Should We Control? The Interplay Between Cognitive Control and Information Integration in the Resolution of the Exploration-Exploitation Dilemma
Should We Control? The Interplay Between Cognitive Control and Information Integration in the Resolution of the Exploration-Exploitation Dilemma
复制标题
DOI：
10.1037/xge0000546
复制
发表时间：
2019-06-01
影响因子：
4.1
通讯作者：
Alexander, William
中科院分区：
文献类型：
--
作者：
Dezza, Irene Cogliati;Cleeremans, Axel;Alexander, William
在他们的日常决策中，人类和动物经常面临着选择一个有回报的熟悉选项（即，开发）或用于新颖的、不确定的选项，然而，其可能在不久的将来产生更好的回报（即，勘探）。尽管进行了广泛的研究，但人类解决这种探索-开发困境的认知机制仍然知之甚少。在这项研究中，我们挑战了流行的假设，即剥削是一个全球性的默认策略，必须通过认知控制机制来抑制，以使探索性策略。为了做到这一点，我们要求参与者参与一项具有挑战性的工作记忆任务，同时在赌博任务中进行重复选择。结果表明，操纵认知控制资源完全阻碍了参与者以定向，有意的方式探索环境的能力。此外，在某些情景下，采用剥削策略也依赖于认知控制资源的可用性。使用最近的信息整合计算模型进行的其他分析表明，增加认知负荷特别会干扰联合收割机将奖励和信息结合起来以告知选择的能力。我们的研究结果揭示了支持解决困境的认知机制，并提供了一个正式的基础，通过它来探索目标导向行为的病理。
In their daily decisions, humans and animals are often confronted with the conflicting choice of opting either for a rewarding familiar option (i.e., exploitation) or for a novel, uncertain option that may, however, yield a better reward in the near future (i.e., exploration). Despite extensive research, the cognitive mechanisms that subtend the manner in which humans solve this exploration-exploitation dilemma are still poorly understood. In this study, we challenge the popular assumption that exploitation is a global default strategy that must be suppressed by means of cognitive control mechanisms so as to enable exploratory strategies. To do so, we asked participants to engage in a challenging working memory task while performing repeated choices in a gambling task. Results showed that manipulating cognitive control resources exclusively hindered participants' ability to explore the environment in a directed, intentional manner. Moreover, under certain scenarios, adopting exploitative strategies was also dependent on the availability of cognitive control resources. Additional analyses using a recent computational model of information integration suggests that increasing cognitive load specifically interferes with the ability to combine reward and information in order to inform choices. Our results shed light on the cognitive mechanisms that underpin the resolution of the dilemma and provide a formal foundation through which to explore pathologies of goal-directed behavior.