一般口演（Oral）

Oral Decision Making 一般口演意思決定

7月28日（日）8:45～9:00　第9会場（朱鷺メッセ　3F　306+307） 4O-09m1-1 A Bayesian model explains how individual and mutual properties of action and outcome affect sense of agency Roberto Legaspi（Legaspi Roberto），Taro Toyoizumi（Toyoizumi Taro） Center for Brain Science, RIKEN Sense of agency (SoA), i.e., the experience that oneself initiated an action that caused an outcome, is a significant concept in philosophy, psychology, legal ethics and cognitive neuroscience. However, the literature still lacks a mathematical elucidation of the principles that underlie SoA. We theorize SoA in the framework of optimal Bayesian cue integration that explains how the individual and mutual properties of action and consequent outcome affect SoA. Our Bayesian model exhibits the following properties for high SoA: (A) the perceived action-outcome effect is consistent with the causation of the outcome by the action, (B) the perceived action and outcome signals are reliable when they happen in isolation, i.e., the amplitudes of sensory jitters are small enough not to increase sensory uncertainty, and (C) the prior belief that the action caused the outcome is strong. Our theory therefore posits SoA as precision-dependent causal agency. We used our Bayesian model to computationally explain the intentional binding effect, although hailed as reliable indicator of SoA, also lacks computational elucidation. Intentional binding shows that the perceived times of action and outcome are shifted toward each other when actions are voluntary (hence, there is SoA), but involuntary actions produced repulsion, i.e., prolonged opposite perception of the action-outcome interval (hence, lack of SoA). These are accounted for by (A). Furthermore, the magnitude of binding or repulsion depended on the reliability of the action and outcome signals, which is accounted for by (B). Lastly, as accounted for by (C), the strength of the prior moved the perception of the action and outcome towards each other, and weakening the prior diminished this effect similar to when the signals are perceived as unrelated. Our Bayesian model therefore computationally explains the nature of the link between SoA and intentional binding. However, our model also explains action and outcome binding even for unintended actions given high confidence in the action causing the outcome and reliable sensory cues. Divorced therefore from the long held notion that SoA can only arise from intentional action, our theory posits intentionality is not strictly necessary for SoA to emerge.		7月28日（日）9:00～9:15　第9会場（朱鷺メッセ　3F　306+307） 4O-09m1-2 The lateral line ablation decreases neural activity in the ventral habenula and leads to the loss of the dyadic fighting in zebrafish Yu Huan Hu（Hu Yu Huan）¹，Jiun-Lin Horng（Horng Jiun-Lin）²，Li-Yih Lin（Lin Li-Yih）³，Chun-Yung Chang（Chang Chun-Yung）¹，Ming-Yi Chou（Chou Ming-Yi）¹ ¹Department of Life Science, National Taiwan University ²Department of Medicine, Taipei Medical University ³Department of Life Science, National Taiwan Normal University When a male zebrafish (Danio rerio) encounters a conspecific male opponent, they fight with each other for short time for exhibiting the hierarchical dominance in an aren. For winning a contest and damage control, many sensory systems are involved in this behavior including the lateral line system (LLS) that enables it to detect environmental water flow. Previous studies indicated that the mechanosensory signals detected by the LLS significantly influence the proper decision making for fighting. However, the neural circuits and the brain regions underlying this process are still unclear. Since the neuromasts, the basic unit of the LLS, are composed of hair cells, we efficiently ablated the LLS in zebrafish by neomycin, an ototoxic aminoglycoside antibiotic. We confirmed the efficiency of neuromasts ablation by rhodamine123 fluorescent staining and swimming performance, and then applied the dyadic fighting paradigm to the lateral-line-ablated fish. We found that the lateral-line-ablated fish showed higher tendency to lose the fights against their control siblings. Fighting duration, latency, the ratio of initial dominance and attacking number did not change. Moreover, social preference, aggressiveness and anxiety levels were also similar to their control siblings. In situ hybridization of c-fos, a marker for neural activity, showed that the neural activity in the ventral habenula (vHb) was significantly lower in the lateral-line-ablated fish than that in control siblings. Our results suggest that the LLS regulates the outcomes of social conflicts through mediating the neural activity in the vHb.

7月28日（日）9:15～9:30　第9会場（朱鷺メッセ　3F　306+307） 4O-09m1-3 報酬処理に関わる神経回路における時間的に安定な期待値信号 Hiroshi Yamada（山田洋）^1,3,4，Yuri Imaizumi（今泉優理）²，Masayuki Matsumoto（松本正幸）^1,3,4 ¹筑波大　医学医療系　生命医科学域 ²筑波大　医療科学類 ³筑波大院　人間総合科学 ⁴筑波大　トランスボーダー医学研究センター Previous literatures indicate that many brain regions are involved in reward-based decision makings by signaling expected values (i.e., probability times magnitudes). However, nature of the signals is poorly understood in terms of temporal dynamics of the expected value signals carried by the neuronal population, and their differences among reward-related brain regions. To this end, we examined the following question across multiple reward-related brain regions: whether the expected value signals appeared stable or instantaneous during the presentation of the cue stimuli. We used simple cued task, in which the expected values of stimulus was indicated by pie-chart visual stimuli to the monkeys with a great precision; 0.1 to 1.0 probability of fluid rewards by 0.1 increment and 0.1 ml to 1.0 ml rewards by 0.1 ml increment. This enables us to map out the neuronal activity in the space of probability and magnitude of rewards (i.e., expected values) from the onset to offset of the cue stimuli. During the presentation of the cue stimuli, we recorded 686 neurons from two monkeys of four brain regions where is known to processe rewards: dorsal striatum (194), ventral striatum (144), Orbitofrontal cortex (area 13M, 190), and medial part of the Orbitofrontal cortex (area 14O, 158). First, a conventional linear regression analysis to detect the expected value modulation as probability and magnitude revealed that all four brain regions carry signals of expected values and its component (i.e., probability or magnitude) as a mixture in its population. By using state-space analysis (Monte et al., 2013), we found that the expected value signals maintained in ventral striatum and orbitofrontal cortex were very stable through cue presentation at the level of population. In contrast, the signals in dorsal striatum and medial orbitofrontal cortex population were unstable and fluctuated. These results potentially provide general framework for understanding how the expected values are computed and maintained through neural circuitry.		7月28日（日）9:30～9:45　第9会場（朱鷺メッセ　3F　306+307） 4O-09m1-4 霊長類尾状核のベータ波活動と神経活動は葛藤下の意思決定の価値と意欲を別々に再現する Ken-ichi Amemori（雨森賢一）¹，Satoko Amemori（雨森智子）²，Daniel J Gibson（Gibson J Daniel）²，Ann M Graybiel（Graybiel M Ann）² ¹京都大霊長研白眉プロジェクト ²McGovern Inst Brain Res, MIT, Cambridge, USA An approach-avoidance (Ap-Av) conflict arises when an individual has to decide between acceptance and rejection of a compound offer that has positive and negative attributes. During the conflict judgment of likes and dislikes, motivational responses simultaneously emerge and influence reaction times and the frequency of behavioral errors. Because of the differential influences of reward and aversiveness, such a conflict offer allows us to dissociate the neural processes of valence and arousal. The primate caudate nucleus (CN) has been implicated in affective judgment, but it is still unclear how the CN neural responses represent the decision-related variables. To address this, we utilized an Ap-Av conflict task and dissociated the neural processes of valence and arousal. In the task, monkeys reported their decisions by moving a joystick within 3 seconds to the Ap or Av target. We recorded spikes and local field potential (LFP) from the CN to find the neural correlates of valence and arousal. We analyzed 450 unit and 667 LFP activities recorded during the Ap-Av conflict task. With the trial-by-trial spike and beta oscillatory activities, we performed all-possible subset regression analyses using five selected behavioral variables, consisting of the offered reward size (Rew), offered airpuff size (Ave), value judgment (ChV: chosen value), reaction time (RT) and behavioral performance (FOE: frequency of omission error). By comparing the unit and beta responses, we found differences in the groups representing value judgment and arousal responses. While CN units represented both positive and negative values (positive: n=26, negative: n=14), the beta responses almost exclusively represented positive value (positive: n=93, negative: n=1), exhibiting a significant difference in the proportion (Fisher's exact test, P < 0.001). While CN units represented both positive and negative FOE (positive: n=17, negative: n=9), the beta responses representing FOE were almost exclusively positive (positive: n=22, negative: n=1), exhibiting a significant difference in the proportion (Fisher's exact test, P < 0.05). These dissociable features of unit and beta responses suggest distinctive roles of CN beta oscillations in facilitating positive value judgment and suppressing arousal motivation, while certain CN neurons respond to a wider range of features in the conflict decision-making.