Surprise signals in the supplementary eye field: rectified prediction errors drive exploration-exploitation transitions.
نویسندگان
چکیده
Visual search is coordinated adaptively by monitoring and predicting the environment. The supplementary eye field (SEF) plays a role in oculomotor control and outcome evaluation. However, it is not clear whether the SEF is involved in adjusting behavioral modes based on preceding feedback. We hypothesized that the SEF drives exploration-exploitation transitions by generating "surprise signals" or rectified prediction errors, which reflect differences between predicted and actual outcomes. To test this hypothesis, we introduced an oculomotor two-target search task in which monkeys were required to find two valid targets among four identical stimuli. After they detected the valid targets, they exploited their knowledge of target locations to obtain a reward by choosing the two valid targets alternately. Behavioral analysis revealed two distinct types of oculomotor search patterns: exploration and exploitation. We found that two types of SEF neurons represented the surprise signals. The error-surprise neurons showed enhanced activity when the monkey received the first error feedback after the target pair change, and this activity was followed by an exploratory oculomotor search pattern. The correct-surprise neurons showed enhanced activity when the monkey received the first correct feedback after an error trial, and this increased activity was followed by an exploitative, fixed-type search pattern. Our findings suggest that error-surprise neurons are involved in the transition from exploitation to exploration and that correct-surprise neurons are involved in the transition from exploration to exploitation.
منابع مشابه
Title : 1 Surprise Signals in the Supplementary Eye Field : Rectified Prediction Errors
1 Surprise Signals in the Supplementary Eye Field: Rectified Prediction Errors Drive 2 Exploration-Exploitation Transitions 3 4 Authors: 5 Norihiko Kawaguchi, Kazuhiro Sakamoto, Naohiro Saito, Yoshito Furusawa, Jun Tanji, 6 Masashi Aoki, and Hajime Mushiake 7 8 Affiliations: 9 Department of Physiology, Tohoku University School of Medicine, 2-1 Seiryo-machi, Aoba-ku, 10 Sendai, 980-8575, Japan 1...
متن کاملCortical Signals for Rewarded Actions and Strategic Exploration
In stable environments, decision makers can exploit their previously learned strategies for optimal outcomes, while exploration might lead to better options in unstable environments. Here, to investigate the cortical contributions to exploratory behavior, we analyzed single-neuron activity recorded from four different cortical areas of monkeys performing a matching-pennies task and a visual sea...
متن کاملFrontal theta reflects uncertainty and unexpectedness during exploration and exploitation.
In order to understand the exploitation/exploration trade-off in reinforcement learning, previous theoretical and empirical accounts have suggested that increased uncertainty may precede the decision to explore an alternative option. To date, the neural mechanisms that support the strategic application of uncertainty-driven exploration remain underspecified. In this study, electroencephalograph...
متن کاملMemory and Decision Making in the Frontal Cortex during Visual Motion Processing for Smooth Pursuit Eye Movements
Cortical motor areas are thought to contribute "higher-order processing," but what that processing might include is unknown. Previous studies of the smooth pursuit-related discharge of supplementary eye field (SEF) neurons have not distinguished activity associated with the preparation for pursuit from discharge related to processing or memory of the target motion signals. Using a memory-based ...
متن کاملAn Accumulative Exploration Method for Reinforcement Learning
Agents in Multi Agent Systems can coordinate their actions by communicating. We investigate a minimal form of communication, where the signals that agents send represent evaluations of the behavior of the receiving agent. Learning to act according to these signals is a typical Reinforcement Learning problem. The backpropagation neural network has been used to predict rewards that will follow an...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- Journal of neurophysiology
دوره 113 3 شماره
صفحات -
تاریخ انتشار 2015