Economic Rationality and Competing Behavioral Rules

该项目专注于回报-单调的行为规则，特别是微观经济学和博弈论中建模的模仿和强化。在第一个筹资阶段，我们已经证明了这两种行为规则在双过程理论意义上都表现出自动过程的特征。我们已经开发了一个简单的分析性双进程模型，该模型根据错误和正确的响应来预测响应时间。具体地说，每当受控进程和自动进程发生冲突时，正确的响应会更慢，但如果两个进程都一致，错误就会更慢。此外，在冲突情况下的决定比在一致的情况下的决定慢。我们首先在贝叶斯更新范式中确认了这些预测，其中强化学习与贝叶斯理性的规定相冲突。这些预测在实验性的古诺寡头垄断中也得到了证实，在这些寡头垄断中，模仿可以被视为一个自动过程，与更受控制的最佳回复行为相冲突。进一步的分析阐明了决策惯性、动机定向和自我控制资源耗竭的作用。使用更多的货币激励和认知负荷操纵，已经取得了令人惊讶的结果。即使准确率低于上限，增加货币激励也不会改善贝叶斯更新范例的性能。复杂决策范例中的认知负荷似乎不会导致明显的性能下降，但令人惊讶的是，响应时间显著减少，这表明过程转变有利于自动化过程。在第二阶段，我们计划在新的但相关的范式(个人和人际)中简朴地测试我们以前的结果和预测的稳定性，以确立双过程方法在整个经济学中的相关性。在这里，我们将利用我们在其他范例方面的经验以及为此目的明确开展的初步研究。我们还计划将部分重点转移到彩票选择上，将其作为风险下决策的自然应用。我们还旨在通过新的范式阐明认知负荷操纵和增加货币激励所获得的显着结果。此外，我们将依靠眼球运动测量和通过脑电获得的事件相关电位，作为与反应时间或选择数据相比更直接的过程测量。最后，我们计划扩展我们的正式-分析双过程响应时间模型，以纳入不同的设置和相关研究，目的也是为研究单位内的其他研究项目和其他项目提供可检验的假设。

The project concentrates on payoff-monotonic behavioral rules, especially imitation and reinforcement as modeled in microeconomics and game theory. In the first funding phase, we have shown that both behavioral rules exhibit the characteristics of automatic processes in the sense of dual-process theories. We have developed a simple analytical dual-process model which delivers predictions for response times in terms of errors and correct responses. Specifically, whenever a controlled process and an automatic one conflict, correct responses are slower, but if both processes are aligned, errors are slower. Further, decisions in case of conflict are slower than decisions in case of alignment. We started by confirming these predictions in Bayesian updating paradigms where reinforcement learning conflicts with the prescriptions of Bayesian rationality. The predictions were also confirmed in experimental Cournot oligopolies where imitation could be seen as an automatic process conflicting with more controlled best-reply behavior. Further analysis has elucidated the role of decision inertia, motivational orientations, and depletion of self-control resources. Surprising results have been obtained using increased monetary incentives and cognitive load manipulations. Increasing monetary incentives does not improve performance in Bayesian updating paradigms, even though accuracy rates are below the ceiling. Cognitive load in complex decisionmaking paradigms does not appear to result in a clear performance reduction, but, surprisingly, response times are significantly reduced, pointing out to a process shift in favor of automatic processes. In the second phase, we plan to parsimoniously test for the stability of our previous results and predictions in new but related paradigms (both individual and interpersonal), in order to establish the relevance of the dual-process approach for economics in full generality. Here we will take advantage of our experience with other paradigms and of preliminary research carried out explicitly for this purpose. We also plan to partially shift the focus towards lottery choice as the natural application of decision making under risk. We also aim to clarify the striking results obtained with cognitive load manipulations and increased monetary incentives by means of new paradigms. Additionally, we will rely on eye-movement measurements and on event-related potentials obtained by means of the EEG, as more direct process measures compared to response times or choice data. Finally, we plan to expand our formal-analytical dual-process model of response times to incorporate different settings and related research, also with the objective of delivering testable hypotheses for the other research lines and other projects within the research unit.