Abstract Pattern In Auditory Sequence: An ERP Study

Learning abstract rule plays an important role in identification, classification, ratiocination and other cognitive activities. The core of it is to detect and generalize the rules. Currently, neural mechanism of learning abstract rule is still controversial. Predictive coding is one of the high-profile hypothesis. According to the predictive coding hypothesis, the architecture of the cortex implements a top-down prediction algorithm that constantly anticipates incoming sensory stimuli. This model is used to generate top-down predictions that are compared with novel incoming inputs. Only the difference, as known as the "predictive error", is transmitted to higher cortical stages, where it can be used to adjust the internal model. The local-global paradigm as well as three kinds of abstract rules "xxxx", "xxxY" and "xxx_" are used in this study. We use EEG recording and eye tracking technology to detect the ERP and pupillary dilation to further explore the predictive coding in auditory sequence at abstract level.Behavior results show that participants can all acquire the three abstract rules. EEG results show that MMN, P3a and pattern-based ERP all have a great amplitude increase. Great difference exists among the ERPs of omission X, omission Y and pure omission trials, which indicates a predictive error. Expectations of Y are greater than the expectations of x, which suggests that the brain has different prediction of different rules. Pupil results show that the variation of stimuli can cause significant pupil dilation while standard stimuli are not. The reason is that the participants are expected to standard stimuli.In conclusion, our experimental results show that predictive coding exists in learning abstract rule. We can detect the predictive coding with specific ERPs and pupillary dilation. Our results provide important evidence for the further study about the predictive coding in learning abstract rule.