Taylor Grant J, Brown Scott D, Evans Nathan J
School of Psychology, University of Queensland, Brisbane, Australia.
School of Psychology, University of Newcastle, Newcastle, Australia.
Psychon Bull Rev. 2025 Apr 3. doi: 10.3758/s13423-025-02680-y.
Finding an optimal decision-making strategy requires a careful balance between the competing demands of accuracy and urgency. In experimental settings, researchers are typically interested in whether people can optimise this trade-off, typically operationalised as reward rate, with evidence accumulation models serving as the key framework to determine whether people are performing optimally. However, recent studies have suggested that inferences about optimality can be highly dependent on the task design, meaning that inferences about whether people can achieve optimality may not generalise across contexts. Here, we investigate one typically overlooked design factor: whether participants spend a fixed amount of time on each block (fixed time) or have a fixed number of trials in each block (fixed trials). While fixed-time designs are typically thought to be the most appropriate for optimality studies, as to maximise the number of correct responses participants must optimise RR, our Experiments 1 and 2 indicate that people are at least as good at optimising reward rate under fixed-trial designs as fixed-time designs. However, Experiment 3 provides some evidence that fixed-trial designs with no instructions may not be at least as good as fixed-time designs with very specific instructions. Importantly, these findings challenge the idea that fixed-time designs are the most appropriate for reward rate optimality studies, and further emphasise the importance of carefully considering study design factors when making inferences about optimality in decision-making.
找到一种最优的决策策略需要在准确性和紧迫性这两个相互竞争的需求之间仔细权衡。在实验环境中，研究人员通常感兴趣的是人们是否能够优化这种权衡，通常将其操作化为奖励率，证据积累模型作为确定人们是否表现最优的关键框架。然而，最近的研究表明，关于最优性的推断可能高度依赖于任务设计，这意味着关于人们是否能够实现最优性的推断可能无法在不同情境中普遍适用。在这里，我们研究一个通常被忽视的设计因素：参与者是在每个区块上花费固定的时间（固定时间）还是在每个区块中有固定数量的试验（固定试验次数）。虽然固定时间设计通常被认为最适合最优性研究，因为要最大化正确反应的数量，参与者必须优化奖励率，但我们的实验1和实验2表明，在固定试验次数设计下，人们在优化奖励率方面至少与固定时间设计一样出色。然而，实验3提供了一些证据表明，没有指示的固定试验次数设计可能不如有非常具体指示的固定时间设计。重要的是，这些发现挑战了固定时间设计最适合奖励率最优性研究的观点，并进一步强调了在对决策中的最优性进行推断时仔细考虑研究设计因素的重要性。
