Rationale

A growing body of research suggests that substance use disorder (SUD) may be characterized as disorders of decision making. However, drug choice studies assessing drug-associated decision making often lack more complex and dynamic conditions that better approximate contexts outside the laboratory and may lead to incomplete conclusions regarding the nature of drug-associated value.

Objectives

The current study assessed isomorphic (choice between identical food options) and allomorphic (choice between remifentanil [REMI] and food) choice across dynamically changing reward probabilities, magnitudes, and differentially reward-predictive stimuli in male rats to better understand determinants of drug value. Choice data were analyzed at aggregate and choice-by-choice levels using quantitative matching and reinforcement learning (RL) models, respectively.

Results

Reductions in reward probability or magnitude independently reduced preferences for food and REMI commodities. Inclusion of reward-predictive cues significantly increased preference for food and REMI rewards. Model comparisons revealed that reward-predictive stimuli significantly altered the economic substitutability of food and REMI rewards at both levels of analysis. Furthermore, model comparisons supported the reformulation of reward value updating in RL models from independent terms to a shared, relative term, more akin to matching models.

Conclusions

The results indicate that value-based quantitative choice models can accurately capture choice determinants within complex decision-making contexts and corroborate drug choice as a multidimensional valuation process. Collectively, the present study indicates commonalities in decision-making for drug and non-drug rewards, validates the use of economic-based SUD therapies (e.g., contingency management), and implicates the neurobehavioral processes underlying drug-associated decision-making as a potential avenue for future SUD treatment.

中文翻译：

---

量化药物和非药物决策动态的基于价值的决定因素

 基本原理

越来越多的研究表明，物质使用障碍（SUD）可能被描述为决策障碍。然而，评估药物相关决策的药物选择研究通常缺乏更复杂和动态的条件，无法更好地接近实验室外的环境，并可能导致关于药物相关价值性质的不完整结论。

 目标

目前的研究评估了雄性大鼠在动态变化的奖励概率、大小和差异性奖励预测刺激中的同构（在相同食物选择之间的选择）和同构（在瑞芬太尼 [REMI] 和食物之间选择）选择，以更好地了解药物价值的决定因素。分别使用定量匹配和强化学习（RL）模型在聚合和逐个选择级别上分析选择数据。

 结果

奖励概率或幅度的降低独立地降低了对食品和 REMI 商品的偏好。奖励预测线索的加入显着增加了对食物和 REMI 奖励的偏好。模型比较表明，奖励预测刺激在两个分析层面都显着改变了食物和 REMI 奖励的经济可替代性。此外，模型比较支持将 RL 模型中的奖励值更新从独立术语重新表述为共享相对术语，更类似于匹配模型。

 结论

结果表明，基于价值的定量选择模型可以准确地捕捉复杂决策环境中的选择决定因素，并证实药物选择是一个多维评估过程。总的来说，本研究表明药物和非药物奖励决策的共性，验证了基于经济的 SUD 疗法（例如应急管理）的使用，并暗示药物相关决策背后的神经行为过程是潜在的为未来 SUD 治疗提供途径。