Department of Psychology, Brain & Cognition Unit, Saarland University, Saarbruecken, Germany.
Neurosci Biobehav Rev. 2010 Jun;34(7):1066-79. doi: 10.1016/j.neubiorev.2010.01.014. Epub 2010 Feb 6.
We present a neurocognitive model of long-term object memory. We propose that perceptual priming and episodic recognition are phenomena based on three distinct kinds of representations. We label these representations types and tokens. Types are prototypical representations needed for object identification. The network of non-arbitrary features necessary for object categorization is sharpened in the course of repeated identification, an effect that we call type trace and which causes perceptual priming. Tokens, on the other hand, support episodic recognition. Perirhinal structures are proposed to bind intrinsic within-object features into an object token that can be thought of as a consolidated perceptual object file. Hippocampal structures integrate object- with contextual information in an episodic token. The reinstatement of an object token is assumed to generate a feeling of familiarity, whereas recollection occurs when the reinstatement of an episodic token occurs. Retrieval mode and retrieval orientation dynamically modulate access to these representations. In this review, we apply the model to recent empirical research (behavioral, fMRI, and ERP data) including a series of studies from our own lab. We put specific emphasis on the effects that sensory features and their study-test match have on familiarity. The type-token approach fits the data and additionally provides a framework for the analysis of concepts like unitization and associative reinstatement.
我们提出了一个关于长期物体记忆的神经认知模型。我们认为，知觉启动和情景识别是基于三种不同类型的表示的现象。我们将这些表示类型标记为类型和标记。类型是用于对象识别的原型表示。在反复识别的过程中，对象分类所需的非任意特征网络会得到加强，这种效应我们称之为类型痕迹，它导致了知觉启动。另一方面，标记支持情景识别。我们提出，边缘结构将内在物体特征绑定到对象标记中，可以将其视为一个整合的知觉物体文件。海马结构在情景标记中整合物体与上下文信息。假设对象标记的恢复会产生熟悉感，而当情景标记的恢复发生时，就会出现回忆。检索模式和检索方向动态调节对这些表示的访问。在这篇综述中，我们将该模型应用于最近的实证研究（行为、fMRI 和 ERP 数据），包括我们自己实验室的一系列研究。我们特别强调了感觉特征及其学习-测试匹配对熟悉感的影响。类型-标记方法适用于数据，并且还为分析像单元化和联想恢复这样的概念提供了一个框架。
