| The thesis reports on four related studies focusing on the development of intention understanding and its related neural process, especially when unexpected outcomes occur (i.e. predictions error occur).From a developmental view, the study reported in chapter two focused on children's attribution of their own intention to unexpected action outcomes. The two critical conditions were:(1) a successful intended action brought about an undesired outcome, or (2) a failed action brought about a desired outcome. Children played two games in which outcomes could mysteriously change to produce unintended outcomes (i.e., causes were unplanned). Children sometimes received a desirable reward, independent of whether their original intention succeeded. Four-and Fiver-year-olds mistakenly claimed they had intended the desirable outcome even when it was unpredicted. Four-year-olds judged that they had not intended a deliberate action, if it did not yield a rewarding outcome. The results show children found difficult to correctly attribute their own intentions in both of the unexpected conditions, especially when positive prediction error occurred (unexpected positive outcomes).To investigate the prediction error in children's intention attribution, we also looked at adults' intention attribution to see if there was a similar effect. The study reported in chapter three tested how adults attributed other people's intention in the following situation:"In order to get A, the agent did action B which would also cause a side effect C". Previous studies had shown that participants were more likely to say the agent was intended to cause C when C is negative, as compare to positive. This phenomenon was called "side-effect effect". To explain this, we proposed the hypothesis "predictions error account" which states that for each event that we encounter, we automatically built up internal expectations of outcomes, and unconsciously compare it to the actual outcome when it occurs. If there is a deviation between them, we have a tendency to attribute an agent responsible for it. The bigger the prediction error, the more likely we tend to say the agent intentionally caused outcome C. According to this account, how much the participants care about outcome C is a critical factor in determining their expectations of the outcome C. The more they care about outcome C, the higher their expectations are, thus when a negative outcome occurs, the greater the prediction errors are, and the more likely they would claim that the agent intentionally caused outcome C. To test this hypothesis, in two experiments, we manipulate participants' importance level of the outcome C. The results from both experiments consistently showed that the more the participants care about C. the more they claimed the agent intentionally caused it. The results provide supportive evidence to the "prediction error account" hypothesis and suggests that as in children, the prediction error also affects a mature mind in attributing intentions.Driven by the finding from the previous two studies that prediction error has an impact on people's intention attribution, using ERP technic, the neural processing of the critical factor "prediction error" is a focus on the study reported in chapter 4. Based previous studies, feedback-related negativity (FRN) is an event related potential (ERP) component related to reward prediction error. However, it is debating whether positive prediction error and negative prediction error elicit the same or different neural process. To clarify this, we employed a modified gambling task, in which probabilistic cues were provided to elicit positive or negative expectations about an impending outcome. ERPs to "final outcome" and "probabilistic cues" were analyzed. The results showed that unexpected negative outcome elicited larger FRN than unexpected positive outcomes, suggesting different neural processes for these two different type of prediction errors. We also found that predictions error generated by probabilistic cues also elicited FRN.Different neural processes of positive and negative prediction error revealed by the above mentioned study lead us a better understanding of its role in modulating intention attribution. It is however,1) not a direct measure of a cognitive process that involves understanding of intention; 2) not providing experimental data from a developmental view. Motivated by these reason, in the study reported in chapter 5 we started exploring children's neural mechanism of intention understanding by reading their electroencephalography (EEG). However, the prevalent EEG technic (mainly ERP) largely limits any body movement that could occur in experiment design. The constrains of the technique also prevent developmental research from widely exploring children's social brain as it is hard to keep children to sit still. Facing this problem, this study focused on two main goals:1) developing an EEG data acquisition-analyzing system that would allow broader use on infant and children's EEG studies; 2) following the prevalent hypothesis that mirror neuron system provides the neural basis of understanding other's action in terms of goals and intentions, we tested young children's mu rhythm activities while observing other people's actions and while executing their own action. The task required children and parents to play a face to face turn taking interactive bubble popping game on a touch screen. Both observation and execution trials were generated. To meet the first goal, we developed a testing and data recording environment that allows children to play video games on a touch screen with their parent while all their behavioral, motion as well as EEG data are synchronized and recorded. Using independent component analysis, we were able to pull out the left hemisphere sensorimotor mu component and evaluate its activity according to the experimental conditions. The results provide supportive evidence to the hypothesis that the mirror neuron system is involved in the understanding of action intention. We observed mu suppression on both observation and execution epochs, compared to the resting epochs (i.e. baseline),In conclusion:1) Four-and Five-year-olds found it is difficult to correctly attribute their own action intentions when the outcomes were unexpected which imply prediction error. It is even more difficult for them if the unexpected outcomes brought positive rewards2) Adults also seem to be affected by prediction errors when attributing other's intentions. For a negative outcome, the more the participants care about it and expect it to be good, the more likely they would think other people intentionally caused it.3) Our brain reacts differently to positive prediction errors and negative prediction errors. Negative prediction error is associated with larger feedback-related negativity while positive prediction error is associated with negative prediction error.4) Mu rhythm activity is related to the understanding of intention of actions. The data shows a desynchronization while children observe other people executing an action, as well as while doing the similar action themselves.
|
PDF Full Text Request