Long-term memory supports the retention, preservation, and prioritization of short-term memory.

摘要

This thesis describes a series of experiments that have effected several empirical tests of the theoretical idea that the short-term retention of information is accomplished by the re-activation of long-term memory (LTM) representations. Short-term memory (STM) refers to the retention of information in an active state when this information is not present in the environment. Working memory (WM) refers to the ability to manipulate or otherwise transform STM information, to protect it in the face of interference, and to use it to guide high-level behaviors. STM and WM are of central importance in the study of high-level cognition because they have been implicated as critical contributors to such essential cognitive functions and properties as language comprehension, learning, planning, reasoning, and general fluid intelligence. This research leveraged a powerful new information-based analysis technique for neuroimaging data -- multi-voxel pattern analysis of functional magnetic resonance imaging (fMRI) data -- in a way that enabled the direct assessment of physiological correlates of the concept of activated LTM representations. The method allowed for the precise tracking of actual information representations in the brain, not simply elevated activity assumed to correspond to information. The strongest criticism of the activated-LTM model has been that it fails to account for the flexibility & dynamism of WM, but the results of this thesis provide converging empirical evidence that the re-activation of LTM representations supports robust and efficient information processing in the human brain. Information-based fMRI analysis revealed, in a way that was not possible before, that: (1) persistent brain activity many not be the neural instantiation of STM but instead may reflect the moment-to-moment contents of attention; (2) the theoretical model of activated-LTM needs to be refined in order to account for processes required for complex human behavior (e.g., preservation and prioritization); (3) parallel encoding (in multiple representational formats) and recoding many indeed by the basis of STM. I therefore contend that activated LTM should be considered a critical component to the broad range of cognitive processes essential for high-level behaviors.