Cognitive skill learning postulates strategy generation and implementation when people learn to perform new tasks. Here we followed self-reported strategy use and objective performance in a working memory (WM) updating task to reveal strategy development that should take place when faced with this novel task. In two pre-registered online experiments with healthy adults, we examined short-term strategy acquisition in a ca 20–30-minute adaptive n-back WM task with 15 task blocks by collecting participants' strategy reports after each block. Experiment 1 showed that (a) about half of the participants reported using a strategy already during the very first task block, (b) changes in selected strategy were most common during the initial task blocks, and (c) more elaborated strategy descriptions predicted better task performance. Experiment 2 mostly replicated these findings, and it additionally showed that compared to open-ended questions, the use of repeated list-based strategy queries influenced subsequent strategy use and task performance, and also indicated higher rates of strategy implementation and strategy change during the task. Strategy use was also a significant predictor of n-back performance, albeit some of the variance it explained was shared with verbal productivity that was measured with a picture description task. The present results concur with the cognitive skill learning perspective and highlight the dynamics of carrying out a demanding cognitive task.

当人们学会执行新任务时，认知技能学习会假定策略的产生和实施。在这里，我们在工作记忆（WM）更新任务中跟踪了自我报告的策略使用和目标绩效，以揭示应对这一新颖任务时应进行的策略制定。在与健康成年人进行的两个预先注册的在线实验中，我们通过在每个区块之后收集参与者的策略报告，检查了大约20-30分钟的自适应n-back WM任务（包含15个任务块）中的短期策略。实验1显示：（a）大约一半的参与者报告说，在第一个任务块中就已经使用了一种策略；（b）在初始任务块中，最常见的是所选策略的更改；（c）更为详尽的策略描述预测得更好任务绩效。实验2主要重复了这些发现，并且还表明，与开放式问题相比，基于列表的重复策略查询的使用会影响后续策略的使用和任务绩效，并且还表明任务期间策略实施和策略变更的发生率更高。策略的使用也是n-back性能的重要预测指标，尽管它所解释的某些差异与通过图片描述任务衡量的口头生产率是相同的。目前的结果符合认知技能学习的观点，并突出了执行一项艰巨的认知任务的动力。并且还表明任务期间执行策略和更改策略的比率更高。策略的使用也是n-back性能的重要预测指标，尽管它所解释的某些差异与通过图片描述任务衡量的口头生产率是相同的。目前的结果符合认知技能学习的观点，并突出了执行一项艰巨的认知任务的动力。并且还表明任务期间执行策略和更改策略的比率更高。策略的使用也是n-back性能的重要预测指标，尽管它所解释的某些差异与通过图片描述任务衡量的口头生产率是相同的。目前的结果符合认知技能学习的观点，并突出了执行一项艰巨的认知任务的动力。