To make adaptive decisions in multi-dimensional environments, animals must infer the relevant features and afford them priority for the control of learning and actions. Prioritizing sources of information is the role of executive control and attention, but very little is known about the mechanisms by which the brain computes relevance or priority. We present a new decision-theoretic view of priority as a cognitive state that implements information gathering policies. This response is distinct from representations of information utility and it is recruited when the information serves instrumental and non-instrumental motives, suggesting that curiosity and instrumental information demand are more closely related than is traditionally assumed. We discuss the broader relevance of these findings for decision making in realistic, complex situations.

为了在多维环境中做出适应性决策，动物必须推断出相关特征并将其优先控制学习和行动。信息源的优先级是执行控制和注意力的作用，但是人们对大脑计算相关性或优先级的机制知之甚少。我们提出了一种新的优先级决策理论观点，将其视为一种实施信息收集策略的认知状态。此响应与信息效用的表示形式不同，它是在信息用于工具性和非工具性动机时才被招募的，这表明好奇心和工具性信息需求比传统上认为的更紧密相关。我们讨论了这些发现与现实，复杂情况下的决策的广泛相关性。