著者
北野 孝太 山岸 厚仁 西森 克彦 佐藤 暢哉
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

There are many studies on a variety of social behavior mediated by oxytocin. Of the studies, several suggest that oxytocin is deeply involved in empathy. Empathy toward other individuals is thought to be necessary to give rise to helping behavior. However, there has been little research on the relationship between oxytocin and helping behavior. We investigated helping behavior in oxytocin receptor knockout prairie voles. Prairie voles are known as socially monogamous rodents with high sociality. To examine helping behavior, we used a paradigm in which voles helped a conspecific soaked in water by opening a door. The prairie voles were housed in pairs. All the pairs were siblings. One of the pairs was assigned to be a soaker vole and the other was assigned to be a helper vole. There were two groups; the oxytocin receptor knockout and wildtype groups. The oxytocin receptor knockout voles were paired with wildtype siblings and were assigned to be the helper. Their wildtype cagemates were assigned to be the soaker. The experimental apparatus was divided into two areas; a pool area and a ground area. These areas were separated by a transparent acrylic plate on which a circular door was attached. The soaker vole was placed in the pool area and the helper vole was placed in the ground area. The door could be opened only from the ground area. We measured door-opening latencies. After the door-opening, the pair of the voles were allowed to interact. At that time, the huddling time was measured. As a result, the oxytocin receptor knockout voles showed significantly longer latencies for opening the door than the wildtype voles. In addition, the oxytocin receptor knockout voles showed shorter huddling time than the wildtype voles. These suggest that oxytocin is important for empathic behavior.
著者
加藤 郁佳 風間 北斗
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

The activity of dopaminergic neurons (DANs) has been most extensively studied in the context of associative learning. Specifically, DANs have been shown to respond to an unconditioned stimulus and a conditioned stimulus as it becomes predictive of the value of unconditioned stimulus through learning (Schultz et al., 1997; Eshel et al., 2015). However, a conditioned stimulus, especially an odor, is associated with an innate value because it can evoke approaching or avoiding behavior even without learning (Badel et al., 2016). Whether and how DANs represent such innate value of odors remain largely unknown. Here, we addressed these issues using an olfactory system of Drosophila melanogaster as a model. We focused on the DANs innervating the mushroom body (MB) where the olfactory information is modulated by reward or punishment (Aso et al., 2014). The MB comprises 15 compartments, each of which receives input from a distinct type of DANs and provides a place for synaptic modification underlying learning. We performed two-photon Ca2+ imaging to record the responses of DANs in all the 15 compartments to a panel of 27 odors with various values quantified based on the innate approach or avoidance behavior (Badel et al., 2016). We found that DANs differentially respond to attractive and aversive odors; the activity of DANs in specific compartments was correlated positively or negatively with the value of odors. Regression analysis showed that information about the value of odors can be decoded from the activity of DANs, and DANs in each compartment have specific contributions to the encoding of odor values. These results suggest that DANs drive not only associative learning but also moment-to-moment adaptive behaviors in an olfactory environment. In this presentation, we plan to discuss the neural circuit mechanism that gives rise to the odor-evoked activity of DANs.
著者
山本 拓都 北澤 茂
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

Perceived colors of objects remain constant under a variety of illuminations. This ability of color constancy has been modeled by using deep convolutional networks that were extensively trained to estimate reflectance of objects. However, it is not likely that our brain acquires color constancy through such rigorous procedures. We here show that color constancy emerges in a simple deep convolutional autoencoder when it involves batch normalization layers. We found that batch normalization is a simple but effective method to cancel any global bias in illumination and achieve color constancy. However, it may still be argued that batch normalization layers are not physiological by itself. We thus proposed a biologically plausible replacement for batch normalization that consists of two tandem inhibitory layers, one with subtractive and another with divisional inhibitory neurons. The model is biologically plausible because there are two groups of inhibitory interneurons in the cerebral cortex, one that performs subtractive inhibition (e.g., somatostatin-expressing interneurons) and another that performs divisional inhibition (e.g., parvalbumin-expressing interneurons). Further, both types of inhibition can be achieved by adjusting parameters of an extended Hodgkin-Huxley neuron. By using the network with biologically plausible Hodgkin-Huxley neurons, we were able to achieve color constancy. We suggest that color constancy emerges as a simple consequence of normalization achieved by the two groups of inhibitory interneurons.
著者
紺野 大地 松本 信圭 鈴木 隆文 池谷 裕二
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

The brain is intrinsically active even in the absence of external stimuli. Although many researches have studied spontaneous brain activity, few studies have examined the differences in oscillatory frequency, it remained whether the patterns of spontaneous brain activity are similar between different oscillatory frequencies. To address this question, we recorded electrocorticograms (ECoGs) from the visual cortex of free moving rats. ECoG is a well-balanced neural signal, which is stably mapped brain surface local field potentials over a wide cortical region with high signal fidelity and minimal invasiveness to the brain tissue. The ECoG probe used in this study had 32 electrodes on a mesh structure to stably contact them onto the brain surface. We found that the across-electrode propagation pattens of spontaneous brain activity differed across oscillatory frequency bands. For examples, the spatial propagations of delta- and alpha-band activity tended to exhibit an inverse correlation. On the other hand, the propagations of beta- and gamma-band activity were similar. In addition, the spontaneous activity patterns were classified into several clusters using uniform manifold approximation and projection (UMAP) and affinity propagation algorithms. These results reveal a complex relationship between spontaneous brain activity and oscillatory frequencies.
著者
宮本 健太郎
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

To explore and survive in an unpredictable, volatile world with multiple alternatives available, people and other animals, such as macaque monkeys, need to estimate uncertainty before making a decision. However, the neural mechanism to enable proactive metacognitive judgements based on evaluation of uncertainties is unknown.In our first study on humans with functional neuroimaging and transcranial magnetic stimulation, we newly invented a prospective metacognitive matching task. In the task, participants were required to estimate their performance (`subjective probability') to classify the direction of ambiguous motion in random-dot kinematogram task. Then they compared this subjective probability with the probability of reward offered by the alternative external cues (`environmental probability') and chose the better probability option in prior to performing the motion classification. Activity in several frontal and parietal areas reflected both subjective and environmental probabilities during perceptual decision making. Anterior lateral prefrontal cortex (alPFC, area 47), however, tracked evidence relating to subjective probabilities both when a choice was taken and when it was rejected. Moreover, fMRI signals in alPFC modulated by subjective probability predicted prospective metacognition performance and ability. These observations suggest that alPFC plays a critical role to assess one's own cognitive skills and mental states proactively to take an optimal choice in the future.In our second study on monkeys with functional neuroimaging and targeted pharmacological intervention, we previously found that the dorsal prefrontal and frontopolar cortices confer decision confidence on experience and ignorance, respectively, during a serial-probe recognition memory task (Miyamoto et al., 2017, Science 355(6321); Miyamoto et al., 2018 Neuron 97(4)). We have newly found that the inferior parietal lobule (area PG) contributes to integrate these confidence read-outs and execute a strategically optimal decision making for post-decision wagering based on self-reflection of performance in the precedent memory task.These human and monkey studies converge to suggest that higher-order processes to proactively evaluate subjective uncertainties are implemented in primate neural networks. The neural mechanism would be essential to convert metacognition into action.
著者
船水 章大 Fred Marbach Anthony M Zador
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

Neurons in auditory cortex encode auditory stimuli, but the precise encoding can depend strongly on task-relevant variables such as stimulus or reward expectation. This raises the question: If the cortical representation of the stimulus varies with task-relevant variables, how can areas downstream of auditory cortex decode these representations? One possibility is that decoding in downstream areas also depends on these task-relevant variables. To address this question, we developed a two-alternative choice auditory task for head-fixed mice in which we varied either reward expectation (by varying the amount of reward, in blocks of 200 trials) or stimulus expectation (by varying the probability of different stimuli). The task was based on our previous study (Marbach and Zador, bioRxiv, 2016) in which mice selected left or right spout depending on the frequency of tone stimuli (low or high). We used two-photon calcium imaging to record populations of neurons in auditory cortex while mice performed the task. We found that varying either reward or stimulus expectation changed neural representations (i.e. stimulus encoding). However, the optimal decoder was remarkably invariant to different encodings induced by different expectations. The discrepancy between the neural encoding and decoding could be that the prior encoding was independent from the sound decoding in the auditory cortex. Our results suggest that stimuli encoded by auditory cortex can be reliably read out by downstream areas, even when the encoding is modulated by task-relevant contingencies.
著者
風間 北斗
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

Sensory systems are known to fulfill two seemingly conflicting requirements in order to adapt to dynamic environment; maintaining stimulus representations over time to ensure reliable encoding of sensory information while modifying/optimizing the representations depending on experience. To understand the circuit mechanisms that control both the stability and flexibility of sensory representations, we have been studying odor representations in the olfactory memory center, the mushroom body (MB), in the Drosophila brain. We conducted comprehensive volumetric Ca imaging and analyzed the responses of all ~2,000 Kenyon cells (KCs), the principal neurons in the MB, to repeated applications of odors. We found that KCs changed their odor responses upon repetitive application of stimuli; some KCs showed depression while some showed facilitation in an odor identity-specific manner, indicating that the odor tuning rather than the general excitability of individual KCs was modified. Similar change was not observed in the upstream primary olfactory center, the antennal lobe, suggesting that odor representations are modified in the MB through odor experience. Notably, the response correlation between different odors was fairly stable across trials when the KC population activity was compared, indicating that the relative odor representations are maintained at a population level despite the changes at a cellular level. We will discuss candidate mechanisms that might underlie the dynamic changes of MB odor representations.
著者
Marco Karl Wittmann
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

Navigating dynamic environments often requires humans and other animals to consider information beyond currently experienced choice values. I present two functional magnetic resonance imaging (fMRI) experiments, one in humans and one in macaque monkeys, that demonstrate the impact of global reward signals on explorative behaviour and neural activity. First, I show that representations of reward rates exist simultaneously at multiple time scales in a human brain network including dorsal anterior cingulate cortex (dACC) and the agranular insula (Ia). Such signals can be used to compute the change in reward rate in the environment and to construct estimates of future values that are different from the values experienced in the past. Their relative strengths govern how we decide to persist in our environment or switch to an alternative course of action; this can lead animals to keep exploring poor environments despite low reward rates in anticipation of future reward. In a second project we scanned macaque monkeys during a 3-armed bandit task. By carefully controlling for the effects of choice-contingent reward and choice repetition, we revealed that their choice strategy was again distinctly influenced by the global reward state. Remarkably, the global reward state affects the way that each choice outcome is valued and influences future decisions so that the impact of both choice success and failure is different in rich and poor environments. Successful choices are more likely to be repeated but this is especially the case in rich environments. Unsuccessful choices are more likely to be abandoned but this is especially likely in poor environments. In other words, a low global reward state incentivized the animals to increasingly explore alternative options even if their current choice was successful. Just as in the first study in humans, we found that dACC and Ia, but in addition the dorsal raphe nucleus, track global reward state as well as specific outcome events.
著者
Vincent D Costa Ramon Bartolo Hua Tang Bruno B Averbeck
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

All organisms, from slime molds to humans, have to decide whether to forego immediate rewards, like food, in order to explore an unknown option and learn if it is better than something already experienced. This trade-off is referred to as the explore-exploit dilemma. To balance exploration and exploitation biological agents need to know when exploration is advantageous. An efficient strategy for managing explore-exploit tradeoffs is to predict the immediate and future outcomes of each available choice option. Predicting whether choices will be immediately rewarded or unrewarded is easily computed based on past experience. Predicting how often choices are rewarded or unrewarded in the future is a much more difficult computation, as it relies on prospection. Yet these predictions can be integrated to decide when exploration is advantageous. Given theoretical and lay beliefs that balancing exploration and exploitation is difficult, prior studies have focused on identifying cortical mechanisms of exploratory decision making, ignoring how subcortical motivational circuits aid in managing explore-exploit tradeoffs. Here, we leverage theoretical advances in the use of partially observable Markov decision process models to understand how reward uncertainty motivates exploration, in order to characterize neural activity in the amygdala, ventral striatum, orbitofrontal cortex, and dorsolateral prefrontal cortex of macaque monkeys as they solve a multi-arm bandit task designed to query specific aspects of novelty seeking and explore-exploit decision making. Our findings challenge the widely held corticocentric view of how the brain solves the explore-explore dilemma, by emphasizing similarities in how subcortical and cortical regions encode value computations critical for deciding when exploration is warranted.
著者
Alicia Izquierdo Alexandra Stolyarova Mohsen Rakhshan Megan AK Peters Hakwan Lau Alireza Soltani
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

Studies in humans have revealed neural correlates of confidence in several regions, including in prefrontal cortex. However, it is still unclear which regions are causally involved in this process. I will present recent work where we trained rats to discriminate between ambiguous visual cues via spatial choices based on a learned stimulus-response rule. Following action selection using a touchscreen, rats expressed their confidence by time-wagering: they could wait for a variable amount of time before they could receive a possible reward or initiate a new trial. This design allowed us to measure confidence trial-by-trial. We found that waiting times increased with discrimination accuracy and were negatively correlated with response times, demonstrating that this measure could be used as a proxy for confidence. Following chemogenetic silencing of anterior cingulate cortex (ACC), waiting times became less diagnostic of perceptual uncertainty. We also computed metacognitive efficiency (meta-d'/d') that assesses how well waiting time tracked discrimination performance (d') across trials (Maniscalco & Lau, 2012), and found that this measure was significantly reduced following ACC inhibition. These results will be discussed in the context of our recent work in the orbitofrontal cortex and how animals may similarly show metacognition in more traditional reinforcement learning paradigms.
著者
REI AKAISHI
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

In contrast to the narrow and local computations in artificial systems, computations in the human and animal brains proceed in both local and global scales in time and space to try to achieve a hierarchy of short-term and long-term goals across both local and global environments. One clue of the unique nature of the local computations in human and animal brains comes from their anomalies: the observed phenomena include history-dependent inertia in decision making and biases in credit assignment in learning. These biases can make the computations in the local environment more efficient by reducing computational costs with successive repetitions of the same behavioral patterns. However, these biases can increasingly narrow down the scope of computations to the local situations, which can be detrimental for their survival. To counteract this tendency for the local adaptation, the brain seems to be equipped with the ability to model and implement long-term and large-scale predictive decisions. This topic has been recently examined especially in the experimental paradigms of foraging decisions that mimic the situations of foraging behaviors of animals in the wild. The animal foraging paradigm, where an agent alternates between global diffusive searches and fine-grained local searches, is actually a broadly relevant model for a wide range of human foraging-like behaviors including memory search, information seeking, problem solving and social learning. By integrating these findings with relevant literature, we would like to suggest that biological intelligent systems operate through a hybrid multiscale architecture of local and global computations.
著者
Will Dabney Zebulun Lloyd Kurth-Nelson Naoshige Uchida Clara K Starkweather Demis Hassabis Rémi Munos Matthew Botvinick
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-16

Since its introduction, the reward prediction error (RPE) theory of dopamine has explained a wealth of empirical phenomena, providing a unifying framework for understanding the representation of reward and value in the brain. According to the now canonical theory, reward predictions are represented as a single scalar quantity, which supports learning about the expectation, or mean, of stochastic outcomes. In the present work, we propose a novel account of dopamine-based reinforcement learning. Inspired by recent artificial intelligence research on distributional reinforcement learning, we hypothesized that the brain represents possible future rewards not as a single mean, but instead as a probability distribution, effectively representing multiple future outcomes simultaneously and in parallel. This idea leads immediately to a set of empirical predictions, which we tested using single-unit recordings from mouse ventral tegmental area. Our findings provide strong evidence for a neural realization of distributional reinforcement learning.The RPE theory of dopamine derives from work in the artificial intelligence (AI) field of reinforcement learning (RL). Since the link to neuroscience was first made, however, RL has made substantial advances, revealing factors that radically enhance the effectiveness of RL algorithms. In some cases, the relevant mechanisms invite comparison with neural function, suggesting new hypotheses concerning reward-based learning in the brain. Here, we examine one particularly promising recent development in AI research and investigate its potential neural correlates. Specifically, we consider a computational framework referred to as distributional reinforcement learning.
著者
栁澤 琢史
雑誌
第43回日本神経科学大会
巻号頁・発行日
2020-06-15

脳情報の解読と制御は神経科学の発展に伴って現実的な技術となり、様々な医療応用が期待されている。脳波や脳磁図、fMRI、NIRSなど様々な脳信号に対して機械学習を適用することで、知覚認知内容や運動状態などを推定できる(脳情報解読、Neural Decoding)。また、Neural decodingの結果に基づいてロボットやコンピュータを脳信号から制御できる(Brain-Computer Interface, BCI)。我々は、人の頭蓋内に電極を留置して脳波を計測する皮質脳波に対してNeural decodingを適用し、ロボットハンドを制御するBCIを開発した。特に筋萎縮性側索硬化症(ALS)により重度の運動機能障害がある患者に対して、感覚運動野へ頭蓋内電極を留置しBCIの有効性を評価する臨床研究を行い、重度麻痺があってもBCIにより意思伝達できることを示した。しかし、ALS患者では進行性に運動野活動が減弱するため、運動情報に基づくBCIには限界がある。そこで、後頭葉や側頭葉などALS患者でも比較的、機能が保たれる領域から皮質脳波を計測することで、意思伝達を実現するBCIを目指している。多様な意味内容の動画を視聴している際の皮質脳波を計測し、動画の意味内容を、word2vecを用いてベクトル化し、これを皮質脳波から推定し、視覚的意味内容推定に基づくBCIを開発した。 BCIは、neural decodingを介して、脳と機械がインタラクションする技術でもある。脳がBCIを介してどの程度の情報を操作できるか、また、BCIの操作に習熟することで、脳にどのような変化が誘導されるのかは、BCIの可能性を知り安全性を高める上で重要な神経科学的問題でもある。我々は脳磁図を用いた非侵襲型BCIを開発し、様々なBCI操作に習熟することによる脳活動及び神経症状の変化を探索した。特に感覚運動野の皮質活動に基づいたBCIによりロボットハンドを制御し、これを上肢に幻肢痛がある患者に適用したところ、BCI使用後には、患者の感覚運動野に可塑的変化が誘導され、幻肢痛も制御されることを明らかにした。同様の方法は視覚認知機能の修飾などにも効果が期待される。 異常な脳活動状態に起因する精神神経疾患に対して、Neural decodingを用いた活動状態の解読と、neurofeedbackによる活動修飾は、新たな治療オプションになると期待される。脳情報の解読と制御を神経科学的に理解し、精神神経疾患の新しい診断·治療につなげる我々の取り組みを紹介する。 .