Altered reward processing and amotivational syndrome in young adult cannabis users: an fMRI study using a reinforcement learning task

Background

Cannabis is the most widely used illicit substance worldwide and has been associated with reduced motivation, often termed an “amotivational syndrome”, as well as lower educational achievement [1]. From a neurocomputational perspective, motivated behaviour can be conceptualised within a reinforcement learning framework, in which actions are updated based on reward prediction errors (RPE), implemented through cortico-striatal dopaminergic circuits [2]. This cannabis-related reduction in motivation appears particularly pronounced in adolescents and young adults, a developmental period characterised by ongoing maturation of fronto-striatal systems underlying reward processing and motivation. However, only a subset of users develop clinically significant motivational impairments, suggesting marked inter-individual vulnerability. Identifying brain-based markers of this vulnerability may support early detection and prevention of cannabis-related psychopathology.

Objectives

This study examined whether regular cannabis use in adolescence/young-adulthood is associated with altered reward processing and reinforcement learning, and whether these changes may serve as neuroimaging markers of early motivational dysfunction and psychopathology.

Methods

The sample included 62 young adults aged 18-25 years with regular cannabis use (at least weekly) and no psychiatric diagnoses, and 46 non-user controls matched for age, sex, and IQ, with no personal or family history of psychiatric disorders. Participants completed a Reinforcement Learning Task (RLT) during fMRI, in which they repeatedly chose between two options with different reward probabilities in two conditions. In the reward condition, one stimulus had a higher probability of reward, allowing learning from feedback, whereas in the bivalent condition both stimuli were rewarded equally (50%), so no learning was possible. A key measure was the RPE, reflecting the discrepancy between expected and actual outcomes and modeled using the Q-learning framework developed by Christopher Watkins and Peter Dayan [3]. fMRI data were analysed using FSL FEAT with a General Linear Model approach (p < 0.05, cluster-forming threshold z > 2.3).

Results

During reward versus bivalent stimulus processing (corresponding to reward anticipation), cannabis users showed increased activation in medial prefrontal regions (cluster size = 389; MNI peak coordinates: x = -26, y = 40, z = 32; Z = 3.69, p = 0.00563). In contrast, during reward outcome processing (reward versus bivalent feedback), they exhibited reduced activation in the superior medial prefrontal cortex (cluster size = 267; MNI peak coordinates: x = -8, y = 56, z = 38; Z = 3.39, p = 0.0421), indicating reduced engagement of outcome evaluation and value updating processes. However, there were no significant differences between groups in RPE encoding, suggesting preserved neural representations of prediction error signals.

Conclusions

These findings suggest a dissociation in reward processing in young cannabis users, with increased medial prefrontal activity during reward anticipation and reduced activity during outcome evaluation, indicating altered integration of reward and feedback signals. This pattern may reflect a neural marker of altered reward processing. Nevertheless, RPE encoding did not differ between groups, suggesting preserved coding of prediction errors.

References

Rovai, L., Maremmani, A. G. I., Pacini, M., Pani, P. P., Rugani, F., Lamanna, F., Schiavi, E., Mautone, S., Dell’Osso, L., & Maremmani, I. (2013). Negative dimension in psychiatry. Amotivational syndrome as a paradigm of negative symptoms in substance abuse. Rivista Di Psichiatria, 48(1), 1-9.
O’Doherty, J. P., Cockburn, J., & Pauli, W. M. (2016). Learning, Reward, and Decision Making. Annual Review Of Psychology, 68(1), 73-100.
Watkins, C. J. C. H., & Dayan, P. (1992). Q-learning. Machine Learning, 8(3-4), 279-292.