Computer training that reduced cannabis approach bias didn't change brain cue responses in teens

CAAT training shifted approach bias toward avoidance while sham training increased approach bias (trend p=0.055). However, no significant neural changes emerged on fMRI between CAAT and sham groups. Small-to-medium effect sizes were observed for decreased amygdala (d=0.36) and medial prefrontal (d=0.48) activation to cannabis cues in the CAAT group.

Sub-study of a larger RCT: 37 cannabis-using youth (ages 17-21) completed pre and post fMRI scans around 6 sessions of CAAT (n=19) or sham training (n=18), with cannabis cue reactivity task.

Computer-based interventions are attractive because they're scalable and cheap. While this particular approach reduced cannabis use behaviorally, the brain mechanism appears not to involve changes in cue reactivity, meaning the intervention may work through other neural pathways.

This is an honest null result. The intervention worked clinically (in the larger trial) but not through the hypothesized brain mechanism (cue reactivity). This kind of finding is crucial for understanding which component of treatment actually drives behavior change.

Very small neuroimaging subsample (n=37). The larger trial showed behavioral effects; the subsample may have been underpowered to detect neural changes. Only examined cue reactivity; other neural measures might show changes. Non-treatment-seeking sample.