報酬・意思決定 Reward and Decision Making
P1-2-177 視覚探索中のサルLIP野ニューロン活動における側抑制 Lateral inhibitory interactions in the activity of macaque lateral intraparietal neurons during visual search ○西田知史1, 田中智洋1, 小川正1 ○Satoshi Nishida1, Tomohiro Tanaka1, Tadashi Ogawa1 京都大学大学院 医学研究科 認知行動脳科学1 Dept Integrative Brain Science, Graduate School of Medicine, Kyoto Univ, Kyoto1 During visual search, visually-responsive neurons in visuomotor areas, such as lateral intraparietal area (LIP), discriminate the target from distractors by exhibiting the stronger activity when the target appears inside the receptive field (target-related activity) than when the distractor appears outside the receptive field (distractor-related activity). Theoretical studies have suggested that lateral inhibitory interactions can play a key role in producing such target-discriminative activity. However, little is known about whether and how lateral inhibitory interactions contribute to shaping the in-vivo neuronal activity during visual search. To address this issue, we recorded neuronal activity from LIP of two monkeys performing a visual search task including target-present and target-absent trials. The monkeys had to make a saccade to a color-singleton target among distractors in the target-present trials, whereas they had to hold their fixation until the trial end in the target-absent trial in which all stimuli were distractors. By using the activity from the target-absent trials as neutral activity, we separately evaluated the elevation of target-related activity (target-related elevation) and the reduction of distractor-related activity (distractor-related reduction) in the target-present trials. We found that (1) not only the target-related elevation but also the distractor-related reduction did occur at the population level, (2) the magnitude of these modulations gradually increased as the time of saccade initiation approached, and (3) the strength of target-related elevation was correlated with that of distractor-related reduction. These results suggest that target-discrimination activity (i.e. the difference between the target- and distractor-related activities) in LIP during visual search is produced by the combination of target-related facilitation and distractor-related suppression in activity via lateral inhibitory interactions.	P1-2-178 複数ステップ報酬課題における線条体の活動が時間割引率を反映する Striatal event-related activity reflects intrinsic discount rate in a multi-step reward task ○田中沙織1, 高岡正法1, 米田紘康1, 榎本一紀2, 鮫島和行2, 春野雅彦3, 木村實2, 大竹文雄1 ○Saori Tanaka1, Masanori Takaoka1, Hiroyasu Yoneda1, Kazuki Enomoto2, Kazuyuki Samejima2, Masahiko Haruno3, Minoru Kimura2, Fumio Ohtake1 大阪大学　社会経済研究所1, 玉川大学2, 情報通信研究機構3 ISER, Osaka Univ, Osaka1, Tamagawa Univ, Tokyo2, NICT, Tokyo3 Midbrain dopamine neurons are known to represent the long-term value of multiple future rewards in monkeys (Enomoto et al., 2011). To establish this fact in the human brain, we applied a multi-step reward task used in monkey neurophysiological studies in a human fMRI experiment. Subjects selected a card among four cards (one wining and three blank) until they drew the winning one and received a monetary reward. Two conditions were used: In the "decreasing" condition, eventual monetary reward began at 400, and was reduced by 100 units after each pick of a blank card, while in the "increasing" condition, reward began at 100, and was increased by 100 units after each pick. Theoretically, prediction error is not affected by the variation in subjects' discount rate during the decreasing condition, and we found that the BOLD signal in the ventral striatum was significantly correlated with prediction error with any discount rate during the decreasing condition. This finding is consistent with monkey studies as the ventral striatum have been shown to receive dopaminergic projections. In contrast, prediction error is strongly affected by subjects' discount rate during the increasing condition; discount rates smaller than 0.7 provide less-than zero prediction error signals when they drew a winning card, while discount rates larger than 0.7 provide more-than zero prediction error signals under the same circumstances. We compared event-related activity produced by drawing winning cards between subjects with discount rates smaller and larger 0.7 as estimated by an economic questionnaire performed after fMRI experiment. We found positive/negative event-related activity in the striatum of subjects with larger/smaller discount rates, respectively. Our findings build bridges not only between neurophysiological (monkey) and imaging (human) studies, but also between neural data, computational models, and economic measurements of the evaluation of multiple future rewards.
P1-2-179 確率共鳴を導入した入力情報が危険認識の際のヒューリスティックな意思決定に与える効果のモデリング―新コミュニケーションシステムの実現に向けて Modeling of effects of input information with the stochastic resonance on heuristic decision making in the risk perception -toward the realization of new communication systems ○平林美樹1, 小嶋寛明1, 大岩和弘1, 大橋弘忠2 ○Miki Hirabayashi1, Hiroaki Kojima1, Kazuhiro Oiwa1, Hirotada Ohashi2 情報通信研・未来ICT・バイオICT1, 東大院・工・システム創成2 Adv ICT Res. Inst, NICT, Hyogo1, Dept Sys Inno, Univ of Tokyo, Tokyo2 For example, when the amount of information is excessive to find a solution systematically, heuristic decision making is often performed to find a rapid solution close to the best possible answer by introducing the reference index to enhance the value of a certain option such as the satisfaction level in order to make it easy to get a solution. We think that the heuristic information processing itself has universal and predictable features caused by the restriction of the brain's information processing capacity. Utilizing these properties to predict behaviors, we aim to provide platform technologies to construct new communication systems for distributing and understanding information correctly. In our previous work, we revealed universal heuristic features of strategic decision making using a neural network model focusing on hippocampal circuits. In the information processing of updating memories to strategy consolidation, it is thought that hippocampal circuits play the roles to select the synapse level of memories and passes them to the system level of memories by "pattern separation" to distinguish the differences and "pattern completion" to search similar memories by a clue. Memories are stored as a set of sensory information and feelings occurred at that time. The hippocampus is located in the place where all sensory information can be received. And its structure is suitable to associate it. Using the neural network model of hippocampus, we verify that we can recognize the importance of the target information through the stochastic resonance by adding the noise information to inputs. We expect that our findings contribute to construct the new communication systems that can prevent the misunderstandings and can express information accurately by appealing to the decision making process directly through the innovation of the ways for information service by considering the features of neural circuits, and to promote the understanding the roles of hippocampus.	P1-2-180 空腹と餌の選択：満足が引き起こす選択行動の乱雑さの増加 Hunger and Food Choice: Satiety-dependent increase of Choice Stochasticity in non-human primate ○山田洋1 ○Hiroshi Yamada1 国立精神・神経医療研究センター1 Center of Neurology and Psychiatry, National Institute of Neuroscience1 Hunger and thirst are fundamental biological processes that drive choices and consumptions of rewards in human and animal. While the value of consumable rewards increase if animals get hungry and thirsty, it is still unclear how the satiety state affect the preference of consumable goods, and hence how the state change their reward values and drives choice behavior. To address this issue, I examine the choice behavior of marmoset monkeys (Callithrix jacchus) by employing standard economic techniques. I applied all possible pair-wise choices among five food items to six captive marmosets under the three different satiety conditions: sated (lunch after regular breakfast), unsated (before regular lunch), and hungry (before regular lunch without breakfast). I measured all of their choices in each monkey each condition, and count the number of transitivity violation (i.e. inconsistent choice) as the definition in the traditional economics. Each monkey showed transitivity violations in the sated and unsated condition frequently. The number of violation, however, dramatically decreased when they were hungry (hungry condition). In addition, preferences of each food items appeared to be stable between different satiety levels. These evidences suggested that preferences are not inconsistent among different satiety level, but monkeys behave less stochastic as they become hungrier. The orbital and medial prefrontal, ventral striatum network that governed by hypothalamus might provide the basis for the state-dependent valuation and choice behavior.
P1-2-181 LIP野における目標識別からサッカードまでの遅延は刺激に依存する Stimulus-dependent delay from neuronal discrimination in LIP to saccade initiation during visual search ○田中智洋1, 西田知史1, 川端美智子1, 小川正1 ○Tomohiro Tanaka1, Satoshi Nishida1, Michiko Kawabata1, Tadashi Ogawa1 京都大学大学院 認知行動脳科学1 Dept of Integrative Brain Science, Kyoto Univ Graduate school of Medicine, Kyoto1 In a visual search task, reaction time tends to be shorter for a efficiently-discriminative target, but longer for a inefficiently-discriminative target. The reaction time of the eye or manual movement is conventionally utilized to evaluate target discriminability in visual search. During visual search, at least two neural processes may take place: visual target selection and response preparation/generation. The evaluation of target discriminability with reaction time is justified under the assumption that only the duration for neuronal target-discrimination can vary associated with visual conditions, but that for response preparation/generation is constant. Although this assumption is widely accepted, neurophysiological evidence are still controversial. To address this issue, we recorded activity from the lateral intraparietal area (LIP), known to be a crucial area for visual search, of monkeys performing a singleton search task. By manipulating stimulus luminance (Light vs. Dim) and target-distractor similarity (Easy vs. Hard), we made four stimulus conditions. Especially, we focused on the activity from the two visual conditions (Dim-easy and Light-hard), which were quite different in visual but can evoke similar saccadic reaction times. Furthermore, in data analysis, we selected the trials so that the distributions of reaction times were almost the same across conditions. Nevertheless, we found that target discrimination times were different between the two conditions, indicating that discrimination time alone cannot predict saccade initiation. Our additional analysis also revealed that by assuming a accumulator-threshold model driven by LIP visual responses, the time of saccade initiation is well explained. Thus, the present study provided the counterevidence against the constancy of post-discrimination time in LIP, and suggests the existence of another process in downstream pathways determining the timing of saccade generation.	P1-2-182 ヒト対ヒトの協調作業が困難であるほど意思決定の脳活動負荷が高まる Brain activation related to human to human communication in the decision making process; an fMRI study ○小野弓絵1, 丹羽政美4, 渡邊和子5, 橘篤導5 ○Yumie Ono1, J. Adam Noah2, Shaw Bronner3, Masami Niwa4, Kazuko Watanabe5, Atsumichi Tachibana5 明治大・理工・電気電子生命1, イェール大・医・精神2, ロングアイランド大・ADAMセンター3, 揖斐厚生病院・診療放射線科4, 星城大・リハビリ5 Dept Elec and Bioinformatics, School of Sci and Eng., Meiji Univ., Kanagawa, Japan1, Dept Psychiatry, Yale Univ. School of Med., CT, USA2, ADAM center, Long Island Univ., Brooklyn, NY, USA3, Dept Radiol., Ibi Welfare Hospital, Gifu, Japan4, Faculty of Rehab & Care, Seijoh Univ., Aichi, Japan5 During cooperation, dynamic decisions must be continuously made which involve outcomes for both self and the other(s). To elucidate brain activation related to human to human communication in the decision making process, we employed a task which consisted of playing the video game 'Tetris' in a cooperation mode in which two players were able to interact in real time while one of the pair underwent an fMRI scan. Five subjects consented to the study prior to their participation. Subjects played the game in the scanner together with the experimenter who stayed outside the magnet room. The game was modified in a block design with 30 s of activity alternating with 30 s of rest, while BOLD signals were continuously measured. At the end of each block, subjects evaluated the performance of their cooperation using a 5 point grading scale (1 being easy to 5 being difficult communications between players). Results compared the brain activities in good cooperation events (rate 1 and 2) to difficult conditions (rate 4 and 5). General activation patterns were similar in both conditions but totally increased when the rating was considered to be difficult or ineffective cooperation. Especially increased activities in the prefrontal cortex (PFC), the anterior cingulate cortex (ACC), and some of the basal ganglia were seen. These results suggest that when making dynamic decisions in a task which necessitates multi-person cooperation, the effectiveness of the ability to work together can influence decision making circuitry in ACC and PFC. When the task was rated behaviorally difficult, increases in activity were seen in the basal ganglia, frontopolar cortex and ACC which may reflect increased working memory, emotional response and cortical processing necessary to compensate for the increase in difficulty.
P1-2-183 海馬と扁桃体間のガンマオシレーションの同期は報酬予測による行動調節に寄与する Coherent high gamma oscillations between hippocampus and amygdala contribute to reward expectations modulating behavior in rats ○寺田慧1, 高橋晋2, 櫻井芳雄1 ○Satoshi Terada1, Susumu Takahashi2, Yoshio Sakurai1 京都大学大学院　文学研究科　心理学専修1, 同志社大 脳科学研究科 神経回路形態部門2 Dept Physiol, Kyoto Univ, Kyoto, Japan1, Dept Neural Circuitry, Doshisya Univ, Kyoto, Japan2 The aim of this study is to examine the role of coherent oscillations in the mechanism of reward expectations modulating goal-directed behavior. We simultaneously recorded spikes and local field potential from hippocampus (HPC) and amygdala (AMY) when the rat performed a light-direction discrimination task with different reward expectations (High/Low). Here, we report the following results. First, rats actually modulated their behavioral performance in accordance with their reward expectations. Second, we found more reward-expectation neurons in AMY and more task-performing neurons in HPC. Third, prominent increase of coherence of high gamma (50-150 Hz) oscillation in HPC and AMY was present during High reward expectation. Fourth, the coherent oscillation had significantly negative correlation with behavioral trial-by-trial fluctuation of goal selection time. These results suggest that the coherent high gamma oscillation contributes to reward expectations modulating goal-directed behavior. We propose that AMY influences firing rates or strength of synchronization of HPC neurons through coherent oscillations, which is a part of the mechanism how reward expectations modulate goal-directed behavior.	P1-2-184 意思決定に関わるシーン予測のデコーディング Decoding scene prediction associated with decision making ○鹿内友美1,2, 石井信1,2 ○Yumi Shikauchi1,2, Shin Ishii1,2 京都大院・情報学1, ATR・脳情報解析2 Grad Sch of Informatics, Kyoto Univ, Kyoto1, BICR, ATR, Kyoto2 Prediction of upcoming scene is an important cognitive function for decision-making in our daily life. Scene prediction is a kind of mental operation in which we imagine future observations by integrating available information including past observations. Mapping from neural signals into mental operation has rarely been reported. The goal of this research is to show that decoding of prediction from fMRI data is possible in a realistic 3D navigation environment. In the fMRI experiment, subjects were requested to discriminate the true next scene from a distracter scene, after being shown several observations useful for identifying the current position (and hence next scene) in the environment. We decoded the predicted scene from activities of fronto-parietal network regions and those regions were compared in terms of the decoding accuracy. When decoding of the frontal view of the next scene, which is most important for subjects to determine the next motion for navigation, the best accuracy was achieved from the parietal cortex. In contrast, the medial prefrontal region shows good performance when decoding the oblique views of the next scene.These results suggest that the scene prediction could be performed in the fronto-parietal network such to reflect the degree of contribution to the subsequent decision-making. Our findings have outlined the decision-making system employed in complicated environments, and implied useful characters of decoders which can be used for brain machine interface of practical navigation systems.
P1-2-185 顔に対する選好判断に関わる腹内側前頭前野の活動パターンにおける性差 Gender differences in the patterns of vmPFC activity associated with preference judgments for faces ○伊藤文人1, 藤井俊勝1, 阿部修士1, 川崎伊織1, 林亜希子1, 上野彩1, 吉田一生2, 境信哉3, 麦倉俊司4, 高橋昭喜4, 森悦朗1 ○Ayahito Ito1, Toshikatsu Fujii1, Nobuhito Abe1, Iori Kawasaki1, Akiko Hayashi1, Aya Ueno1, Kazuki Yoshida2, Shinya Sakai3, Shunji Mugikura4, Shoki Takahashi4, Etsuro Mori1 東北大院・医・高次機能1, 北海道大学保健科学院2, 北海道大学保健科学研究院生活機能学分野3, 東北大学大学院医学系研究科量子診断学分野4 Dept Behav Neurol Cogn Neurosci, Tohoku Univ, Sendai, Japan1, Div Health Sci, Hokkaido Univ, Sapporo, Japan2, Dept Funct Disability, Hokkaido Univ, Sapporo, Japan3, Dept Diagn Radiol, Tohoku Univ, Sendai, Japan4 Previous neuroimaging studies implicated the ventromedial prefrontal cortex (vmPFC) as a key region in a preference judgment. However, gender differences in vmPFC activity during the preference judgment remain unclear. In the present study, we prepared 256 face photographs (64 elderly female, 64 elderly male, 64 young female, and 64 young male) as experimental stimuli. The experiment consisted of two tasks: a pleasantness rating task during fMRI and a preference judgment task after fMRI. During the pleasantness rating task, 33 subjects (17 females, mean age: 21.2 years) were presented with the face photographs one by one, and were asked to rate each stimulus based on how pleasant it was (1; the lowest, 5; the highest). During the preference judgment task, the subjects were presented with the pairs of face photographs side by side and were asked to choose one they preferred. We found that the scores of pleasantness rating tasks were positively correlated with the vmPFC activity. Next, we directly compared the first-level statistical maps of females and those of males using a two-sample t-test, but we found no significant difference between these two groups. For each subject group (i.e., the female group and the male group), we then extracted the signal changes in the cluster of the vmPFC and conducted a ROI analysis using three-way ANOVAs with the gender of face stimulus (female and male), the age of face stimulus (elderly and young), and the subject's preference (preferred and not preferred) as factors. The ANOVA for the data of females revealed only a significant main effect of the subject's preference. The ANOVA for the data of males showed significant main effects of the gender and the age of face stimulus, but showed no significant main effect of the subject's preference. These findings suggest that the age and gender differences of face stimulus impact on the patterns of the vmPFC activity in males but not on those in females.

上部に戻る

前に戻る