脳の確率的な計算原理 Probabilistic Principles of Brain Computation
S1-4-2-1 さえずる小鳥からの細胞内膜電位記録に基づく聴覚フィードバックモデルの検証- さえずりの分布を決める基底核様回路と，精緻なタイミングをつくる神経回路同士の相互作用 - Intracellular recordings from singing birds revealed no auditory feedback in sensorimotor nucleus HVC, - Interaction between neural circuits that involves stochastic song exploration and precise timings - ○濱口航介1 ○Kosuke Hamaguchi1, Katherine A Tschida1, Inho Yoon2, Bruce R Donald2,3,4, Richard Mooney1 Duke University Medical Center1, Dept Computer and Electrical Engineering, Duke University, Durham, USA2, Dept Computer Science, Duke University, Durham, USA3, Dept Biochemistry, Duke University, Durham, USA4 For the maintenance of elaborated behaviors such as playing musical instruments, the balance between the precise motor control and stochastic motor exploration is necessary, and the evaluation of resultant motor quality based on auditory feedback plays an important role. However, how the real-time auditory feedback interacts with motor circuitry in the brain to maintain fast motor gestures was not well understood. Songbirds use auditory feedback to maintain their vocalizations, thus providing an attractive organism to identify synaptic mechanisms for auditory-motor integration. Recent imaging study from our group has found that, deafening weakens synapses on a specific class of sensorimotor neurons (HVCX cells) that project to an anterior forebrain pathway (AFP) important to audition-dependent vocal plasticity. The lesions made in the output of AFP, LMAN, can prevent deafening-induced song degradation and inactivating LMAN can reverse shifts in syllable pitch induced by distorted auditory feedback (DAF), suggesting that the AFP has access to auditory feedback-related information about song performance. These studies elevate HVCX cells as a potential source of auditory feedback information to the AFP, however, HVCX cell action potential outputs are insensitive to these feedback perturbations. This suggests that they may integrate feedback only at subthreshold levels during singing. Using intracellular recordings in singing finches, we found that the synaptic activity of HVCX cells is insensitive to distorted auditory feedback. Furthermore, in vivo multiphoton imaging combined with AFP lesions revealed that deafening-induced changes to HVCX synapses depend on intact AFP output. These findings show that deafening-induced changes in HVCX cells do not simply result from reduced auditory drive to HVC and also suggest that the AFP accesses feedback information independent of HVCX cells.	S1-4-2-2 大脳皮質自発発火活動と確率的記憶状態 Internal noise and stochastic memory states in cortical networks with long-tailed synaptic connections ○寺前順之介1, 坪泰宏2, 平谷直輝3, 深井朋樹2,3 ○Jun-nosuke Teramae1, Yasuhiro Tsubo2, Naoki Hiratani3, Tomoki Fukai2,3 大阪大院・情報1, 立命館大学情報理工学部知能情報学科2, 東京大院・新領域3 Grad School of Information Science and Technology, Osaka Univ1, Department of Human & Computer Intelligence, Ritsumeikan University2, Dep of Complexity Science and engineering, Univ of Tokyo, Tokyo, Japan3 Recent experiments revealed that excitatory synaptic potentials (EPSPs) between cortical pyramidal neurons obey a long-tailed, typically lognormal, distribution. Such a distribution creates a synaptic spectrum from majority of weak synapses to a small fraction of extremely strong synapses. However, the implications of the highly skewed amplitude distribution of EPSPs remain largely unknown for signal processing and memory state in cortical networks. Here, we numerically and analytically study the significance of the long-tailed EPSP distribution for spike-based signal processing and memory states in network models of cortical neurons. We asked whether reverberating synaptic input generated by such a distribution is sufficient for the genesis of stable spontaneous activity, and whether this internal noise provides an solution for efficient spike-based communication between neurons and is beneficial for associative memory capacity. We show that the networks can robustly generate internal noise optimal for spike transmission between neurons with the help of a long-tailed distribution in the networks. We then construct an associative memory network based on the long-tailed weights distribution. In the memory network, only neurons responsible for stored memory patterns have relatively higher membrane potentials than other neurons owing to the internal noise. Consequently, the probability of responding to inputs increases for the encoding neurons. Our results imply that the internal noise is also useful for associative memory recall.
S1-4-2-3 認知発達研究における注視行動の統計的モデリング Statistical Modeling of Eye Movements in Cognitive Developmental Studies ○日高昇平1 ○Shohei Hidaka1 北陸先端科学技術大学院大学知識科学研究科1 School of Knowledge Science, Japan Advanced Institute of Science and Technology1 The study of cognitive development hinges, largely, on the analysis of infant looking. But analyses of eye gaze data require the adoption of linking hypotheses: assumptions about the relationship between observed eye movements and underlying cognitive processes. We develop a general framework for constructing, testing, and comparing these hypotheses, and thus for producing new insights into early cognitive development. We first introduce the general framework applicable to any infant gaze experiment and then demonstrate its utility by analyzing data from a set of experiments investigating the role of attentional cues in infant learning. The new analysis uncovers significantly more structure in these data, finding evidence of learning that was not found in standard analyses and showing an unexpected relationship between cue use and learning rate. Finally, we discuss general implications for construction and testing of quantitative linking hypotheses.	S1-4-2-4 ヒトの社会的直観・熟慮メカニズムの計算・確率論的アプローチ Computational approach to understand human social intuition and reflection mechanisms ○春野雅彦1 ○Masahiko Haruno1 NICT脳情報通信融合研究センター1 NICT Center for Information and Neuralnetwork1 Much decision making requires balancing benefits to the self with benefits to the group. There are marked individual differences in this balance such that individualists tend to favor themselves while prosocials tend to favor the group. Understanding the computational mechanisms underlying this difference has important implications for society and its institutions. Using behavioral and magnetic resonance imaging data collected during the performance of economic games, we first show that individual differences in social preferences for resource allocation depend on intuitive processes instantiated in the nucleus accumbens and amygdala rather than on the conscious reflection on fairness associated with activity in anterior cingulate and dorsolateral prefrontal cortex. More specifically, our data suggest that the nucleus accumbens is involved in intuitive behavioral selection for both prosocials and individualists, while the amygdala is involved in intuitive perception of inequity. We will also examine the roles of the nucleus accumbens and the amygdala by using a decoding method of fMRI data. In the last part, by contrast, we will also discuss the roles of prefrontal cortices during prosocial behavioral choice by reflection.

上部に戻る

前に戻る