In alcohol dependence, differences in brain responses were seen in tasks of facial emotion processing. Blunted ACC activity was observed in recently detoxified alcohol-dependent individuals in response to both fear and disgust compared with controls [17]. In line with this, higher rACC activation in recently detoxified alcohol-dependent patients in response to aversive faces vs neutral cues was associated with better treatment outcomes [18]. In the same study, blunted activation was also observed in the fusiform gyrus [18]. Lower prefrontal and left anterior insula activation, and less hippocampal connectivity, was observed in response to happy, sad, and angry faces in alcohol-dependent individuals during both early and sustained remission compared with controls [19]. Therefore, it is possible that blunting of brain activity in response to facial emotions may persist into longer-term abstinence and recovery.

In line with this, recently abstinent alcohol-dependent individuals compared with controls demonstrated deficits in recognising fearful faces in conjunction with lower insula and orbitofrontal cortex (OFC) activity [20]. In the same study, alcohol-dependent participants with a history of repeated detoxifications demonstrated lower insula and OFC activation and insula-amygdala connectivity compared with those who had fewer detoxification attempts [20]. Lower amygdala and hippocampal activation was also observed in response to emotional faces relative to neutral faces in recently detoxified long-term alcohol-dependent men, without the same repeated detoxification history, compared with controls [21]. However, another study found no differences in amygdala activity in response to negative emotional faces between recently detoxified inpatient alcohol-dependent men and controls [17].

Blunted ACC and mPFC activity were observed in recently detoxified alcohol-dependent individuals in response to anticipatory anxiety and threat [22]. Similar blunting in ACC, midcingulate cortex (MCC) and posterior cingulate cortex (PCC) connectivity was seen in response to individualised stressful experience cues in recently detoxified alcohol-dependent individuals compared with healthy controls [23]. Blunted activation was also observed following exposure to individualised stressful experience cues relative to neutral in the vmPFC and ACC in recently-detoxified alcohol-dependent individuals [24]. This higher activation in the vmPFC and ACC during the neutral condition was correlated with high levels of stress-induced craving in the recently detoxified alcohol-dependent group [24]. Lower activation in the right amygdala was observed in a recently detoxified alcohol-dependent population compared with healthy controls in response to aversive images [25]. Right amygdala activity was also negatively correlated with severity of alcohol abuse [25]. Recently detoxified alcohol-dependent individuals demonstrated blunted activation in the posterior insula compared with controls whilst focusing on, i.e. ‘sustaining’, emotional images compared with neutral images in an emotion reappraisal task [26]. However, in response to high-arousal negative images in the absence of alcohol cues, recently detoxified alcohol-dependent individuals demonstrated heightened activation in the bilateral insula and amygdala compared with controls [27], in contrast to the aforementioned studies which mostly demonstrated blunted activity.

Together, these findings suggest that alcohol-dependent individuals demonstrate blunted activation to negative affective stimuli and negative emotional faces compared with controls. In response to negative emotional faces, alcohol-dependent individuals showed blunted activation in the ACC, insula, fusiform gyrus, hippocampus, OFC, and amygdala compared with controls, and reduced amygdala-insula and hippocampal connectivity. In line with this, the ACC, insula, and fusiform gyrus have previously been shown to be involved in facial emotion processing [28, 29]. Therefore, blunted responses in these regions may be associated with deficits in facial emotion recognition, and subsequent poorer treatment outcomes [18]. In response to aversive images or cues, alcohol-dependent individuals also demonstrated overall blunted cingulate connectivity, and blunted activation in the ACC, insula, mPFC, vmPFC, and amygdala. However, some results were conflicting, as heightened activation was also observed in the insula and amygdala. Overall, these results suggest that alcohol dependence may be associated with blunted corticolimbic activity, as the hypoactivation in these regions may cause deficits in facial emotion recognition and processing of negative emotional stimuli. Further studies should clarify whether deficits in facial emotion recognition only result in diminished recognition capabilities or whether this may lead to misinterpretation of emotions, which could have social implications. Furthermore, it is challenging to account for the impact of variation in abstinence length on brain activity during the tasks (Table 1), which could be more clearly elucidated through longitudinal studies.

Although amygdala hyperactivation is commonly demonstrated in neuroimaging studies investigating emotional processing in negative affective disorders [43, 44], this review found inconsistent amygdala activation in substance dependence. Amygdala dysregulation was reliably observed in the reviewed studies in alcohol dependence, although inconsistent in direction. Differential amygdala activation was observed between groups in response to negative images [27], but not to emotional faces [17, 18, 20] or high threat stimuli [22]. Together, these findings suggest that the type of emotional stimulus may affect the direction of the amygdala reactivity in those with alcohol dependence. In contrast, in cocaine dependence, amygdala hyperactivity was only observed in Potenza, Hong [31], suggesting other regions may be more centrally involved in the contribution to the emotional dysregulation and negative reinforcement observed in cocaine dependence. In Potenza, Hong [31], amygdala hyperactivity was observed in women compared with men, suggesting women with cocaine dependence may be more susceptible to amygdala dysregulation following the experience of stress. However, more studies are required to elucidate the impact of sex differences in cocaine dependence. In opioid dependence, dysregulation of amygdala activity was reliably observed across the reviewed studies. However, the nature of the dysregulation was inconsistent, as higher activity and connectivity were observed in response to emotional faces during heroin treatment [36, 37], but not in response to negative images during abstinence [39], or the emotional oddball task during suboxone treatment [38]. Therefore, the divergent results may be attributable to the varying abstinence statuses of the participants. As previously discussed, longer-term abstinence may help attenuate or normalise dysregulated brain reactivity to negative affective stimuli. Treatment with opioid agonists like heroin, such as in Schmidt, Borgwardt [36], Schmidt, Walter [37], or other OST, are also likely to have inhibitory effects in the amygdala due to the high expression of inhibitory mu opioid receptors. Further, the variety of tasks used are likely to engage different cognitive processes and thus recruit amygdala activity in differing ways.

Hypoactivity in the amygdala could be attributable to down-regulation by modulatory ACC/mPFC networks during tasks requiring more cognitive effort [45]. For example, there was no amygdala activation during emotion reappraisal in alcohol dependence [26]. This is in line with neuroimaging research, which demonstrates heightened activity in prefrontal areas alongside reduced activity in the amygdala [46]. As prefrontal regions are involved in emotion regulation and the amygdala is considered emotion generating, lack of amygdala activity during an emotion regulation task is expected. Additionally, blunted amygdala responses may be due to habituation following repeated exposure of participants to negative emotional stimuli [16, 47], such as in Wang, Zhang [39]. However, in people with alcohol use disorder and adverse childhood experiences, a lack of habituation in the amygdala was observed in response to negative emotional faces [48]. The results suggest repeated exposure could cause sensitisation to negative affective stimuli, which may be mediated by neuroplastic alterations in the amygdala [49]. These changes may perpetuate the dysphoria and anhedonia that drive negative reinforcement [49].

The ACC is a key region implicated in addiction, as the neural networks mediating emotion regulation and cognitive control are dysregulated. In the studies currently reviewed, individuals with substance dependence generally demonstrated dampened ACC activity during tasks of emotional processing and regulation [17, 22–24, 32, 34, 38]. Together, these findings suggest that regulatory control governed by the ACC may be compromised in dependence, resulting in impaired emotional regulation. Impaired top-down control of emotion generating regions such as the amygdala may manifest as heightened impulsivity and craving. This effect likely occurs following exposure to conditioned cues including emotional states like stress, which subsequently triggers craving characteristic of the preoccupation/anticipation stage [3]. In line with this, reduced ACC activity is observed during tasks of attention and inhibitory control in users of cocaine, heroin, alcohol and cannabis [14]. This impairment in inhibitory control subsequently heightens impulsivity and contributes to relapse [3]. Modulating the rACC, considered the emotional hub of the ACC, may prevent craving caused by emotional states, by ameliorating hypohedonia or negative affect, thus preventing the transition to the preoccupation/anticipation stage [12].

Multiple roles governed by the PFC are dysregulated in dependence during negative emotional processing and contribute to the negative reinforcement observed in addiction. As previously described, in alcohol dependence, lower activation was observed compared with controls in the dlPFC, mPFC and anterior portions of the PFC in response to emotional faces [19], and in the OFC in response to fearful faces [20]. These subregions are collectively involved in emotion regulation and cognitive control [50], characterised by increased activity in response to negative affective stimuli to downregulate emotion generating regions like the amygdala [51]. Given its role in cognitive appraisal, diminished activity in the dlPFC suggests impairments in evaluating and interpreting emotional facial stimuli. Indeed, reduced activity in the dlPFC is associated with impairments in negative facial emotion processing in people with mood disorders [52], suggesting similar patterns of emotion dysregulation in people with alcohol dependence. Connections of the OFC and mPFC to the amygdala are known for modulating social cognition and emotional responses [53, 54]. Thus, lower activity in the mPFC and OFC in alcohol dependence may indicate impairments in the ability to process emotional faces due to their aberrant connectivity with the amygdala [55, 56]. Lower vmPFC activity was observed in response to stress imagery, which was predictive of heavier drinking after relapse [24]. In line with this, blunted vmPFC response to stress in young healthy adults was also associated with increased binge drinking and emotional dysregulation [57]. Given its role in integrating affective valuations made by regions like the amygdala [58], diminished activity in the vmPFC is indicative of more dysregulated emotional processing, which may confer greater risk of relapse, heavy drinking, and treatment failure [24, 59, 60].

In opioid dependence, increased amygdala-OFC connectivity was observed under placebo in heroin-maintained dependent individuals compared with controls in response to fearful faces, which was reduced by heroin treatment [61]. Amygdala-OFC connectivity is a known connection mediating anxiety responses [53, 54], suggesting that heroin dependence may disrupt the normal connectivity of this response and increase anxiety responses to fearful faces in the absence of heroin, which is restored following heroin treatment [61]. Similarly, amygdala-OFC connectivity and OFC activity was heightened in cannabis dependence in response to negative affective images [41], and reduced mPFC activity was also observed in long-term heavy cannabis users during evaluation of emotional stimuli [42]. The patterns of emotional dysregulation in cannabis dependence are similar to that of alcohol dependence, suggesting cannabis and alcohol may recruit similar neural pathways that subsequently become dysregulated with repeated use.

In cocaine dependence, higher activation in the dlPFC and higher dlPFC-amygdala connectivity was observed during reappraisal [30], whilst hyperactivation was observed in the dmPFC, dlPFC, vmPFC, and vlPFC in in response to stress imagery [31]. These results suggest that people with cocaine dependence demonstrate increased sensitisation to negative emotions compared with controls and subsequent dysregulation of the associated emotional state [30]. This is in contrast to alcohol dependence, where prefrontal activity was blunted in response to negative affective stimuli and facial emotions. Interestingly, in polydrug users with primary cocaine dependence, activity was lower in the mPFC in response high arousal negative and positive images compared with controls [34]. As prefrontal activity diminished in polydrug dependence, it is possible that heightened prefrontal activity is facilitated by purer cocaine use, whilst the dampening of its activity may be due to additional drugs like alcohol and cannabis. Together, the results suggest that different drugs of dependence may lead to variance in PFC functioning, disrupting a variety of emotional regulation processes.

In anxiety disorders, the role of the insula in perceiving the internal state of feeling anxious can become maladaptive and manifest as heightened subjective anxiety [10, 62]. However, this effect is less consistently observed within substance dependence. The results of the studies reviewed showed inconsistent patterns of insula reactivity and connectivity during negative emotional processing. Insula activity was heightened during tasks involving passive viewing of aversive stimuli in alcohol [27, 31, 40], and dampened during tasks of reappraisal in cocaine [26, 30]. During emotion reappraisal in the healthy population, the insula consistently demonstrates reduced activity [16, 46, 63], which suggests downregulation by regions like the ACC and mPFC [16, 45, 63]. Across all substances, reactivity of the insula was consistent depending on the type of task, i.e. heightening of insula reactivity during passive viewing of aversive stimuli and dampening of insula reactivity during reappraisal. This suggests that the type of task used, and the subsequent cognitive process employed, may be the driving factor in modulating insula activity, rather than the prolonged use of the drug of dependence.

Tasks of facial emotion processing showed different patterns of insula activity in alcohol dependence [17, 19, 20]. In response to happy, sad and angry faces, anterior insula activity was lower compared with controls in Alba-Ferrara, Muller-Oehring [19], but higher in Salloum, Ramchandani [17]. In response to fearful faces, anterior insula activity was lower in O’Daly, Trick [20] but higher in Salloum, Ramchandani [17]. Additionally, anterior insula activity was lower in response to disgust [17]. Given the differences in reactivity to the same emotions, it is possible these differences can be attributed to abstinence length of the participants. Insula activity occurs following homeostatic changes, such as during withdrawal from drugs. Therefore, the length of time the participants were abstinent likely affects insula activity during emotion processing. It is likely that the longer the participants were abstinent, the less hyperactive the insula would be during tasks of negative emotional processing. This is in line with the reviewed studies, as participants in Alba-Ferrara, Muller-Oehring [19] were either in early remission (<12 months) or late remission (>12 months), compared with Salloum, Ramchandani [17] and O’Daly, Trick [20], where all the participants were in early abstinence for <4 weeks. As early remission was considered up to 12 months in Alba-Ferrara, Muller-Oehring [19], but early abstinence was considered up to 4 weeks in Salloum, Ramchandani [17] and O’Daly, Trick [20], participants were likely abstinent for considerably longer in Alba-Ferrara, Muller-Oehring [19]. Therefore, reduced insula hyperactivity following exposure to facial emotions, particularly negative emotions, during longer periods of remission may indicate therapeutic benefit by reducing the negative subjective experience of feelings, which may help prevent craving and relapse.

In the studies reviewed, the patterns of reactivity following negative emotional processing were variable amongst the discussed brain regions and substances. Blunting in activation was broadly seen in alcohol dependence in response to emotional faces and affective images. This may suggest that emotional networks that are recruited whilst processing affective stimuli may subsequently become dysregulated during alcohol dependence. In cocaine dependence, the results were also variable but broadly demonstrated heightened activation in response to stressful and aversive stimuli and blunted activation in response to positive affective stimuli. This is in line with previous evidence demonstrating blunted responding to natural rewards in cocaine addiction [14, 64, 65]. Patterns in brain reactivity were also variable in the studies investigating opioid dependence, although the amygdala was consistently implicated. Fewer conclusions can be drawn regarding cannabis dependence due to the lower number of studies. Although, regions that were implicated in other studies like the mPFC, anterior insula and ACC were also dysregulated in cannabis dependence. Together, the findings demonstrate alterations in activity and connectivity of these regions, which may result in impairments to negative emotional processing in drug dependence. This effect could be due to pre-existing maladaptations or functional adaptions following repeated drug exposure, or a combination of both, although is difficult to ascertain without longitudinal studies. Results following the completion of the ABCD and IMAGEN studies will hopefully shed light on the causality of developing substance use and addictive behaviour [66, 67]. Together, the results indicate that there may not be a broad level mechanism of dysfunction across substance dependences, but rather each class causes dysregulation in diverging ways.

Given the lack of neuroimaging research on emotional dysregulation in the substance dependence field, conclusively characterising brain reactivity patterns during negative emotional processing remains a challenge. Particularly, differences in study design, statistics and participant demographics contribute variability to the results, making interpretation increasingly challenging. Brain regions with significant anatomical overlap and indistinct boundaries may also present challenges when attributing dysfunction to a specific region. The ambiguity between boundaries increases potential for misinterpreting findings, making it difficult to attribute the results within the context of one specific region. This may lend itself to interpreter bias, whereby the effect may be interpreted in line with the hypothesis or current trends in research. The best way to account for this is with Activation Likelihood Estimation (ALE) meta-analyses, which can estimate the probability of overlap between activations across various studies [68]. However, given there are insufficient studies in this domain to make ALE possible, whilst the field grows, it is important to ensure all data being published is transparent moving forward. Following new results being published in the future, previously published data can be more easily re-evaluated and re-interpreted considering the new findings and be utilised in quantitative meta-analyses. In particular, whilst data in the field remains relatively sparse, a Bayesian approach may be appropriate as it can combine prior knowledge with small or heterogenous study results, reducing uncertainty and stabilising effect sizes.

Since the first studies, the methodology and statistical approaches used in fMRI have developed hugely, resulting in more caution about conclusions from earlier studies. Statistical methods commonly used in early fMRI statistical packages were found to allow substantially higher false positive rates than initially thought [69]. Some of the earlier studies in this review may not have accounted for multiple comparisons (Table 5), increasing the likelihood of erroneous interpretations. Some statistics in early fMRI studies were uncorrected, and instead the significance threshold was increased to help prevent false positives. However, this is less effective than FWE or FDR, which are more powerful at controlling for false positives when conducting multiple hypothesis tests [70]. Furthermore, studies using whole brain analyses only (see Table 1) are subject to increased variance in the results. Using a priori ROI hypothesis-driven approaches may help mitigate the problems associated with greater variance at the whole brain level. Studies using fMRI are typically underpowered due to small sample sizes. Increasing sample sizes can help improve confidence in results by reducing the likelihood of both type I and type II errors [71]. Recently, sample sizes have been recommended to be in the thousands to improve replicability and decrease effect size inflation [72]. However, increasing sample size is often extremely challenging in addiction research.Table 5

Overview of statistical thresholding and anatomical normalisations.

A critical factor contributing variability in results is sex differences, as there are broad neurobiological differences between males and females [73]. Across the substance dependence literature, the majority of study participants are typically men. This prevents the results from being accurately attributed to how addiction manifests in females. In the reviewed studies, only Potenza, Hong [31] and Canterberry, Peltier [34] addressed sex differences in cocaine dependence. In Potenza, Hong [31], hyperactivity was observed in corticostriatal-limbic circuitry in women compared with men following exposure to stress cues. Meanwhile, no differences between sexes were observed in Canterberry, Peltier [34] in response to negative images. It is possible that the divergence in findings may be explained by co-occurring drug dependencies, as dependence on alcohol and cannabis were allowed in Potenza, Hong [31] and Canterberry, Peltier [34] respectively. Together, these discrepancies highlight the need to better address the impact of sex differences on emotional dysregulation in substance dependence, as well as comorbid polydrug dependencies.

It is rare that an individual is dependent on one substance, particularly if nicotine is included, but such ‘polydrug’ dependence is rarely acknowledged. Nicotine dependence is particularly common in substance dependences and so is often allowed in studies. This makes it particularly challenging to disentangle the effect of nicotine relative to the substance being studied. Further, the majority will meet criteria for other psychiatric diagnoses, such as anxiety, depression, post-traumatic stress disorder and personality disorders [74]. As these conditions share similar but also distinct neural networks, attributing the results to a particular condition is incongruous [44]. Individuals with substance use disorders and comorbid psychiatric disorders demonstrate alterations in brain regions associated with reward processing, emotional regulation, and cognitive control [3]. However, these brain changes can vary depending on the specific combination of disorders, as well as the severity of symptoms and the stage of addiction. Additionally, the frequency of comorbid substance dependences and psychiatric diagnoses creates a recruitment challenge, as recruiting a ‘clean’ sample of individuals without current or past psychiatric disorders or substance dependencies is extremely difficult. Together, these challenges highlight the importance of carefully characterising the study population and accounting for the complexity of polydrug use and psychiatric comorbidity when designing and interpreting research studies.

As previously discussed, differences in abstinence length of the participants likely contributes to the disparity in findings across studies. During early abstinence, within days to a few weeks, the brain initially adapts to the lack of the substance and is more hyperreactive to negative affective stimuli, manifesting as increases in anxiety, emotional distress, craving and irritability, as it rebalances without the influence of substance [75]. Heightened reactivity to negative affective stimuli may intensify the feelings of negative emotions, culminating as heightened stress or anxiety responses [76]. As abstinence continues into weeks and months, neural circuitry associated with emotion processing may begin to stabilise, which may be accompanied by reduced craving and reactivity to stressful triggers. Longer-term abstinence, from months to years, may facilitate further recovery of brain reactivity to negative stimuli [77]. However, long-term neuroadaptations can persist into protracted withdrawal [78]. However, individual variation must be considered, due to factors such as dependence duration and severity, co-morbid psychiatric diagnoses, and concomitant behavioural intervention [79].

To further improve the reliability of findings in task-based fMRI, we should standardise the tasks being used, as this will help improve replicability [16]. Many studies used different tasks, falling into different categories regarding the paradigms used (Table 2), making comparisons in results between task categories difficult. Further, even studies using similar tasks but with small differences (Table 3), can produce profound differences in brain activation (Table 4). Small differences in task design may recruit different neural processes and engage different emotion networks [41], making it challenging to compare studies with each other directly. From a psychological perspective, the tasks utilised only measure singular aspects of emotion processing, which is a more complex construct. Given the variation in task categories and individual design within categories (Table 3), the differences between tasks likely mean that various processes are being measured that have neural underpinnings that are distinct from one another. Further, with other variables contributing to heterogenous results, such as sex, abstinence length and co-morbid diagnoses, using consistent task designs should improve replicability. Whilst the IAPS is good for standardising images used within studies examining affect, further standardisation in emotional processing task design is necessary. A shift towards consistency through the utilisation of reporting checklists [80], akin to what is observed in the cue reactivity literature with the formation of the ENIGMA Addiction Cue Reactivity Initiative [81, 82], would help improve reliability and replicability of results. A similar initiative should be developed within the field of emotional processing. A collaborative network like ENIGMA would help define and validate standardised tasks, which can be used consistently across different studies and research groups and ensure their sensitivity to measuring the relevant cognitive and behavioural processes. Alongside providing a centralised database where researchers can access the standardised tasks, together, this approach would help facilitate replication and comparison of findings in the field of emotional processing in substance dependence going forward.

In addition to standardising tasks, standardising terminology across the wider emotional processing field would greatly impact the interpretation of results. In the emotional processing field, terms such as ‘emotion’, ‘affect’ and ‘mood’ are often used interchangeably, although they refer to distinct psychological constructs [83]. The DSM-5 characterises ‘affect’ as observable expressions and objective measures of emotion, and ‘mood’ as a more prolonged and sustained emotional state [84], whilst ‘emotion’ generally refers to an immediate and transient response to a stimulus [85]. However, these terms are used inconsistently across the field, which may influence how findings are conceptualised, compared across studies, and applied in clinical context. For example, inconsistent use of terminology may hinder the synthesis of results in meta-analyses or systematic reviews. Without clear definitions for these terms, studies may not be investigating the same emotional constructs that they appear to be, which may subsequently compromise the ability to accurately interpret the findings and draw conclusions. Translating this into the clinic presents more challenges, as the subjective experience of the patient may be the same across different psychological constructs. For example, stress and anxiety are often conflated even though their neurobiological and psychological processes differ. Moving forward, precise terminology should be adopted across the emotional processing field, similar to Wellcome’s initiative for identifying and adopting common questionnaires in mental health research [86], in order to ensure that research outcomes are valid and that findings can be compared across studies.

Future studies should incorporate more techniques to modulate dysregulated brain processes in therapeutically beneficial ways. At the neural level, fMRI studies should include more pharmacological or psychological intervention. Incorporating these interventions into fMRI may help further elucidate biomarkers of emotional dysregulation by characterising whether they modulate brain reactivity in a manner consistent with therapeutic benefit. Protocols should carefully consider the timing of drug administration relative to the imaging session and have this kept consistent across studies.

Moreover, non-invasive brain stimulation (NIBS) techniques, including repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), and temporal interference (TI) stimulation have recently emerged as having potential clinical utility in treating substance dependence [87, 88]. In comparison to fMRI, which is primarily used as a tool to identify brain reactivity patterns and subsequent neural markers of psychiatric disorders, NIBS is advantageous as it can potentially modulate brain function in a therapeutic manner. Further, NIBS has been shown to modulate brain networks rather than individual regions, suggesting it may be useful for treating complex psychiatric disorders where multiple regions within a network are dysregulated [89]. Recent research has highlighted the dlPFC as an efficacious target region for substance dependence and may be a beneficial stimulation site across various substances [90]. rTMS of the dlPFC was shown to modulate emotion regulation during an emotion reappraisal task in alcohol use disorder [91]. Relative to sham procedure, rTMS reduced the emotional experience to positive and negative images in alcohol use disorder participants, whereas rTMS increased the emotional experience to neutral and positive images in healthy controls. The results suggest a possible role for rTMS modulating emotional states related to relapse, by reducing the negative emotional experience to negative images. However, since there were no associated reductions in subjective craving, further work is needed to characterise the effect of NIBS on emotion processing networks on clinically relevant measures of substance dependence.

However, whilst few studies exist examining the effect of NIBS on emotion regulation in substance dependence, more studies have been conducted investigating NIBS in cue reactivity paradigms [92]. Studies have demonstrated a personalised approach to NIBS, where individualised stimulation of deeper-brain targets in combination with individualised cue-reactivity paradigms have substantially decreased craving and substance use in heroin dependence [93, 94]. Recent advances in NIBS have advanced the ability to modulate deeper-brain structures more commonly associated with emotional circuitry, further enhancing the translational relevance of NIBS for substance dependence [93–95]. Further incorporation of NIBS with fMRI studies will help continue establishing the causal relationships between brain reactivity patterns and cognitive and behavioural processes. As a relatively new field, there are important limitations to consider, such as cost and accessibility [96], studies with small sample sizes and lack of controls, as well as methodological differences between studies resulting in inconsistent findings [87]. Nonetheless, the therapeutic potential of NIBS in modulating neural circuits that are implicated in addiction in an individualised manner is substantial and warrants further investigation to elucidate its treatment potential.