• Top page
  • Timetable
  • Per session
  • Per presentation
  • How to
  • Meeting Planner



報酬 2
Reward 2

開催日 2014/9/12
時間 16:00 - 17:00
会場 Room H(304)
Chairperson(s) 坂上 雅道 / Masamichi Sakagami (玉川大学脳科学研究所 / Tamagawa University, Brain Science Institute, Japan)
設楽 宗孝 / Munetaka Shidara (筑波大学医学医療系 / University of Tsukuba, Faculty of Medicine, Japan)

Weighting of 'action value' and switching of 'internal state' by secondary reinforcer in blindsight monkeys

  • O2-H-4-3
  • 加藤 利佳子 / Rikako Kato:1 高桑 徳宏 / Norihiro Takakuwa:1,2 Zeghbib Abdelhafid / Abdelhafid Zeghbib:3 Redgrave Peter / Peter Redgrave:3 伊佐 正 / Tadashi Isa:1,2 
  • 1:生理学研究所 認知行動発達機構研究部門 / Dept Dev Physiol, Nat Inst Physiol Sci, Okazaki, Japan 2:総合研究大学院大学 / Graduate Univ for Advanced Stud, Hayama, Japan 3:Dept Psychol, Univ of Sheffield, Sheffield, UK / Dept Psychol, Univ of Sheffield, Sheffield, UK 

Basic algorism of reinforcement learning is weighting of action values before appearance of reinforcer. It is suggested that this fundamental learning system for adaptation to environment contributes both to formation of goal-directed and automatic habitual control. However, the role of subjective awarenes in reinforcement learning remains unclear. To examine effects of subjective visual awareness on weighting of action values, we used monkeys with a unilateral lesion of primary visual cortex (V1) as an animal model of blindsight. We compared their behavior when a visual cue was presented to their intact or lesion-affected visual field. Blindsight is phenomenon in which human subjects with V1 lesion show an ability to localize visual stimuli in their affected visual field without a visual awareness.
We designed a learning task in which monkeys had to search for an unseen target area (TA) on a screen by moving their eyes. When their gaze entered the TA, a visual cue was presented either in their intact or affected visual field. With successive trials, the search time, the number of saccades and the searching distance, were all reduced in both conditions. The monkeys frequently selected saccades of similar direction and amplitude preceding cue presentation. In contrast, the pattern of saccades after cue presentation was significantly different. After the cue was presented to the intact field, the monkeys stopped searching and moved their gaze away from the TA. This suggests that the monkeys were subjectively aware of the cue and switched their internal sate from searching. When the cue was presented in the affected field, the monkeys often continued searching suggesting they were unaware the visual cue. Despite this, when the cue was presented in the affected field it was able effectively to reinforce the discovery of the TA.
Previous studies have suggested that weighting of action values is mediated by dopamine neurons which can encode reward prediction errors. Therefore, our results suggested that a residual visual pathway after V1 lesion, presumably via the superior colliculus, could have an access to dopamine reinforcement mechanism.

Copyright © Neuroscience2014. All Right Reserved.