Intelligent Systems and Informatics Laboratory, Mechano-Informatics Department of Graduate School of Information Science and Technology, The University of Tokyo, Eng. Bldg.2, 7-3-1, Hongo, Bunkyo-ku, Tokyo, 113-8656, Japan.
Sci Rep. 2020 Dec 9;10(1):21525. doi: 10.1038/s41598-020-78635-1.
Global self-esteem is a component of individual personality that impacts decision-making. Many studies have discussed the different preferences for decision-making in response to threats to a person's self-confidence, depending on global self-esteem. However, studies about global self-esteem and non-social decision-making have indicated that decisions differ due to reward sensitivity. Here, reward sensitivity refers to the extent to which rewards change decisions. We hypothesized that individuals with lower global self-esteem have lower reward sensitivity and investigated the relationship between self-esteem and reward sensitivity using a computational model. We first examined the effect of expected value and maximum value in learning under uncertainties because some studies have shown the possibility of saliency (e.g. maximum value) and relative value (e.g. expected value) affecting decisions, respectively. In our learning task, expected value affected decisions, but there was no significant effect of maximum value. Therefore, we modelled participants' choices under the condition of different expected value without considering maximum value. We used the Q-learning model, which is one of the traditional computational models in explaining experiential learning decisions. Global self-esteem correlated positively with reward sensitivity. Our results suggest that individual reward sensitivity affects decision-making depending on one's global self-esteem.
全球自尊是个体人格的一个组成部分，会影响决策。许多研究讨论了在面临个人自信心受到威胁时，不同的决策偏好取决于全球自尊。然而，关于全球自尊和非社交决策的研究表明，决策会因奖励敏感性而有所不同。这里的奖励敏感性是指奖励改变决策的程度。我们假设，全球自尊较低的个体奖励敏感性较低，并使用计算模型研究了自尊和奖励敏感性之间的关系。我们首先检验了在不确定条件下学习中期望价值和最大值的影响，因为一些研究表明，显著值（例如最大值）和相对值（例如期望价值）分别影响决策。在我们的学习任务中，期望价值影响决策，但最大值没有显著影响。因此，我们在不考虑最大值的情况下，根据不同期望价值的条件来建模参与者的选择。我们使用了 Q 学习模型，这是解释经验学习决策的传统计算模型之一。全球自尊与奖励敏感性呈正相关。我们的结果表明，个体的奖励敏感性会根据其全球自尊而影响决策。
