EEG profiles recurred across groups, but individual reliability was low

The increasing prevalence of Cannabis Use Disorder (CUD) necessitates measures that reliably predict outcomes. Based in the incentive sensitization model of addiction, the late positive potential (LPP) has been used to predict choice of drug versus affective cues and classify individuals into clinically-relevant neuroaffective reactivity profiles derived from viewing affective and drug-related pictures in individuals who use substances. However, this approach has not been conducted on a sample of individuals who use cannabis nor has its retest reliability been assessed. This study replicated and extended the neuroaffective reactivity profile findings in individuals with CUD and tested the one-week profile reliability. At two lab visits one week apart, 44 participants completed a probabilistic choice task and viewed cannabis and affective pictures with EEG recording. k-means cluster analyses were applied to the resulting LPPs to create neuroaffective reactivity profiles. Consistent with past literature, at both lab visits, participants were organized into two clusters characterized by differences in the cannabis and pleasant LPP amplitudes (one with greater cannabis than pleasant LPPs and one with greater pleasant than cannabis LPPs), but LPP clusters did not differ in choice behavior or other indicators of problematic use. Despite this reliable classification at the group level, individual level retest reliability for the profiles was low, outperformed by residualized LPP scores and raw LPPs. These findings add to literature suggesting a common pattern of neuroaffective reactivity profiles across addiction populations but caution their clinical utility considering the unresolved reliability and lack of convergent validity across units of analysis.