In this study, a total of 150 individuals with MUD were recruited from a compulsory drug rehabilitation center in China. Participants were required to meet the following inclusion criteria: (1) Meet the DSM-5 criteria for stimulant use disorder; (2) Have used MA at least twice a month for six consecutive months within the past 2 years (counting from the admission date to the rehabilitation center); (3) Be male, aged 18–55, and had normal color vision as confirmed by the Ishihara test; (4) Have an education level above primary school. The exclusion criteria were as follows: (1) have a history of polysubstance abuse (counting from the admission date to the rehabilitation center); (2) have a personal or family history of mental illness, or severe physical diseases.

This study was approved by the Ethics Review Committee of the Second Xiangya Hospital of Central South University in 2019, with the approval number (2019) Ethical Review [Science] No. (071). Each participant personally signed a paper-based informed consent form.

This was a cross-sectional study conducted from April 2019 to June 2019. Participants were screened in accordance with the inclusion and exclusion criteria, and their written informed consent was duly obtained. Subsequently, the researchers conducted interviews and questionnaire assessments with the participants. In a quiet and comfortable treatment room, the participants underwent a 10-min practice session with the VR neutral scenario. After that, during the formal experiment, the subjective craving scores of the participants in the standardized VR scenarios were collected. After the VR assessment, researchers interviewed each participant to understand their states after cue induction. Relaxation training and interventions were provided as needed. Finally, the participants were invited to perform the picture-based addiction Stroop task. The picture-based addiction Stroop task was administered at least 3 days after the VR assessment. To alleviate any potential distress following VR cue exposure, all participants received a standardized, brief (5-min) guided breathing relaxation exercise administered by a trained researcher. The exercise involved instructing participants to close their eyes and focus on taking slow, deep breaths. The same script was used for all participants.

SPSS 20.0 was used to organize and analyze the data. Descriptive statistics were employed to analyze the general information of the participants, their addiction assessment results, and psychological craving scores. For the attentional bias assessment, data were first screened: participants with less than 90% correct responses in the formal task or reaction times outside 3 standard deviations of the entire sample’s mean were excluded (42). The specific evaluation metrics were the reaction times (in milliseconds) for correct responses to MA-related and neutral pictures. The mean reaction time for proper reactions to each category of picture stimuli was calculated and analyzed with descriptive statistics. Paired-sample t-tests were used to compare reaction times for the two categories of picture stimuli. Pearson correlation analysis was used to correlate attentional bias with psychological craving in people with MUD. The attentional bias score was calculated as the difference between the correct reaction time to the MA-related pictures and that to the neutral pictures. Since the VR drug use scene induces stronger cue craving, in this study, the craving score under the cue of the VR drug use scene was used as the cue craving value for further analysis. For further exploring the relationships among attentional bias, withdrawal craving, and cue craving, the different craving level scores were categorized into three groups based on the distribution of craving scores (full spacing of 1–10 points) by taking the upper and lower percentiles (p33, p66). Thus, withdrawal craving scores equal to 1 were delineated as the no craving group (n = 39), 2–3 as the low craving group (n = 48), and 4–10 as the high craving group (n = 47). According to the VR drug use scenarios, craving scores equal to 1 were delineated as the no cue craving group (n = 32), 2–3 as the low cue craving group (n = 45), and 4–10 as the high cue craving group (n = 57).

A repeated measures ANOVA was used to compare the groups of participants with different levels of withdrawal craving and cue craving in terms of their responses to different stimulus pictures (neutral pictures, MA-related pictures). A one-way ANOVA was used to analyze the groups of participants with different levels of withdrawal craving and cue craving in terms of attention bias. A multiple linear regression analysis was used to identify the effect of cue craving and withdrawal craving on attentional bias. A 95% confidence interval was set, and all statistical tests were two-sided with a significance level of p < 0.05. The Pearson correlation analysis was subjected to Bonferroni correction for multiple comparisons, with the significance threshold set as α = 0.05/2 = 0.025.

In this study, we recruited a total of 150 participants with MUD. Of these, 137 completed the addiction Stroop task while 13 did not. Additionally, three participants were excluded due to their accuracy rates in the task below 90% (50, 88.75, and 88.75% respectively). Consequently, the valid data of 134 MUD participants were finally included in this study. Their sociodemographic characteristics and substance use information are summarized below.

This study used the addiction Stroop task paradigm to assess the attention bias of MUD participants toward MA-related image cues. As shown in Table 3 and Figure 1, participants’ reaction times to MA-related image stimuli were significantly longer than those to neutral image stimuli (t = −6.114, p < 0.001). This indicates that MUD participants exhibited a pronounced attention bias toward MA-related cues.

This study found that compared to neutral cues, individuals with MUD in long-term abstinence displayed a significant attentional bias toward MA-related cues. This finding is consistent with previous studies (27, 43–46), which provide strong evidence of attentional bias in various addictive behaviors, including opioid and cocaine use disorders, as well as nicotine and alcohol use disorders (21, 26, 31). Such attentional bias persists across different paradigms used to assess attentional bias (46). The selective attention to MA-related cues among MUD individuals serves as the foundation for this attentional bias. According to incentive-sensitization theory, substance-related cues possess motivational properties and can capture the attention of individuals with MUD. This results in a significant increase in their motivation to obtain the substance, and in turn, triggers drug-seeking behavior (47). From a neurobiological perspective, long-term MA use leads to structural and functional impairments in the brain, particularly in the prefrontal cortex. Damage to this area reduces the ability of individuals with MUD to control attention, making them unable to suppress irrelevant stimuli during tasks (48).

Unlike the traditional Stroop task for vocabulary addiction, the present study used pictures as the priming stimulus. Previous research has found that pictures elicit more priming effects than words (28). This study compared the reaction times (neutral image reaction time: 921.90 ± 142.58 ms, MA-related image reaction time: 956.58 ± 164.47 ms) with those from Zhu’s study (49), which investigated attentional bias in MUD participants using a lexical addiction Stroop task (neutral word reaction time: 723.30 ± 118.56 ms, MA-related word reaction time: 728.47 ± 118.05 ms). It was found that MUD individuals exhibited longer reaction times to images compared to words. Additionally, the variability in image reaction times was higher than that for word reaction times. These results are consistent with Hester et al.’s findings on attentional bias in cocaine users (50). This suggests that MUD individuals’ selectively attend to both neutral and MA-related images, possibly due to their preferences and the inherent characteristics of image and word stimuli. Despite this, the image-based addiction Stroop task remains a sensitive tool for studying attentional bias (50).

This study found that attentional bias in MUD individuals was significantly positively correlated with both withdrawal craving and cue craving. This finding is consistent with most addiction-related research (26). A meta-analysis showed a strong relationship between attentional bias and craving, which is influenced by various factors (45). Attentional bias and craving are two interrelated psychological phenomena that may share similar underlying mechanisms. Specifically, some studies suggest that attentional bias and psychological craving are causally linked (21). The subjective craving for MA can lead to an attentional bias toward MA-related cues, and this bias, in turn, further increases craving levels in MUD individuals. This process is modulated by multiple factors, including impulsivity, impaired inhibitory control, the deliberate suppression of craving, and the avoidance of drug cues. Additionally, from a classical conditioning perspective, attentional bias can cause MA-related cues to elicit expectations of drug use, further reinforcing attentional bias toward those cues (21). This interrelationship implies that simultaneous intervention targeting both psychological craving and attentional bias may be more effective in guiding clinical interventions.

Attentional bias can be used to differentiate individuals with varying levels of withdrawal craving and cue craving. Specifically, individuals with low withdrawal craving or no cue craving did not show significant differences in reaction times between MA-related and neutral picture stimuli, while those with high withdrawal craving or high cue craving exhibited longer reaction times to MA-related stimuli. Notably, effect sizes in the results indicate that the high-craving group exhibited larger effect sizes compared to the low or no-craving groups. Furthermore, MUD individuals with different withdrawal craving levels demonstrated intergroup differences in attentional bias. Although no statistically significant differences in attentional bias were found across cue craving groups, as illustrated in Figure 2, attentional bias scores increased with higher cue craving levels. This study supports the notion that MUD individuals with different levels of psychological craving exhibit varying degrees of attentional bias. A review by Field et al. suggests that attentional bias is related to the intensity of craving at the moment (26). And in the context of MA withdrawal, this craving represents a withdrawal craving state. Previous research has also shown that nicotine-dependent individuals experiencing tobacco deprivation, those with more intense craving exhibit a stronger correlation between their craving and attentional bias (51, 52). However, a study on attentional bias and subjective craving in MUD individuals found that both those with and without reported craving after cue exposure showed significantly higher reaction times to drug-related cues than to neutral cues. This result differs from our findings, possibly due to variations in cue exposure strength, in the type of attentional bias paradigm or in the method of measurement (53, 54). Furthermore, attentional bias related to illegal substances (e.g., cannabis, cocaine, and heroin) shows a stronger relationship with subjective craving compared to substances like tobacco and alcohol (45).

However, the results further confirm that only withdrawal craving—not cue-induced craving—is independently and specifically associated with attentional bias. It is important to note that the time interval between assessments may also have contributed to the negative results regarding cue-induced craving and attentional bias. Since cue-induced craving is an acute state that decays rapidly over time, the multi-day delay likely weakened any observable relationship. Subsequent studies should aim to minimize this delay by measuring attentional bias immediately after cue exposure within the same experimental session to better capture their dynamic relationship.

These findings also allows us to frame our results within the stage-based model of addiction. Prolonged abstinence does not eliminate the neuroadaptive changes from chronic addiction; instead, it shifts the core motivational driver from acute reward seeking to persistent distress relief. The withdrawal/negative affect stage is defined by negative emotional states (e.g., dysphoria, stress) and weakened reward system function—such as reduced dopamine in the nucleus accumbens (55). Notably, this dopamine deficit may heighten the brain’s prioritization of stimuli linked to distress relief-related stimuli, directing attentional resources automatically toward cues that alleviate withdrawal symptoms and thereby strengthening the aforementioned association. Critically, the context of long-term compulsory abstinence powerfully amplifies this stage. The enforced absence of drugs and drug-related cues suppresses the transient reactivity of the preoccupation/anticipation stage, allowing the persistent, endogenous distress of withdrawal to become the dominant motivational focus. In this environment, attentional bias may thus function primarily as a mechanism of negative reinforcement, aimed at relieving this chronic aversive state.

In contrast, the preoccupation/anticipation stage centers on cue-induced craving and executive function deficits, driven by dysregulation in the prefrontal cortex and insula (15). As cue-induced craving is often triggered by transient external stimuli rather than persistent internal distress, its motivational pull may not be strong enough to consistently shape attentional allocation—may explain why it shows no robust association with attentional bias. Since attentional bias is linked to withdrawal craving rather than cue-induced craving, it logically reflects the cognitive prioritization of alleviating withdrawal-related distress rather than the cue-driven motivational processes (15). This theoretical refinement offers a more precise understanding of when and for whom attentional bias is most salient. Future research could further validate this by examining how attentional bias changes across different stages of addiction and whether stage-specific interventions yield better outcomes.

There are several limitations to this study. First, as a cross-sectional survey, it could not determine the causal relationship between attentional bias and craving. Second, the present study did not include female participants, voluntary MA abstainers, and MUD individuals in short-term abstinence, which limits the generalizability of our findings. Third, the multi-day interval between the assessments of cue-induced craving and attentional bias may have led to the decay of the transient craving state, potentially confounding the observable relationship. In addition, objective physiological data such as electrocardiogram, electrodermal activity, and electroencephalogram were not collected simultaneously. Larger scale, more comprehensive assessments, and longitudinal cohort studies should be conducted to investigate the characteristics of attentional bias among MUD participants across different stages. Specifically, these attentional bias characteristics at different abstinence stages is a critical direction for future research.