Department of Psychiatry, University of California, San Diego, CA, USA.
Neuropsychology. 2013 Mar;27(2):243-55. doi: 10.1037/a0031515.
A fundamental dissociation for most working memory (WM) theories involves the separation of sensory-perceptual encoding of stimulus information from the maintenance of this information. The present paper reports tests of this separability hypothesis for visually presented pseudowords at both mathematical and neuroimaging levels of analysis.
Levels of analysis were linked by two experimental manipulations-visual degradation and pseudoword length variation-that coupled findings from a mathematical modeling study of WM performed in a separate sample to findings from an event-related functional MRI (fMRI) study reported in the present paper. Results from the mathematical modeling study generated parametric signatures of stimulus encoding and WM rehearsal and displacement. These signatures led to specific predictions about neurophysiological responses to study manipulations in a priori regions of interest (ROI).
Results demonstrated predicted dissociations of activation signatures in several ROIs. Significant patterns of brain response mirroring the encode signature were observed only during the task encode interval and only in the visual cortex and posterior fusiform gyrus. In contrast, significant brain response mirroring the rehearsal/displacement signature was observed only in the dorsolateral prefrontal cortex, inferior frontal gyrus, and supramarginal gyrus.
Present findings support the separability hypothesis insofar as brain regions that underlie sensory-perceptual processes demonstrated encode signatures whereas brain regions that support WM maintenance demonstrated the rehearsal/displacement signature. These results also provide evidence for the utility of combining mathematical modeling with fMRI to integrate information across cognitive and neural levels of analysis.
大多数工作记忆 (WM) 理论的一个基本区别是将刺激信息的感觉知觉编码与信息的维持分开。本文报告了在数学和神经影像学分析水平上对视觉呈现的伪词进行这种可分离性假设的测试。
通过两种实验操作——视觉退化和伪词长度变化——将分析水平联系起来，这两种操作结合了在单独样本中进行的 WM 数学建模研究的发现和本文报告的事件相关功能磁共振成像 (fMRI) 研究的发现。数学建模研究的结果生成了刺激编码和 WM 复述和置换的参数特征。这些特征导致了对先验感兴趣区域 (ROI) 中研究操作的神经生理反应的具体预测。
结果表明，在几个 ROI 中观察到激活特征的预测性分离。仅在任务编码间隔期间并且仅在视觉皮层和后梭状回中观察到与编码特征相匹配的大脑反应的显著模式。相比之下，仅在背外侧前额叶皮层、额下回和缘上回中观察到与复述/置换特征相匹配的大脑反应。
目前的发现支持可分离性假设，因为支持感觉知觉过程的大脑区域表现出编码特征，而支持 WM 维持的大脑区域表现出复述/置换特征。这些结果还为结合数学建模和 fMRI 提供了证据，以整合认知和神经分析水平的信息。
