Artificial Grammar Implicit Learning New Exploration

The exploration of implicit learning (IL) lists among the most important themes in cognitive psychology. The most dominant research paradigm of the IL research is Artificial Grammar Learning (AGL) with the Finite-State Grammars (FSG). There are two kernel questions about the AGL: whether the knowledge is acquired implicitly and whether such knowledge is abstract. Till now, psychologists haven't reached consensus about the nature of the AGL, especially whether or not the knowledge acquired is really abstract. Based on the analysis of previous researches, the author have gained new comprehension of the nature of the FSG that the rules of the FSG are accumulative concepts and such rules cannot be completely- expressed through the AGL paradigm. After analyzing the configuration of the strings used in learning as well as testing stages of the former researches, the author proposes that, under present AGL paradigm, the subjects may not be able to acquire the exact rules designed by the researchers.In this thesis, the author attempts to attest the proposed flews in the previous researches through three experiments. The first experiment revolves around the configuration of learning strings in the AGL, and the results show that the performance of the subjects is influenced by the asymmetry of the strings distributed among different paths of the FSG Specifically, subjects performed relatively poorly on the paths whose exemplar strings don't appear in the learning stage. The second experiment examines the setting of the testing strings and the results show that the subjects' performance can also be affected by (1) the asymmetric distribution pattern of the testing strings, and (2) the compatibility between the distribution patterns of the learning strings and testing strings. The last experiment manipulates the location of error letters among the testing strings and found that the subjects' performance can be influenced by the distribution patterns of error strings generated in different ways.In general, the thesis reaches the conclusion that the present AGL paradigm is inadequate with respect to the generation method and distribution pattern of the exemplar strings. With the improvement of the research design, it will benefit the understanding of the Artificial Grammar Learning and the implicit learning as a whole.