A Stochastic Judgment Model of Recall: Separating Measurement, Memory and Correla

描述（由申请人提供）：人类的记忆可以分为识别（“那是我的车吗？”）而不是回忆（“我把车停在哪里了？”）。对识别记忆的研究，比如在警察队伍中识别个人，常常借助于信心评级的收集（“我有95%的把握那就是偷我车的人”）。相比之下，回忆传统上被认为是一个高度准确的要么全有要么全无的过程（即从零开始创建犯罪素描）。然而，最近的实验表明，回忆往往是错误的，也许最好把它看作是一个类似于识别的分级连续体。在我们提出的研究中，我们试图通过收集回忆反应和各种信心评级来调查回忆的分级性质。在某些情况下，我们将在学习之后，但在回忆之前收集这些信心评级（前瞻性信心），而在其他情况下，我们将在回忆响应之后收集这些信心评级（回顾性信心）。我们开发了一种基于信号检测理论的数学工具，使我们能够使用这些置信度评级来确定1)作为回忆基础的记忆强度的分级性质；2)与使用信心量表相关的问题（例如，有些人在确定性上是保守的，而另一些人是自由的）；3)与实际记忆强度相比，信心基于不同因素的程度。这项工作很重要，不仅因为它将促进对回忆的更全面的理解，而且还因为它将决定在什么情况下信心预测或表明准确的回忆。就前瞻性预测回忆而言，这对涉及自主学习的教育环境具有重要意义。就回顾性地表明回忆而言，对法律环境和涉及目击者证词的其他情况有重要影响。最后，这些技术和结果不仅对评估正常成年人的记忆很重要，而且对已经证明可以将回忆表现与其他形式的记忆分离开来的患者群体和药物治疗也很重要。只有通过收集信心评级和应用我们正在开发的数学工具，才能确定这些分离是否反映了潜在记忆强度的真正差异，或者它们是否反映了信心和记忆访问的损害。公共健康相关性：记忆障碍传统上是通过无法回忆以前研究过的项目来衡量的，但这些缺陷可能反映了与潜在反应相关的确定性的变化，而不是潜在记忆强度的缺陷。通过收集信心判断和回忆反应，本研究将开发一套测量工具来区分回忆不足的可能来源。这些测量工具将与记忆障碍的鉴别诊断以及旨在减轻记忆问题的临床治疗相关。

DESCRIPTION (provided by applicant): Human memory can be separated into recognition ("Is that my car?") versus recall ("Where did I park my car?"). The study of recognition memory, such as in identifying individuals in a police lineup, is often aided by the collection of confidence ratings ("I'm 95 percent sure that's the guy who stole my car"). In contrast, recall is traditionally viewed as a highly accurate all or none process (i.e., creating a criminal sketch from scratch). However, recent experiments demonstrate that recall is often fallible, and is perhaps better viewed as lying along a graded continuum similar to recognition. In the proposed studies, we seek to investigate this graded nature of recall by collecting not only recall responses, but also various kinds of confidence ratings. In some cases we will collect these confidence ratings after learning, but before recall (prospective confidence), while in other cases we will collect these confidence ratings following recall responses (retrospective confidence). We have developed a mathematical tool based on signal detection theory that allows us to use these confidence ratings to determine 1) the graded nature of the memory strength that underlies recall; 2) issues related to use of the confidence scale (e.g., some people are conservative while others are liberal in their certainty); and 3) the extent to which confidence is based on different factors than actual memory strength. This work is important not only because it will foster a fuller understanding of recall, but additionally because it will determine under what circumstances confidence predicts or indicates accurate recall. In terms of prospectively predicting recall, there are important implications for educational settings involving self-paced learning. In terms of retrospectively indicating recall, there are important implications for legal settings and other situations involving eyewitness testimony. Finally, the techniques and results will be important not only for assessing memory in normal adults, but also in terms of patient populations and drug treatments that have been demonstrated to dissociate recall performance from other forms of memory. Only by collecting confidence ratings and applying the mathematical tools that we are developing can it be determined whether these dissociations reflect real differences in the underlying memory strength, or whether they instead reflect impairments in confidence and memory access. PUBLIC HEALTH RELEVANCE: Memory disorders are traditionally measured through the inability to recall previously studied items but these deficits may reflect a change in the certainty associated with potential responses rather than a deficit in the underlying memory strength. By collecting confidence judgments as well as recall responses, the proposed work will develop a set of measurement tools that can differentiate between the possible sources of a recall deficit. These measurement tools will be relevant to differential diagnoses of memory disorders as well as clinical therapies designed to alleviate memory problems.