IntroductionCannabis is one of the most widely used psychoactive substances globally, with an estimated 228 million users in 2022, representing over 4% of the global adult population (United Nations Office on Drugs and Crime, 2023). Although the risks of harm associated with cannabis are generally considered lower than for substances such as alcohol or opioids (Nutt et al., 2010), regular use is associated with cognitive impairments, psychosocial problems and increased risk of developing cannabis use disorder (CUD), particularly among daily users (Connor et al., 2021; Solmi et al., 2023). These risks are exacerbated by rising Δ9-tetrahydrocannabinol (THC) potency (ElSohly et al., 2021) and decreasing public concern about the harms of use (Harrison et al., 2023; Waddell, 2022).
The primary psychoactive compound in cannabis, THC, activates CB1 receptors, which are densely distributed in brain regions critical to executive function, emotional regulation and reward processing (Mackie, 2005). Regular cannabis use has been associated with neuroadaptations in these areas, particularly in the prefrontal cortex (PFC), anterior cingulate cortex (ACC) and orbitofrontal cortex (OFC; Gilman et al., 2014; Lorenzetti et al., 2019; Zhou et al., 2018). These changes may contribute to cognitive impairments, including deficits in attention, working memory, processing speed and inhibitory control (Crean et al., 2011; Lovell et al., 2020).
Inhibitory control is a core component of addiction vulnerability and maintenance (Luijten et al., 2014; Yücel et al., 2019) and is often studied using behavioural paradigms such as the Go/No-Go task (GNG) and the Stop-Signal Task (SST; Logan and Cowen, 1984). Evidence suggests cannabis users may exhibit poorer performance on the SST, while findings for the GNG are less consistent (Liu et al., 2019). Notably, some functional Magnetic Resolution Imaging (fMRI) studies have identified functional differences in cannabis users compared to controls - such as ACC hypoactivity and increased PFC activation - despite comparable task performance, possibly indicating compensatory neural activity (Hester et al., 2009; Tapert et al., 2007). These tasks reflect distinct subcomponents of motor inhibition: action restraint and action cancellation (Schachar et al., 2007). The GNG captures restraint (i.e., preventing a response before it is initiated) while the SST assesses cancellation (i.e., interrupting an ongoing motor response). Although both forms of inhibition are relevant to addiction, they may rely on partially dissociable neural substrates and show differential sensitivity to cannabis-related impairments.
Standard inhibitory control tasks typically use neutral stimuli, which may not fully reflect the influence of drug-related attentional bias. Cannabis users demonstrate increased neural and behavioural reactivity to cannabis-related cues (O’Neill et al., 2020; Sehl et al., 2021), and such cue-reactivity is implicated in craving and relapse (Berridge and Robinson, 2016; Zilverstand et al., 2018). Cannabis-related versions of inhibitory control tasks can help assess whether attentional biases and reward signalling impair inhibitory control (Field and Cox, 2008).
This study used functional near-infrared spectroscopy (fNIRS), a non-invasive imaging method suitable for measuring cortical haemodynamics, to assess PFC and OFC activity during two inhibitory control tasks (GNG and SST), and a measure of attentional bias (Cannabis Stroop task), each incorporating cannabis-related stimuli.
It was hypothesised that cannabis users, relative to non-user controls, would show reduced task performance and differences in neural activity – specifically differences in PFC activity during inhibition and OFC activity in response to cannabis cues.
DiscussionThe current study aimed to investigate differences in behavioural and neural responses during cannabis cue-specific inhibitory control tasks between regular cannabis users and non-user controls using fNIRS. Contrary to the hypotheses, the results revealed no significant group differences in behavioural performance or neural activation (i.e. oxyHb and deoxyHb levels) across all three inhibitory reward tasks (i.e. GNG, Cannabis Stroop and SST). Furthermore, there was no significant effect of total lifetime dose, age of onset or current frequency of use on behavioural or neural responses.
The behavioural results are partly in line with previous GNG studies which have largely found no significant difference in errors between cannabis users and controls (Liu et al., 2019). However, deficits in the SST, such as those identified by Liu et al. (2019), may not be present across all cannabis-using populations and might reflect specific patterns of use.
Although fMRI studies have reported neural differences between cannabis users and controls (Hester et al., 2009; Tapert et al., 2007), the fNIRS data in this study did not reveal any group differences in neocortical regions of the PFC, including the OFC. This discrepancy might be due to the inability of fNIRS to access subcortical regions (Pinti et al., 2020), where differential neural activities may have occurred in this sample.
Performance on the Cannabis Stroop Task was similar across groups, contradicting previous research showing poorer performance among heavy or dependent users (Field, 2005). This suggests that the attentional bias defecits reported previously may be specific to more severe patterns of cannabis use, or may reflect effects that are not consistently replicable in non-dependent populations.
The lack of group differences in response to cannabis-related cues is somewhat surprising, given prior findings of increased PFC activation in response to such stimuli (O’Neill et al., 2020; Sehl et al., 2021). However, those studies often used passive cue exposure, whereas the present design embedded cues in active tasks. This suggests that context plays a crucial role in neural cue reactivity. Additionally, it is possible that cue-based reward processes occur at a subcortical level and are thus not detectable using fNIRS (Pinti et al., 2020).
There are several considerations when interpreting these null findings. First, the current sample may not reflect heavy or dependent cannabis use. Participants reported at least weekly cannabis use over the previous 3 months, but this threshold may capture more moderate, non-problematic users. Previous studies that identified significant neural or behavioural effects often included heavier or daily users (Borissova et al., 2022; Field, 2005; Hester et al., 2009; Liu et al., 2019; Sehl et al., 2021; Tapert et al., 2007). Second, the use of a 3-month minimum criterion may have excluded individuals with CUD, reducing the likelihood of detecting neuro-adaptive changes. Prominent addiction models posit that such adaptations emerge with chronic or dependent use (Berridge and Robinson, 2016; Luijten et al., 2014; Zilverstand et al., 2018). This view is supported by evidence that structural and functional brain changes are more likely in dependent populations (Chye et al., 2017, 2019; Lorenzetti et al., 2020, 2024). Stratifying users by diagnostic tools such as the Cannabis Use Disorders Identification Test-Revised (Adamson et al., 2010) may help capture relevant group differences in future studies.
Another limitation is the reliance on self-reported substance use. Although practical and ethically less invasive, self-report can be affected by recall and social desirability bias (Del Boca and Noll, 2000; Hjorthøj et al., 2012; Latkin et al., 2017). No urine screening was conducted to confirm recent cannabis use or rule out other substances. While biological verification offers more accuracy (Huestis et al., 2007; Lorenzetti et al., 2022), such methods remain resource-intensive and are not always feasible in community-based designs.
Methodologically, the SST may have lacked sensitivity. Average stopping accuracy was 18.37%, below the recommended 50%, and the task included fewer stop trials (n = 40) than advised (Verbruggen et al., 2019). These factors may have reduced the reliability of SSRT estimation. Task duration may also have influenced fNIRS findings. While the Cannabis Stroop followed block design conventions, the GNG and SST used 4–5-minute blocks. This may have attenuated the haemodynamic response due to filtering effects (Miezin et al., 2000; Scholkmann et al., 2014), despite parameter adjustments. Future research should adopt standard block designs to optimise fNIRS sensitivity.
Future work should stratify cannabis users by severity and usage pattern, broaden the age range and consider multimodal neuroimaging approaches (e.g. fNIRS-fMRI) to assess both cortical and subcortical effects. Longitudinal studies are also needed to clarify developmental trajectories of cannabis-related cognitive and neurophysiological change, as cross-sectional methods may not capture dynamic changes over time (Hanson et al., 2010; Lorenzetti et al., 2024).
In summary, this study found no significant differences in behavioural performance or neural activation between regular cannabis users and non-user controls during cue-specific inhibitory control tasks. These null findings may reflect non-dependent use patterns and task- or method-specific limitations. Further research is needed to examine whether impairments in inhibitory control and brain function emerge specifically with heavier or dependent cannabis use.