Gomes C F A, Brainerd C J, Nakamura K, Reyna V F
Department of Human Development, Cornell University, Ithaca, NY 14853, telephone: 607- 793-6099,
Department of Human Development, Cornell University, Ithaca, NY 14853, telephone: 607-255-2592,
J Math Psychol. 2014 Apr 1;59:50-64. doi: 10.1016/j.jmp.2013.07.003.
A half-century ago, at the dawn of the all-or-none learning era, Estes showed that finite Markov chains supply a tractable, comprehensive framework for discrete-change data of the sort that he envisioned for shifts in conditioning states in stimulus sampling theory. Shortly thereafter, such data rapidly accumulated in many spheres of human learning and animal conditioning, and Estes' work stimulated vigorous development of Markov models to handle them. A key outcome was that the data of the workhorse paradigms of episodic memory, recognition and recall, proved to be one- and two-stage Markovian, respectively, to close approximations. Subsequently, Markov modeling of recognition and recall all but disappeared from the literature, but it is now reemerging in the wake of dual-process conceptions of episodic memory. In recall, in particular, Markov models are being used to measure two retrieval operations (direct access and reconstruction) and a slave familiarity operation. In the present paper, we develop this family of models and present the requisite machinery for fit evaluation and significance testing. Results are reviewed from selected experiments in which the recall models were used to understand dual memory processes.
半个世纪前，在全或无学习时代的开端，埃斯蒂斯表明，有限马尔可夫链为他在刺激抽样理论中设想的条件作用状态转变这类离散变化数据提供了一个易于处理的综合框架。此后不久，这类数据在人类学习和动物条件作用的许多领域迅速积累，埃斯蒂斯的工作激发了马尔可夫模型的蓬勃发展以处理这些数据。一个关键成果是，情景记忆、识别和回忆的主要范式的数据，经证明分别非常近似于一阶段和两阶段马尔可夫过程。随后，识别和回忆的马尔可夫建模几乎从文献中消失，但现在随着情景记忆的双过程概念又重新出现。特别是在回忆方面，马尔可夫模型正被用于测量两种检索操作（直接访问和重建）以及一种从属的熟悉度操作。在本文中，我们开发了这一系列模型，并给出了拟合评估和显著性检验所需的方法。我们回顾了一些选定实验的结果，在这些实验中，回忆模型被用于理解双重记忆过程。
