The Brain-Hijacking Model
The dominant model of addiction rests on a simple claim: drugs flood the brain's reward circuits with dopamine, rewire decision-making, and leave the user unable to choose anything else. This framework shapes everything from DARE presentations to pharmaceutical research to clinical guidelines. Block the receptors, manage withdrawal, suppress the craving. The person is secondary to the molecule.
For decades, the evidence base reinforcing this model came from a remarkably uniform experimental setup: put a rat alone in a cage, give it a lever that delivers cocaine, heroin, or methamphetamine, and watch it press that lever until it stops eating, stops sleeping, sometimes stops breathing. Conclusion: the drug is irresistible.
A team at the National Institute on Drug Abuse asked a question nobody had tested with proper operant methods: What if the rat had something else worth pressing a lever for?
Two Levers, One Cage
Marco Venniro and colleagues at the NIH designed custom chambers with two active levers: one delivered an intravenous drug infusion, while the other opened a sliding door to a small adjacent chamber containing a familiar peer rat for a brief social interaction. The setup forced a genuine choice, and the researchers tested every variable they could think of to break it.
They tried methamphetamine and heroin at escalating doses, testing both male and female rats, and retested after 15 and 30 days of forced abstinence to check whether the preference was durable. They ran rats through the established extended-access escalation model, where animals ramp up drug intake over weeks until their behavior mirrors compulsive use, and in Experiment 1 alone, social reward prevented methamphetamine self-administration across four different dose levels, with a statistical comparison producing F1,9 = 201.1 (P < 0.001).
The effect was not marginal. It was categorical.
Even the 'Addicted' Rats Chose a Friend
The most striking test came in Experiment 2, where the researchers trained 42 male rats on methamphetamine for 50 daily sessions, then classified each animal using three DSM-IV-based addiction criteria: compulsive seeking during non-drug periods, elevated progressive-ratio responding, and persistence through punishment. Eight rats scored "High," nine scored "Medium," and twenty-five scored "Low."
Then they gave all of them the choice.
The addicted rats chose social interaction just as strongly as the non-addicted ones, producing an F-statistic for social versus drug preference across all groups of 1,641.8 (P < 0.001), with no effect of addiction-score group and no interaction between group, reward type, and session. That number is unusual in behavioral neuroscience: it means the variation between conditions dwarfed the variation within them so completely that the signal buried the noise, and the correlation between addiction score and social-reward choice was exactly zero.
Social Connection Did What Isolation Could Not
Experiment 4 tested whether social-choice-induced abstinence could prevent what addiction researchers call "incubation of craving," the progressive increase in drug-seeking that occurs during forced abstinence and makes relapse deadlier over time, a phenomenon well-documented in both rodent models and human clinical settings. In rats forced to abstain in home cages, methamphetamine seeking escalated sharply by day 15 (F1,20 = 6.0, P = 0.02). In rats that abstained by choosing social interaction, it did not.
This protection persisted for weeks. Even after 30 additional days of forced home-cage abstinence following the social-choice period, craving remained suppressed. The researchers identified a neural correlate: social-choice abstinence activated PKCฮด-expressing inhibitory neurons in the central amygdala and suppressed activity in the anterior insular cortex, brain regions previously linked to relapse after contingency management.
Critically, social connection outperformed food as an alternative reward in a direct comparison: rats that abstained via food choice still showed incubation of craving at day 15, while rats that abstained via social choice did not, and when the researchers tried to devalue the social reward through satiety by cohousing rats with their social partners for up to six days, it had no effect on preference. Food satiety, by contrast, pushed rats back toward drugs.
The Strongest Counterargument
The most important objection is also the most obvious: these are rats. Carl Hart at Columbia and others have argued that rodent models systematically distort addiction science because they strip away the cognitive, social, and economic complexity of human substance use. A rat pressing a lever for 40 minutes of peer contact in a clean laboratory does not model a person with a decade of opioid dependence, disrupted relationships, untreated PTSD, and no stable housing.
There is also a design-level critique that cuts at real-world applicability: when social reward was delayed by more than 60 seconds or punished with footshock, rats shifted back to methamphetamine. This means the protective effect requires immediate, reliable access to social contact. For humans in addiction, social support is rarely instant, rarely uncomplicated, and rarely free of the very punishment the study used to break the effect.
What We Didn't Prove
No rat study can demonstrate that the same mechanism operates in humans. The Sprague-Dawley rats used here had never experienced trauma, poverty, chronic pain, or psychiatric comorbidity, and the social partner in the experiment was always available, always familiar, and always positive, conditions that bear no resemblance to the fractured social worlds most addicted humans actually inhabit.
The sample, while large for behavioral neuroscience, remains a single-species finding from a single laboratory, and although the same group has replicated and extended these results to cocaine, heroin, and methamphetamine using an automated procedure (Venniro et al. 2022, J. Neuroscience), and a 2025 Nature Communications study identified dopaminergic ensembles that encode social reward's competitive advantage over drugs, fully independent replications from laboratories outside the Shaham group remain limited. The neural mechanism (PKCฮด neuron activation in central amygdala) is correlational: the study showed association, not necessity or sufficiency.
The Bottom Line
For decades, the standard addiction experiment tested what happens when a social animal is locked alone in a cage with a drug lever. This study tested what happens when you give that animal something else to choose. The answer was unambiguous: even rats behaviorally classified as addicted abandoned methamphetamine and heroin the moment a social alternative appeared. The finding doesn't prove that connection cures addiction. But it does suggest the field's foundational experimental setup was never really testing the drug's power. It was testing the absence of everything else.
What You Can Do
If you work in addiction treatment, investigate the community reinforcement approach (CRA), one of the most effective behavioral interventions for substance use disorders. It works by rebuilding social, vocational, and recreational reinforcers that compete with drug use. This study provides a biological rationale for its effectiveness.
If someone you care about is struggling with substance use, understand that maintaining connection is itself a form of intervention. Isolation is not merely a consequence of addiction; this research suggests it is a structural driver. Staying present, even imperfectly, may matter more than any conversation about quitting.
If you design policy around addiction, reconsider whether treatment models built on isolation (carceral settings, residential programs that restrict outside contact) are compatible with the neuroscience. The protocol is publicly available: Venniro & Shaham, Nature Protocols, 2020.