) in a 1-year follow-up study. 8n conclusion, fMRI studies demonstrated neural habituation effects for sequences as short as 4 stimuli, ranging to long-term effects over several runs, mostly located in the cingulate cortex, insula and somatosensory cortices (see Supplementary Table 1, available at http://links.lww.com/PAIN/B920).The S2 and insula were sensitive to both short-term and long-term habituation, whereas activity in the ACC/MCC seems to be unique for shortterm habituation.By contrast, activity in the sgACC/rACC seems to be related to long-term habituation only (Fig. 4).

Expectations and mechanisms of habituation
Specific genotypes, 56 the practicing of meditation, 57 or administration of haloperidol (a D2-dopamine receptor antagonist) 6 have all been suggested to affect brain activity during repeated painful stimulation.The effect of haloperidol is of specific interest because it suggests that especially the dopaminergic network is involved in the process of short-term habituation. 6The results of this randomized, double-blind, within-subject fMRI study showed that, only when participants received the placebo, they habituated (over 2 runs of 3 different stimulus intensities), as indicated by decreased activity in the left postcentral gyrus and left MCC.This habituation effect was not found when those participants received haloperidol.In fMRI studies investigating the effect of a nocebo manipulation, the nocebo group showed increased brain activity over time in the right operculum (although uncorrected for multiple comparisons). 78Similarly, another study reported increased activity over time in the opercular cortex for the nocebo group. 25In conclusion, a nocebo manipulation has shown to affect the habituation process both on a self-report and on a neural level.The dopaminergic network is suggested to underlie habituation to pain because of the habituation-blocking effect of a dopamine antagonist.

Habituation to pain: association between self-report and electroencephalography/functional magnetic resonance imaging
Although about half of the EEG and fMRI studies also included pain ratings (with sometimes opposing effects between neural and self-reported measures), only a limited number of studies directly tested this association.For fMRI, Ellerbrock et al. 25 reported that ratings were positively related to activity in the insula and periaqueductal gray (Table 4).For EEG, Greffrath et al. 32 reported that pain ratings and the N2-P2 amplitude were positively related, but Matre et al. 61 found no relation between the 2 measures (Table 3).A positive relation was reported between a decrease in pain ratings and decrease in the N2-P2 amplitude, thereby specifically testing the habituation to pain effect. 41EEG using multilevel modelling reported both positive and negative associations between pain ratings and EEG amplitude. 105Moreover, this association was found to decrease with a higher trial number, that is, habituation affects the association. 105The preliminary results on this topic thus suggest an association between neural habituation (as measured with EEG) and self-report ratings.However, it should be noted that most studies did not measure behavioral and neural habituation concurrently or showed opposing (but not statistically tested) effects.Clearly, this issue needs further in-depth investigation and remains an important topic for follow-up studies.

Discussion
In this scoping review, we included 63 studies on habituation to pain through measures of self-report or brain activation (EEG and/or fMRI).We found a large variety in methods, experimental settings, and contexts, suggesting that habituation is a ubiquitous phenomenon.Self-report studies (section 3.2) have shown a large influence of expectations and stimulation site, as well as the presence of individual differences in habituation to pain in healthy individuals.EEG (section 3.3) and fMRI (section 3.4) studies have shown characteristics of neural habituation to pain such as involvement of the N2-P2 amplitude and several identified brain areas (cingulate cortex and somatosensory cortices).Furthermore, the measured timescale showed different effects of habituation (eg, within run or across), and it is important to further develop and use statistical methodology to improve analyses and interpretations of these effects.
The majority of studies have used heat stimuli with a thermode, followed by heat using a laser and electrical stimuli (Fig. 3).9 and Supplementary Table (available at http://links.lww.com/PAIN/B920).For long-term habituation (lower 2 rows), spheres indicate areas in which activity decreased over time, except for the rACC/ sgACC, which showed increased activity (column 4, row 4).
Although the effects of heat stimuli with a thermode as compared with heat stimuli using a laser and electrical stimulation have not been systematically compared, all types of stimulation have shown to result in habituation to pain.Peripheral effects might occur more often when using thermode or laser heat stimuli on the same site, with short intervals or with intraepidermal electrical stimulation. 32,47,65In one such study, it was shown that both the N2-P2 amplitude and pain ratings decreased, suggesting that these peripheral effects are also reflected in brain activity. 32urthermore, habituation effects shown after changing the stimulation site suggest the involvement of central processes in habituation.However, results from self-report studies were inconclusive. 28,76This inconsistency suggests that peripheral and central effects related to habituation are likely intertwined.The underlying mechanisms of habituation to pain are still under investigation.Although at the micro-level habituation has been suggested to involve a decrease in cell excitability or homosynaptic depression of excitatory neurotransmission, 62 it is still unclear how this translates to the macro-scale effects.Besides, habituation studies often show similar outcomes as compared with other endogenous pain modulations, such as CPM and/or DNIC 24,97 or conditioning using low-frequency stimulation. 35,50urthermore, there is preliminary evidence for the involvement of the dopamine system. 6In addition, several specific cortical areas were reported to be involved in habituation to pain, particularly the cingulate cortex, insula, and somatosensory cortices. 68The EEG studies mostly reported effects on the N2-P2 amplitude.For LEPs, this N2-P2 effect has been hypothesized to originate from the cingulate cortex and S2. 29,100

Challenges in the field
This review highlights the diversity of approaches and variations in experimental settings.The sample sizes used in the various studies were relatively small, which, in combination with the variations in methodology, makes interpretation and the formulation of general conclusions difficult.As a result, the evidence related to the degree and/or presence of habituation to pain cannot be regarded as very robust.For example, population characteristics such as age and sex showed mixed results in selfreport (3.2.2) and EEG (3.3.2) measures.Conversely, expectations have shown robust effects on self-report ratings (3.2.3), either increasing or decreasing habituation effects.The latter indicates that the process of habituation is inherently subject to experimental settings such as the context and the given instructions, which should be taken into account when comparing studies.Future work could improve standardization across participants in terms of task context and expectations about habituation or sensitization.Furthermore, terminology is not always consistent.For example, "adaptation," "decay," and "habituation" have been interchangeably used in the literature to describe highly similar response decrements.Moreover, in the field of neuroimaging, "repetition suppression" has been used to indicate diminished neural activation. 5Therefore, it is important to have a uniform use of terminology.Similarly, distinctions between levels of habituation and/or sensitization should be clearly indicated.For example, when "diminished habituation" is reported, this could also indicate sensitization to pain (ie, when there is an increase with respect to the first stimulus) and should be described as such.
Another point of terminology is the concept of dishabituation, which is closely related yet distinct from habituation.Dishabituation is defined as "presentation of a different stimulus that results in an increase of the decremented response to the original stimulus" (Rankin et al., 2009,  p. 137). 75Dishabituation should not be confused with sensitization because dishabituation indicates an increase in a previously habituated response and not merely an increase.Although outside the scope of this review, it is important to note that several studies have investigated dishabituation.For example, within repeated triplets of stimuli with 1 second of interstimulus interval, a decrease was shown in the N2 and P2 peak, but changing the modality or increasing the stimulus intensity of the third pulse induced an increase in amplitude (ie, dishabituation); however, a change in spatial location did not affect the amplitude. 44,79,96,99There are also studies using other paradigms that discuss dishabituation.For example, Scheuren et al. 85 included an intervention to induce secondary hyperalgesia, and this has shown to result in dishabituation of the original evoked potential.Thus, various paradigms have been developed to increase our understanding of dishabituation.
Results and paradigms are also often named after the intended effect (eg, habituation paradigm).However, it is likely that there is no specific paradigm to test for either habituation or sensitization because these processes largely overlap (with the exception of temporal summation, although even there, habituation effects have also been shown; see Edwards and Fillingim 21 ).The results of this review indeed show that the few studies that report individual differences include a large variability over time with sensitization instead of habituation effects in some participants.This might be in line with the dual process theory (see introduction Section 1.0), stating that habituation and sensitization occur independently but may interact and then result in a behavioral response. 34Furthermore, a low stimulus intensity has been shown to result more often in habituation as compared with sensitization. 18,47This phenomenon may result from the lower threat it poses to the body. 15However, in some cases, the rate of habituation is similar for different intensities, and only the amplitude is increased for higher stimulus intensities (ie, level of the rating). 28,109Therefore, it remains unclear whether habituation and sensitization are a personspecific "trait" and stable under the same experimental settings or whether the influence of expectations, mood, and social context results in a "state" effect.
The inclusion of ratings as a measure of habituation to pain remains a topic of debate in the field.Including ratings after each trial might increase attention to the pain stimuli; however, including ratings only at the end of a block or experiment might lead to biased ratings through the habituation effects itself (eg, habituation occurred, and because most recent information is recalled ("recency effect"), lower ratings are reported).
The definition of habituation states "a decrease in response after repeated stimulation" but does not require any minimal or maximum number of repetitions.This results in a great variety in the number of repetitions being used in the field (see also Fig. 2), and it can be questioned whether 2 repetitions are sufficient to describe a habituation effect.Habituation effects also differed across timescales.Thus, in the case of averaging trials, this might mask effects on shorter timescales.Figure 5 shows an example, which indicates individual data where there is an increase of the pain intensity ratings over trials within a run, but when averaged, only a decrease across runs is visible.As for now, ratings per trial are encouraged because they provide more insight into the trajectories of habituation and sensitization to pain.Furthermore, multilevel models support including all trials and random effects for intercepts and slopes, thereby allowing for individual changes over time.Moreover, a multilevel model can accommodate effects both within and across runs.Therefore, this type of modeling is encouraged when analyzing effects over time.From our review, we conclude that it has been challenging to align self-report ratings with neural habituation.Conceptually, it is unclear whether habituation to pain on the neural level (ie, a decrease in activity or amplitude) is sufficient to indicate habituation effects or whether self-report ratings are needed.A decrease in brain activity may be related to several underlying neural mechanisms, but these mechanisms not necessarily need to be related to habituation to pain.For example, Paul et al. 68 demonstrated a decrease in fMRI BOLD activity in the cortical areas MCC, IFG, and S2 but reported no decrease in pain ratings.Interpreting the observed decrease in activity as habituation to pain 68 thus remains a topic of further discussion.Moreover, changes in neural processes (eg, brain activity) may be better described as nociception instead of pain. 1 Although most literature described habituation to pain, this remains an important challenge in the field with respect to EEG and fMRI research.Our review further shows that effects are often not aligned, in particular with respect to the direction of effects based on neural (EEG, fMRI) or self-report measures (see Section 3.5).Future studies will help to determine whether there are similar processes underlying self-report and neural habituation or that these 2 levels are separate forms of habituation to pain.Christoffersen 13 already stated very early on that "at the least, it may serve the purpose of avoiding conceptual confusion if habituation at the 2 levels were named separately until a causative link between these levels has been proven to exist in each case" (pp.61-62).

Future recommendations
Based on the current state of the field, the gaps in the literature, and the challenges in the field, we propose the following recommendations for future research.
First of all, with respect to terminology, studies should clearly describe the measured effects.Care should be taken with comparison of different operationalizations of habituation, such as measuring the pain threshold, ratings, skin conductance, or neural measures.It is important to specifically describe which type of measurement a study used, without making a conclusion based on several measurements.For example, there might be habituation of the pain threshold (which involves a process of tolerance) but not of pain stimuli above the threshold.Therefore, we recommend clearly describing the specific effect, for example, habituation to pain intensity ratings.
Second, care should be taken when discussing and comparing different timescales.Habituation was reported to differ based on various timescales and should be described as such, for example, within and across run effects.Within-run effects are more subject to peripheral habituation, especially when the same location is stimulated and/or if a break between the runs is present.Across run effects might be more robust and indicate central processes.However, habituation is also known to become more pronounced over series (also called potentiation). 75In addition, studies using long-term timescales demand a different involvement of participants and show habituation effects resulting from longer stimulation periods.
Third, large individual differences exist with respect to habituation to pain.Thus, studies should focus more on individual characteristics, which may lead to increased personalized approaches and treatments.Moreover, individual differences will lead to more insight into the expected effect, for example, whether a certain paradigm elicits habituation in each participant or only in a subset of participants.
Fourth, the use of individual calibration of the pain stimulus is encouraged, especially with self-report or the combination of selfreport with EEG or fMRI (to ensure a standard procedure across modalities).With calibration differences in skin parameters and pain receptors taken into account (ie, limiting peripheral effects), differences in experienced pain are reduced (ie, reducing floor or ceiling effects where pain is too high for some participants), and the intensity of the stimulus may affect the likelihood of habituation and sensitization.Although a recent large study did not report a relation between perceived pain intensity and brain activation, 40 calibration to individual thresholds may be very important for habituation and sensitization processes and the study of their neural correlates.
Fifth, because habituation may occur under many circumstances and contexts, it is of importance that studies involving repeated painful stimulation, test for effects over time or repetition and include this as a covariate in the statistical model even if the investigation of habituation or sensitization is not the primary focus.
A sixth, more generic, recommendation concerns the use of larger participant groups with preregistration and sharing of analyses/code.The present review clearly shows that most studies are based on a relatively low sample size (Fig. 2B).Large sample sizes will help in generalization of effects and identifying individual differences, whereas preregistration will make the expectations about habituation and/or sensitization to pain more transparent and will increase confirmatory analysis approaches. 37

Limitations of the review
In this review, we did not include studies using behavioral measures such as reflexes, habituation of the pain threshold, and habituation of nonpainful stimuli, although the literature using these measures is substantial.Moreover, we deviated from the preregistered protocol by not including the literature on habituation to pain in chronic pain patients.This was decided because there was abundant literature to describe self-report and neural measures of habituation first in healthy individuals, whereas studies on habituation to pain in chronic pain patients have formed the fundament of a separate review. 102urthermore, studies where sensitization to pain was the main outcome were excluded.The present review clarifies that further investigation is needed to highlight the distinctions between habituation and sensitization to pain and the possible underlying individual characteristics.
Another limitation is that effects of habituation to pain are not always described uniformly (eg, habituation, adaptation, decay or simply decrease in pain), potentially leading to some missed articles.Furthermore, a few articles quantified a decrease in pain but did not describe (or define) it as habituation.We decided to not make interpretations beyond those of the original articles, which then may result in not including these articles.This also applies to the use of the term pain instead of nociception.In the case of EEG and fMRI studies, the term habituation to nociceptive stimuli might have been more applicable.However, because this issue is outside the scope of the current review, the descriptions as used by the authors of the original papers were used (ie, habituation to pain/painful stimuli).In addition, some studies might have examined habituation effects but did not find and/or report this in the article (ie, null effects).

Conclusions
The results from this scoping review show that habituation to pain is a fundamental process that occurs under a variety of circumstances and is expressed both with self-reported and neural measures.Habituation to pain is present on very short (few stimuli) to long (several days) timescales, and the context (eg, expectations) and experimental settings (eg, stimulus intensity, stimulation site) can influence the degree of habituation based on self-reported and/or neural (EEG, fMRI) measures.
Future well-designed studies should address individual differences, population characteristics, and the relation between neural and self-report measures.In addition, the discrepancy between habituation and sensitization should be further verified, accompanied by consistent and clear use of terminology.
As of now, the insights from this review on habituation to pain in healthy individuals can be used to optimize designs and methods for measuring habituation to pain, and this will aid our understanding of the underlying mechanism.Individualized approaches can be used for exploring pain modulation in chronic pain patients.
Treister et al. (2010) compared different modes of endogenous pain modulation, namely, habituation and diffuse noxious inhibitory control (DNIC, see Table

Figure 3 .
Figure 3. Type of stimulation used in studies on habituation to pain and evaluated based on self-report (left), EEG (middle), or fMRI (right).

Figure 4 .
Figure 4. Overview of fMRI studies involving habituation to pain projected on an MNI brain.Centers of peak activity are visualized as 5-mm spheres.Only data corrected for multiple comparisons are included, and we did not include between-group analyses for factors of influence.For short-term habituation (upper 2 rows), spheres represent decreasing activity over time.Several timescales are included, which can be found in Table9and Supplementary Table (available at http://links.lww.com/PAIN/B920).For long-term habituation (lower 2 rows), spheres indicate areas in which activity decreased over time, except for the rACC/ sgACC, which showed increased activity (column 4, row 4).

Figure 5 .
Figure 5.In panel A, individual pain intensity ratings are visualized that show an increase over trials within each run (sensitization).In panel B, the averages (grey diamond) of the same trials per run are visualized, which instead show a decreasing trend across runs (habituation).

Table 2 (
continued) -report results as part of EEG or fMRI studies are included in Tables3 and 4. The number of stimuli is reported for the habituation measurement and not necessarily referring to the total number of stimuli used in the study.If participants received 2 conditions with each 10 stimuli and habituation is calculated per condition, 10 stimuli are reported for the habituation analysis. Self

Table 3
Habituation to pain: EEG studies.

Table 4
Habituation to pain: fMRI studies.

Table 4 (
continued) * Studies that included self-report.The number of stimuli is reported for the habituation measurement and not necessarily referring to the total number of stimuli used in the study.If participants received 2 conditions with each 10 stimuli and habituation is calculated per condition, 10 stimuli are reported for the habituation analysis.ACC, anterior 1cingulate cortex; BOLD, blood-oxygen-level-dependent; IFG, inferior frontal gyrus; MCC, midcingulate cortex; PAG, periaqueductal gray; S1, primary somatosensory cortex; S2, secondary somatosensory cortex.