Department of Psychology, University of Victoria, Victoria, British Columbia, Canada.
Department of Experimental Psychology, Ghent University, Ghent, Belgium.
Psychophysiology. 2022 Jun;59(6):e14004. doi: 10.1111/psyp.14004. Epub 2022 Feb 19.
The reinforcement learning (RL) theory of the reward positivity (RewP) proposes that RewP indexes a reward prediction error (RPE) signal processed in the anterior cingulate cortex (ACC). According to this theory, RewP is an event-related potential (ERP) that is more positive-going for feedback stimuli that predict better-than-expected outcomes (positive feedback) than for feedback stimuli that predict worse-than-expected outcomes (negative feedback). Despite strong evidence for this hypothesis, findings have been equivocal for tasks involving painful outcomes. We hypothesized that the RewP is modulated by high-level task goals such that outcomes that are congruent with the goals elicit positive RPEs even if their immediate consequences are negative. Accordingly, changes in high-level task goals should modulate RewP amplitude for tasks that involve seeking pain compared to tasks that involve avoiding pain. We recorded the electroencephalogram from participants who were instructed to navigate a virtual T-Maze to find reward-predictive feedback in a reward condition and pain-predictive feedback in a pain condition. We expected more positive-going ERPs to reward feedback in the reward condition and more positive-going ERPs to pain feedback in the pain condition. Despite behavioral results indicating that participants complied with task instructions, contrary to our predictions, we did not find a RewP to pain feedback. We suggest that pain feedback interfered with the effect of high-level task goals on RewP amplitude, which is indicative of conflict at different levels of task hierarchy.
强化学习（RL）理论的奖励正波（RewP）提出，RewP 指数化了在前扣带皮层（ACC）中处理的奖励预测误差（RPE）信号。根据这一理论，RewP 是一种事件相关电位（ERP），对于预测优于预期结果（正反馈）的反馈刺激，RewP 更为正向，而对于预测差于预期结果（负反馈）的反馈刺激则为负向。尽管有强有力的证据支持这一假设，但对于涉及疼痛结果的任务，研究结果一直存在争议。我们假设，RewP 受到高级任务目标的调节，即与目标一致的结果会引发积极的 RPE，即使它们的直接后果是负面的。因此，对于涉及寻求疼痛的任务与涉及避免疼痛的任务相比，高级任务目标的变化应该会调节 RewP 幅度。我们记录了参与者的脑电图，他们被指示在一个奖励条件下导航一个虚拟 T 型迷宫以寻找奖励预测性反馈，在一个疼痛条件下导航一个虚拟 T 型迷宫以寻找疼痛预测性反馈。我们预计在奖励条件下，奖励反馈会产生更正向的 ERP，而在疼痛条件下，疼痛反馈会产生更正向的 ERP。尽管行为结果表明参与者遵守了任务指令，但与我们的预测相反，我们没有发现疼痛反馈的 RewP。我们认为，疼痛反馈干扰了高级任务目标对 RewP 幅度的影响，这表明在不同层次的任务层次结构中存在冲突。
