Department of Neuroscience, University of Minnesota.
Behav Neurosci. 2021 Aug;135(4):469-486. doi: 10.1037/bne0000460. Epub 2021 Jun 24.
Many foraging experiments have found that subjects are suboptimal in foraging tasks, waiting out delays longer than they should given the reward structure of the environment. Additionally, theories of decision-making suggest that actions arise from interactions between multiple decision-making systems and that these systems should depend on the availability of information about the future. To explore suboptimal behavior on foraging tasks and how varying the amount of future information changed behavior, we ran rats on two matching neuroeconomic foraging tasks, Known Delay (KD) and Randomized Delay (RD), with the only difference between them being the certainty of the cost of future opportunities. Rats' decision-making strategies differed significantly based on the amount of future certainty. Rats on both tasks still showed suboptimality in decision-making through a sensitivity to sunk costs; however, rats on KD showed significantly less sensitivity to sunk costs than rats on RD. Additionally, on neither task did the rats account for travel and postreward lingering times as heavily as prereward foraging times providing evidence problematic for the Marginal Value Theorem model of foraging behavior. This suggests that while future certainty reduced decision-making errors, more complex decision-making processes unaffected by future certainty were involved and likely produced these decision-making errors within subjects on these foraging tasks. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
许多觅食实验发现,在给定环境奖励结构的情况下,被试在觅食任务中表现不佳,等待时间超过了他们应该等待的时间。此外,决策理论表明,行动是由多个决策系统相互作用产生的,而这些系统应该取决于对未来信息的可用性。为了探索觅食任务中的次优行为以及未来信息的数量如何改变行为,我们在两种匹配的神经经济学觅食任务中对老鼠进行了测试,即已知延迟(KD)和随机延迟(RD),它们之间唯一的区别是未来机会成本的确定性。老鼠的决策策略基于未来确定性的数量而有显著差异。在两种任务中,老鼠的决策仍然表现出对沉没成本的敏感性,这表明它们存在次优性;然而,KD 任务中的老鼠对沉没成本的敏感性明显低于 RD 任务中的老鼠。此外,在这两项任务中,老鼠并没有像预奖励觅食时间那样,将旅行和奖励后逗留时间与奖励后觅食时间一样重要,这为觅食行为的边际价值定理模型提供了有问题的证据。这表明,虽然未来的确定性减少了决策错误,但涉及到更多不受未来确定性影响的复杂决策过程,并且可能在这些觅食任务中的被试中产生了这些决策错误。