%A ZHU Shuqing, ZHAI Yu, JIA Shiwei %T Local context dependence in feedback evaluation: An ERP study %0 Journal Article %D 2019 %J Acta Psychologica Sinica %R 10.3724/SP.J.1041.2019.01198 %P 1198-1207 %V 51 %N 11 %U {https://journal.psych.ac.cn/acps/CN/abstract/article_4554.shtml} %8 2019-11-25 %X

Previous research has indicated that feedback evaluation depends on its embedded context, and feedback related negativity (FRN) is also context-dependent. However, it is still disputed whether feedback evaluation merely relies on the global context (the range of all possible outcomes in a block) or also on the local context (the range of possible outcomes on a given trial). Thus, the current study investigated whether the context-dependent effect could be extended from the global context to the local context. We assumed that if feedback evaluation is in a global context-dependent manner, then all non-reward feedback in a given block should elicit a related negativity; however, if feedback evaluation is in a local context-dependent manner, then a related negativity should be discovered for the relative unfavorable feedback compared to the relative favorable feedback in a given trial context.

In this study, twenty-five healthy participants (17 females; Mage = 21.96 years old) performed a cued time-estimation task while EEG was recorded. Gain (a “+” surrounded with a circle)/loss (a “-” surrounded with a circle) cues were presented trial-by-trial at the beginning of each trial to indicate the context of this trial. Participants needed to estimate one second and then received factual feedback in accordance with their performance. In the gain context, correct and incorrect responses resulted in gain (+ ¥4; gain-favorable feedback) and no-gain (¥0; gain-unfavorable feedback), separately. In the loss context, correct and incorrect responses resulted in no-loss (¥0; loss-favorable feedback) and loss (- ¥4; loss-unfavorable feedback), separately.

The behavioral results indicated that larger trial-to-trial changes in time estimation were observed after relative unfavorable feedback than relative favorable feedback. The ERP results revealed that more negative FRN occurred after relative unfavorable feedback than relative favorable feedback in both the gain and loss contexts. In addition, there was no significant difference between gain-favorable and loss-favorable feedback or between gain-unfavorable and loss-unfavorable feedback. In addition, gain-unfavorable feedback elicited a more negative FRN than loss-favorable feedback, although, in fact, both were zero-value feedback.

The above results indicate that feedback evaluation relies on the local context. Based on the previous literature, we infer that the task and facticity of feedback are key factors giving rise to the discrepancies of the results between the previous studies and the current study. In passive tasks (i.e., the participants do not need to react), the good-bad binary evaluation of FRN cannot be observed as well as the context dependence effect. In active tasks with false feedback, global context dependence is found. In active tasks with factual feedback, the context dependence extends to the local level.