A Bayesian model for implicit effects in perceptual identification.
A Bayesian model for implicit effects in perceptual identification.
复制标题
DOI：
10.1037/0033-295x.108.1.257
复制
发表时间：
2001
影响因子：
5.4
通讯作者：
L. Schooler;R. Shiffrin;J. Raaijmakers
中科院分区：
文献类型：
--
作者：
L. Schooler;R. Shiffrin;J. Raaijmakers
有效地从记忆中提取（REM; R。M. Shiffrin & M. Steyvers，1997），一个情景记忆模型，被扩展到内隐记忆现象，即知觉识别研究报告R。拉特克利夫和G. McKoon（1997年）。在这些研究中，当单词呈现得最简短，强迫选择的目标和陪衬最相似时，先前研究的影响最大。R.拉特克利夫和G. McKoon用这些数据来反驳先前的研究改变了一个词的表征的模型。一个模型，在该模型中，先前的研究通过添加上下文信息来改变单词的表示，以适应他们的数据;在测试中，该模型使用贝叶斯决策过程来比较与测试闪光相关联的感知和上下文特征与存储的痕迹。先前研究的影响是由于匹配额外的上下文信息，并且当替代方案共享许多特征时，先前研究的影响更大，从而减少了减弱这些影响的噪声。
Retrieving effectively from memory (REM; R. M. Shiffrin & M. Steyvers, 1997), an episodic model of memory, is extended to implicit memory phenomena, namely the perceptual identification studies reported in R. Ratcliff and G. McKoon (1997). In those studies, the influence of prior study was greatest when words were presented most briefly and when forced-choice targets and foils were most similar. R. Ratcliff and G. McKoon use these data to argue against models in which prior study changes a word's representation. A model in which prior study changes a word's representation by adding context information is fit to their data; at test, the model uses a Bayesian decision process to compare the perceptual and context features associated with the test flash to stored traces. The effects of prior study are due to matching extra context information and are larger when alternatives share many features, thereby reducing noise that attenuates these effects.