Abstract
Learning to adapt behavior based on changes in the visual environment is a hallmark of human visual cognition and implicated in a variety of neuropsychiatric disorders. Here, we examined the role of different frontal oscillatory mechanisms using high-density electroencephalography (EEG). Human subjects performed a probabilistic cognitive reinforcement learning task that allowed us to distinguish individuals based on their motivational baises to learn by avoiding negative feedback versus pursuing positive outcomes. We found that negative feedback related theta power (4 - 8 Hz) in medial-frontal regions was stronger in negative learners, whereas positive feedback related beta power (18 - 24 Hz) in ventral-medial regions was stronger in positive learners. Brain-behavior correlations showed that individual preferential biases to learn more from negative than positive feedback were associated with relatively more medial-frontal theta power, whereas biases to learn more from positive than negative feedback were related to relatively greater ventral-medial beta power. To further examine whether mechanisms of feedback valence were distinguished in frequency-brain space, we sought to manipulate the medial-frontal theta effect using 4 Hz high definition transcranial alternating current stimulation (HD-tACS). This noninvasive stimulation selectively disrupted medial-frontal theta and negative-learning behavior, leading to performance impairments in negative learners. However, no such stimulation-induced changes were observed in ventral-medial beta rhythms or positive-learning behavior, resulting in the stimulation having no significant impact on individuals who learned by pursuing positive outcomes. The results provide new evidence for theories in visual cognition that propose separate neurophysiological mechanisms underlie choice-behavior and learning guided by visual information from the environment. The results also show that alternating current can be used to manipulate the signal power of ongoing neural oscillations in frontal cortex important for executive control and decision-making.
Meeting abstract presented at VSS 2017