A Distributed Representation of Internal Time
A Distributed Representation of Internal Time
复制标题
DOI：
10.1037/a0037840
复制
发表时间：
2015-01-01
影响因子：
5.4
通讯作者：
Criss, Amy H.
中科院分区：
文献类型：
--
作者：
Howard, Marc W.;Shankar, Karthik H.;Criss, Amy H.
这篇文章追求的假设，规模不变的代表性的历史可以支持性能在各种学习和记忆任务。这种表征保持了对所发生的事情的联合表征，当它对过去越来越远的事件越来越不准确时。简单的行为模型使用几个操作，包括扫描，匹配和恢复历史的先前状态的“时间跳回”，描述了一系列行为现象。这些行为应用包括近因任务在短尺度和长尺度上的判断的典型结果，情节回忆中跨尺度的近因和邻近效应，以及条件反射中的时间映射现象。越来越多的神经数据表明，几个大脑区域的神经表征具有由时间历史表征预测的定性特性。综上所述，这些结果表明，时间历史的尺度不变表示可能作为学习和记忆中认知的物理模型的基石。
This article pursues the hypothesis that a scale-invariant representation of history could support performance in a variety of learning and memory tasks. This representation maintains a conjunctive representation of what happened when that grows continuously less accurate for events further and further in the past. Simple behavioral models using a few operations, including scanning, matching and a "jump back in time" that recovers previous states of the history, describe a range of behavioral phenomena. These behavioral applications include canonical results from the judgment of recency task over short and long scales, the recency and contiguity effect across scales in episodic recall, and temporal mapping phenomena in conditioning. A growing body of neural data suggests that neural representations in several brain regions have qualitative properties predicted by the representation of temporal history. Taken together, these results suggest that a scale-invariant representation of temporal history may serve as a cornerstone of a physical model of cognition in learning and memory.