用PDP模型模拟Stroop任务的实验研究

本实验用基于ＰＤＰ原理构造的Ｓｔｒｏｏｐ任务模型模拟了Ｓｔｒｏｏｐ颜色－词的干扰现象。模型采用ＢＰ学习算法。经过训练，模型学会字词阅读和颜色命名。由于对模型采用６种不同的学习次数（这里为：５００，１０００，２０００，３０００，５０００，１００００）以及４种不同比例（这里为：４：１，６：１，８：１，１０：１）字词模式和颜色模式的训练，其结果字词通道的信息加工比颜色通道的加工更快更强，并产生不同的干扰模式，其字词加工的次数越多，干扰越大，次数越少，干扰也越小。这很可能说明语言的熟练程度对干扰模式有着更大的影响。模型还体现了加工过程的连续性质。但这个模型难以区分不同的加工任务。

A STUDY ON SIMULATION OF STROOP TASK BASED ON PDP PRINCIPLE

The experiment simulated performance in the Stroop task with a model of the Stroop color-word task based on PDP principle. The result showed that processing of information in the word pathway was faster and stronger than in the color pathway after training on the word-reading and color-naming tasks. This might be caused by the frequency effect of word training(the ratio between words and colors are 10: 1 8:1, 6: 1, 4: 1). The more word training times given, the stronger the interferences appeared The model showed the continuous nature of automatic processing. The simulated result suggests that the patterns of between-and within-language interference are determined more probably by language proficiency, but it is difficult for the model to distinguish between different processing tasks.