Conflict Paradigms and the Instance Theory of Automatization

本研究将探讨注意在控制和克服因自动处理无关信息而产生的冲突中的作用。来自自动处理的冲突在现实世界中很明显，我们当前的目标与既定习惯背道而驰，比如离开通常的回家路线在商店停下来。Stroop任务提供了一个实验室的例子：要求人们说出呈现单词的颜色，当单词本身命名为冲突的颜色(例如，红色写成绿色)时，必须克服自动阅读反应。Stroop任务的表现及其许多变化(例如，判断3+4=12是否正确)提供了注意力和自动性理论必须解释的主要数据基础。这项研究将把洛根的自动性实例理论应用于冲突范式，该理论是为解释技能习得而发展的。实例理论假设，新手的表现基于执行任务的一般算法，并且随着实践，该算法逐渐被对特定解决方案的记忆所取代。这个学习过程是以精确的数学方式在理论中建模的。实例理论比其他技能获得理论更好地解释了学习，因为它解释了反应时的整个分布的变化，而不仅仅是平均表现。就目前而言，实例理论在冲突范式中做出了错误的预测，预测反应时不会减慢，错误率接近100%，而大量实验表明反应时显著减慢，错误率略有增加。这项研究将发展克服这一问题的理论，并检验新的发展。理论的发展将涉及到假设两种不同的表征，一种是知觉表征，另一种是概念表征，这两种表征通过注意力的过程联系在一起。这项研究将包括三个系列的实验。第一部分将考察将知觉表征与概念表征绑定的过程；第二部分将分离感知表征和概念表征；第三部分将通过检验学习效果来检验绑定过程的长期结果。本研究将通过对学习过程的计算机模拟，将实例理论与其他现有的技能习得理论进行比较，从理论上探讨绑定过程。

This research will investigate the role of attention in controlling and overcoming conflict engendered by automatic processing of irrelevant information. Conflict from automatic processing is apparent in real-world situations in which our current goals run counter to established habits, such as departing from the usual route home to stop at the store. The Stroop task provides a laboratory example: People are asked to name the color in which words are presented and must overcome the automatic reading response when the word itself names a conflicting color (e.g., RED written in green). Performance on the Stroop task and its many variations (e.g., deciding whether 3 + 4 = 12 is true) provides a major data base which theories of attention and automaticity must explain. This research will apply Logan's instance theory of automaticity, developed to account for skill acquisition, to conflict paradigms. The instance theory assumes that novice performance is based on a general algorithm for performing the task and that, with practice, the algorithm is gradually replaced by memory for specific solutions. This learning process is modeled in the theory in a precise mathematical fashion. The instance theory accounts for learning better than other theories of skill acquisition because it accounts for changes in the entire distribution of reaction times, not just average performance. As it stands, the instance theory makes incorrect predictions in conflict paradigms, predicting no slowing of reaction time and an error rate close to 100%, in contrast to vast numbers of experiments which show a substantial slowing in reaction time and a slight increase in error rate. This research will develop the theory to overcome this problem and then test the new developments. Theory development will involve assuming two different representations, a perceptual representation and a conceptual representation, that are bound together by the process of attention. The research will involve three series of experiments. The first will examine the process that binds perceptual representations to conceptual representations; the second will separate perceptual and conceptual representations; and the third will examine long- term consequences of the binding process by examining learning effects. The research will investigate the binding process theoretically using computer simulations of the learning process that compare the instance theory with other current theories of skill acquisition.