dc.description.abstract | Statistical learning is an important ability in the human life, which can help us extract necessary information from the environment effectively and further transfer the information to be specific rules for living. Therefore, an increasing number of studies have paid much attention on this aspect of research to explore the nature of statistical learning ability. Previous studies have shown that when there are certain relationships embedded in the information, human beings could learn the complex and innate rules from the information by using statistical learning ability. Moreover, the process underlying such learning has been proposed to be largely implicit and unconscious to participants. When the information is perceived by the brain through different sensory modalities, it is firstly transferred to form inner representations, and then is organized and processed by the brain for the extraction of the input information. Such process to extract the specific rules without awareness has been claimed to rely on individual ability of statistical learning. Besides, there is significant individual difference during processing different sensory inputs. Therefore, the specificity of statistical learning ability within the same learners cannot be ignored. It seems that even the same learners process different sensory materials or learn different kinds of rules, the performance based on the statistical learning ability is still different. However, most of the previous studies mainly focused on statistical learning of the dependency of adjacent items, while few studies explored the statistical learning ability of dependency of non-adjacent items, which is also frequent in the information in the real world. Given such background, the present study attempted to design new statistical learning tests to measure the non-adjacent statistical learning ability based on the previous tests of adjacent statistical learning and artificial-grammar learning. This series of tests not only include the visual version but also the auditory version to compare the levels of statistical learning ability when processing adjacent and non-adjacent materials in different modalities. Besides, other tests of general cognitive abilities, including an IQ test, a verbal working memory test, a visual spatial working memory test and an executive function test, have also been applied to observe these cognitive abilities across different learners, and to explore whether the statistical learning ability is independent of the other cognitive functions. The main aims of this study is to explore 1) the individual difference of the different aspects of statistical learning ability and 2) the relationship between the different statistical learning abilities learning different types of rules or from different modalities within the same group of participants. The results showed that the learners could not only learn the adjacent rules but also the non-adjacent rules with enough inputs. Moreover, the performance of statistical learning was affected by different sensory inputs, as reflected by uncorrelated performance in the visual and auditory modalities, especially when the rule to be learned was dependency among adjacent items. Interestingly, the performance of statistical learning in the visual and auditory modalities was correlated significantly when the rule to be learned as dependency among non-adjacent items. In sum, the findings in the present study further support the current theoretical model in demonstrating modality specificity in statistical learning ability. | en_US |