The opioid-receptor antagonist naltrexone (NTX) reduces goal-directed alcohol drinking in rats presumably by blunting alcohol reward. However, different operant conditioning behavior can be produced by different reinforcement schedules, with goal-directed operant behavior being more sensitive to changes in reward value than less flexible, habit-associated models. We tested the hypothesis that NTX more effectively reduces alcohol drinking and seeking in a goal-directed than in a habit-associated operant model, and more effectively reduces alcohol versus sucrose self-administration, consistent with diminished alcohol reward.
Rats were trained to self-administer 10% alcohol or 1.5% sucrose in a lever-press task and then underwent a within-subject assessment of NTX (0.1 to 1 mg/kg) effects on operant behavior. A fixed-ratio (FR5) reinforcement schedule was used to model goal-directed behavior, and a variable-interval (VI30) schedule was used to model habitual behavior.
As predicted, NTX reduced fluid deliveries earned by the FR5-alcohol group significantly more than all other groups. However, NTX reduced lever presses during self-administration sessions in VI30-trained rats without reducing earned deliveries, due to the low contingency between rate of pressing and fluid deliveries under that schedule. Interestingly, when fluid delivery was withheld (extinction), NTX reduced reward-seeking in all rats. Finally, NTX blocked reinstatement of reward-seeking upon presentation of 0.2 ml alcohol or sucrose and associated cues in the FR5-trained but not VI30-trained rats.
NTX reduced goal-directed alcohol drinking compared with other operant conditions. In addition, NTX blocked reinstatement of reward-seeking in rats trained on the goal-directed FR5 reinforcement schedule but not in rats trained on the habit-like VI30 reinforcement schedule. However, NTX also exerted nonspecific effects on reward-seeking that were revealed under low-effort contingency conditions or absence of reward. Together, these data support the hypothesis that NTX is less effective in conditioning models that are more habit-associated.