The Relation between First-Order Performance and Metacognitive Judgments in Implicit Learning Research

在内隐学习情境中，对序列的有意识是如何发展的，这是一个重要但尚未解决的问题（例如，Cleeremans，2014; Scott & Dienes，2008; Esser & Haider，提交）。我们以前的研究结果表明，加强内隐获得的表征是不足以解释意识的出现，在一个内隐学习的情况。相反，与当前意识领域中关于高阶思维理论的讨论相一致，我们假设内隐知识是以一阶表征的形式来表征的，而意识表征则以高阶思维或元认知表征的生成为前提。在过去的几年里，一阶表征和元认知表征之间的相互作用在许多不同的领域引发了大量的研究（参见。决策，Fleming & Daw，2017）。这里仍然悬而未决的核心问题是，一阶和高阶判断是否依赖于相同的信号。或者，可以想象，在类似的情况下，以前的经验，另外调制的元认知表征。第一个系列的目的是调查哪些信息被整合到元认知判断。我们假设，在执行当前任务时，参与者开发了一个越来越稳定的元认知模型。该模型不仅包含了一阶性能的信息，而且还包含了类似情况下的经验。本系列的第二个目标是研究如何通过识别预测和经验元认知判断（预测错误）之间的差异来调节当前任务的元认知模型。在这些发现的基础上，我们转向了一个问题，即有意识的意识是如何在内隐学习的情况下出现的。我们假设，一个大的元认知预测误差与一个稳定的元认知模型，导致重新评估的有效性，目前的元认知模型。不是基于预测误差调整当前模型，而是触发用于更好拟合模型的显式搜索过程，这可以导致对所获得的隐式序列知识的有意识洞察。第二组实验的目的是研究预测误差的大小是否以及如何调节关于潜在序列的有意识知识的发展。总的来说，该建议有助于我们理解元认知判断。与通常的二元决策任务相反，内隐学习范式允许研究先前的学习经验如何调节元认知判断。此外，该建议还将有助于我们理解无意识和意识知识之间的相互作用。

How conscious awareness about a sequence develops in an implicit learning situation is an important but not yet solved question (e.g. Cleeremans, 2014; Scott & Dienes, 2008; Esser & Haider, submitted). Our former findings suggest that strengthening of implicitly acquired representations is not sufficient to explain the emergence of consciousness in an implicit learning situation. Rather, and in line with the current discussion in the field of consciousness concerning Higher-Order Thought theories, we assume that implicit knowledge is represented in terms of first-order representations whereas conscious representations presuppose the generation of higher-order thoughts or metacognitive representations. In the last years, the interaction between first-order and metacognitive representations triggered a lot of research in many different domains (cf. decision making, Fleming & Daw, 2017). The central still open question here is whether first-order and higher-order judgments rely on the same signals. Alternatively, it is conceivable that former experiences in similar situations additionally modulate the metacognitive representations.The proposal comprises two series of experiments. The first series is aimed to investigate which information is integrated into metacognitive judgments. We assume that while performing a current task, participants develop an increasingly more stable metacognitive model of the situation. This model not only includes information derived from the first-order performance but also experiences made in similar situations. The second goal of this series then is to examine how metacognitive models of a current task are modulated by recognizing a discrepancy between predicted and experienced metacognitive judgements (prediction error). Building on these findings, we then turn to the question how conscious awareness emerges in an implicit learning situation. We assume that a large metacognitive prediction error in conjunction with a stable metacognitive model leads to a reappraisal of the validity of the current metacognitive model. Instead of adjusting the current model on the basis of the prediction error, an explicit search process for a better fitting model is triggered, which can lead to conscious insight into the acquired implicit sequence knowledge. The goal of this second series of experiments is to investigate whether and how the size of the prediction error modulates the development of conscious knowledge about the underlying sequence. Overall, the proposal contributes to our understanding of metacognitive judgments. Contrary to the usual binary decision tasks, implicit learning paradigms allow to investigate how former learning experiences might modulate the metacognitive judgments. Furthermore, the proposal will also contribute to our understanding of the interaction between unconscious and conscious knowledge.