視覚 Visual System
P2-1-124 幼若マウス大脳新皮質5層錐体細胞間における電気的結合の細胞位置および細胞タイプ特異性 Position- and cell type-specificity in electrical coupling between neocortical layer 5 excitatory neurons in neonatal mice ○中川直1, 松本直実1, 細谷俊彦1 ○Nao Nakagawa1, Naomi Matsumoto1, Toshihiko Hosoya1 理研脳センター・局所神経回路研究チーム1 Lab. for Local Neuronal Circuits, RIKEN BSI, Wako, Japan1 One of the major questions in neocortical research is the extent to which neuronal organization is stereotyped. Our previous analyses have revealed that the arrangement of subcerebral projection neurons (SCPNs), which are the major pyramidal neuron subtype in mouse layer 5, was highly organized. SCPNs align radially to form columnar units, and these units were periodically located. This periodic columnar organization was observed in multiple cortical areas and the typical wavelength of the periodicity was 30 μm. Under specific visual stimulation, neurons in single columnar units exhibited strongly correlated expression of the immediate early gene c-Fos. Therefore, SCPNs have a periodic columnar arrangement, and neuronal activity leading to c-Fos expression is similar among neurons in the same repeating units.It is not known what mechanisms generate the columnar functional organization. One of the possible mechanisms is electrical coupling mediated by gap junctions, which are present in developing cortical neurons. To investigate electrical coupling between layer 5 excitatory neurons in the neonatal mouse brain, we performed whole-cell patch clamp recordings from multiple neurons in acute brain slices and detected electrical coupling by measuring responses to electrical stimulation in neighboring neurons. Detected electrical coupling was confirmed to be sensitive to gap junction blockers. In the first postnatal week, electrical coupling was found between neighboring SCPNs with high probability (40-50%). This electrical coupling showed stringent cell-type specificity. Moreover, radially-arranged SCPNs were coupled more frequently than those arranged tangentially. The electrical coupling was specific to the early cortical development, as it almost disappeared in the second postnatal week. These results raise the possibility that electrical coupling is involved in the construction of the periodic columnar organization.	P2-1-125 Pumaはラット網膜の発生過程においてコリン作動性アマクリン細胞の樹状突起で発現が持続する Prolonged expression of Puma in cholinergic amacrine cells during the development of rat retina ○若林毅俊1, 小阪淳2, 森徹自1, 山田久夫1 ○Taketoshi Wakabayashi1, Jun Kosaka2, Tetsuji Mori1, Hisao Yamada1 関西医科大学　医学部1, 岡山大院・医歯薬・細胞組織2 Dept Anat Cell Sci, Kansai Med Univ, Osaka1, Dept Cytol Histol, Okayama Univ, Okayama2 During development of the nervous system, large numbers of neurons are overproduced and then eliminated by programmed cell death. Puma is a BH3-only protein that is reported to be involved in initiation of developmental programmed cell death in rodent retinal neurons. However, the expression and cellular localization of Puma in retinal tissues during development are not well known. Here we report the expression pattern of Puma during retinal development in the rat. During the period of programmed cell death in the retina, Puma was expressed in some members of each of the retinal neurons, including retinal ganglion cells, amacrine cells, bipolar cells, horizontal cells and photoreceptor cells. The amacrine cells are classified into different types on morphological and pharmacological characteristics. Although the developmental programmed cell death of cholinergic amacrine cells is known to be independent of Puma, this protein was expressed in almost all of their dendrites and somata of cholinergic amacrine cells at postnatal age 2 to 3 weeks and continued to be detected in cholinergic dendrites in the inner plexiform layer for up to 8 weeks after birth. Puma expression in the inner plexiform layer was enhanced by light deprivation, which inhibits synaptic activity in the retina. Our results suggest that Puma has some significant roles in retinal neurons after eye opening, especially that of cholinergic amacrine cells, in addition to programmed cell death of retinal neurons before eye opening.
P2-1-126 サル下側頭皮質における倒立顔とサッチャー顔の集団コーディング Population coding of inverted and thatcherized faces by face responsive neurons in monkey inferior temporal cortex ○松本有央1, 菅生-宮本康子1, 河野憲二2 ○Narihisa Matsumoto1, Yasuko Sugase-Miyamoto1, Kenji Kawano2 産総研・ヒューマンライフ1, 京都大院・医・認知行動脳科学2 Human Tech Res Inst, AIST, Tsukuba, Japan1, Dept Integrative Brain Sci, Kyoto Univ, Kyoto, Japan2 Neurons in the monkey anterior inferior temporal (AIT) cortex are known to respond to faces. We previously reported that face-responsive neurons in the AIT cortex represent information about a global category, i.e. human vs. monkey vs. simple shapes, earlier than information about more detailed categories about the faces, i.e. facial expression and/or identity. To investigate the effect of face inversion and thatcherization upon the temporally represented facial information, we recorded activities of 119 single neurons from the AIT cortex in two rhesus monkeys (Macaca mulatta) performing a fixation task. Test stimuli were colored pictures of monkey faces (4 expressions of 4 monkeys), human faces (4 expressions of 3 humans), and rectangles and circles (each in 1 of 4 colors). Modifications were made on each face-picture, and its 4 variations were used as the stimulus: upright-normal, inverted-normal, upright-thatcherized, and inverted-thatcherized faces. Population activity vectors consisting of the responses to the 4 variations of the faces were computed in a 50-ms time window. The clustering analysis of the vectors revealed clusters of human, monkey, and shapes in the [115, 165] ms window. For the vectors of the upright-normal or upright-thatchered faces, clusters of human identities and monkey expressions appeared in the [140, 190] ms window. For the vectors of the inverted-normal or inverted-thatchered faces, the separability of human identities and of monkey expressions decreased in the [140, 190] ms window. Furthermore, for only the monkey faces, the Euclid distance between the vectors for the upright-normal and upright-thatcherized faces was significantly larger than the one between the vectors for the inverted-normal and inverted-thatcherized faces after the [255, 305] ms window (paired t-test, p<0.05). These results suggest that the decrease in the separability between the fine category-vectors by the inversion would cause the Thatcher effect.	P2-1-127 マウス視覚野におけるターゲット特異的結合のin vivoカルシウムイメージング In vivo calcium imaging of target specific projections in mouse visual cortex ○松井鉄平1, 川島尚之1, 大木研一1 ○Teppei Matsui1, Takashi Kawashima1, Kenichi Ohki1 九大・医・分子生理1 Dept Mol Physiol, Kyushu, Fukuoka1 In the visual information processing stream of mammals including mice, information arrived at the primary visual cortex (V1) is passed to extrastriate cortical areas specialized for the processing of different features of visual information. However, due to technical difficulty of selectively recording activity of neurons projecting to particular target areas, it is still unclear how this selective processing is achieved. To overcome this difficulty, here we used in vivo two-photon calcium imaging of individual axon terminals of neurons expressing genetically encoded calcium indicator (GCaMP). In this approach, the source and target of particular projection neurons can be specified by selective expression of GCaMP in the source area and by imaging of axonal calcium activity in selected target area, respectively. To test this imaging-based approach, we first investigated spatial organization of orientation preference of efferent connections from V1 to extrastriate visual areas in mice. Individual axon terminal boutons as well as their calcium responses to visual stimuli could be clearly identified. Orientation preference of nearby axonal boutons were often different, however, boutons belonging to the same axon showed matched orientation preference as well as highly correlated calcium activities. These results not only show spatially intermingled pattern of feedforward V1 output for orientation information, but also demonstrate reliability of our recording of individual axonal activity. Mapping of spatial frequency (SPF) and temporal frequency (TF) tuning of individual axonal boutons also revealed intermingled organization of inputs tuned to wide range of SPF and TF. Using this method, we are currently investigating target area dependence of visual information sent through cortico-cortical connections between V1 and extrastriate areas.
P2-1-128 大脳新皮質第5層の皮質外投射ニューロンが構成する微小機能単位の機能解析 Characterization of functional units composed of a major subtype of pyramidal neurons in neocortical layer 5 ○丸岡久人1, 中川直1, 佐伯麻衣1, 松本直実1, 大倉正道2, 中井淳一2, 細谷俊彦1 ○Hisato Maruoka1, Nao Nakagawa1, Mai Saeki1, Naomi Matsumoto1, Masamichi Ohkura2, Junichi Nakai2, Toshihiko Hosoya1 理研脳センター・局所神経回路研究チーム1, 埼玉大・脳科学融合研究センター・脳機能解析部門2 Lab. for Local Neuronal Circuits, RIKEN BSI, Wako, Japan1, The Category of Analysis of Brain Function, Saitama Univ. Brain Science Institute2 The organization of functional subtypes of neurons shapes the basic structure of neocortical circuits. Our previous analysis revealed that the tangential arrangement of subcerebral projection neurons (SCPNs), which are a major pyramidal neuron subtype in mouse neocortical layer 5, was not random but significantly periodic. Under specific visual stimulation, SCPNs in a single repeating unit exhibited strongly correlated c-Fos expression, suggesting related functions (Maruoka et al., 2011). This analysis, however, had a limited temporal resolution because c-Fos expression was induced by one to two hours of stimulation. In order to improve the temporal resolution for a better characterization of the function of SCPNs in repeating units, we developed a method of in vivo two-photon calcium imaging of SCPNs in layer 5. G-CaMP6, a genetically encoded calcium indicators, was expressed in excitatory neurons in the deep layers of the mouse neocortex by in utero electroporation. Electrophysiological study revealed that fluorescence response evoked by action potentials can be optimized by expressing G-CaMP6 in a transient manner rather than constitutively. SCPNs were labeled by retrograde tracing before in vivo imaging. The optimized method enabled imaging of sensory-evoked responses of SCPNs in layer 5 of the binocular visual cortex. To identify the periodic arrangement of the cells, we performed three-dimensional analysis of cell positions. After in vivo imaging, the fixed brain was made transparent using the Sca/e method, and scanned by confocal microscopy. Based on the z-stack images in vivo and in vitro, three-dimensional coordinates of SCPNs were determined using custom-made computer programs. Detail analysis will be presented.	P2-1-129 図形残効パラダイムによる大きさの錯視 Size illusion in the figural after-effect paradigm ○樋口義久1, 田中雄一朗1, 青木俊太郎2, 藤田一郎1,2,3 ○Yoshihisa Higuchi1, Yuichiro Tanaka1, Shuntaro C. Aoki2, Ichiro Fujita1,2,3 大阪大・基礎工・生物工学1, 大阪大院・生命機能2, 大阪大・NICT, CiNet3 Sch Eng Sci, Osaka Univ, Osaka1, Osaka Univ, Grad Sch Frontier Biosci, Osaka2, Osaka Univ and NICT, CiNet, Osaka3 Inspection of a visual stimulus for a brief period can change the perceived position, shape, brightness, and size of a subsequently presented stimulus. This illusion has been collectively referred to as figural after-effects (Koehler and Wallach, 1944), but most likely includes various phenomena with different mechanisms. Here we performed a series of psychophysical experiments in human subjects to investigate the nature of size illusion caused by a figural after-effect protocol. After an inspection stimulus (IS) was flashed 4 times for a period of 250 ms with an interval of 250 ms, a test stimulus and a reference stimulus were presented side-by-side near and away from the IS for 300 ms. The reference was always 2º in size, and the size of the test stimulus was varied across trials over a range of 1.6-2.4º (80-120% of the reference). The position of the two stimuli was randomly interchanged. The subjects judged which of the two stimuli appeared larger. By measuring the point of subjective equality, we assessed the effect of an IS on the perceived size of a test stimulus. The change in the perceived size depended on the size of an IS; the perceived size became larger following an IS of 1º, while it became smaller after an IS of 1.68º or larger. The bi-directionality of the effect suggests that biased spatial attention does not fully explain the illusion. The shrinkage effect by the IS smaller than the test and reference stimuli (1.68º) indicates that a simple antagonism between presumed smallness and largeness representations in the brain does not explain the illusion. The illusion was also different from color after-effect and motion after-effect in that it occurred when there was no spatial overlap between an IS and a test stimulus. Finally, this size illusion occurred when an IS was presented to one eye and test/reference stimuli to the other eye, suggesting that the neural mechanism resides beyond the stages after the convergence of binocular information.
P2-1-130 Gabor wavelet領域逆相関法を用いたサルV2野細胞の反応特性の解析 Application of Gabor wavelet domain reverse correlation to visual neurons in macaque V2 ○稲垣未来男1, 佐々木耕太1,2, 橋本肇1, 伊藤南3, 浅川晋宏4, 大澤五住1,2 ○Mikio Inagaki1, Kota S. Sasaki1,2, Hajime Hashimoto1, Minami Ito3, Kunihiro Asakawa4, Izumi Ohzawa1,2 阪大院・生命機能1, 脳情報通信融合研究センター2, 東京医科歯科大学3, 東京慈恵会医科大学4 Grad Sch Frontier Biosciences, Osaka Univ, Osaka1, Center for Information and Neural Networks (CiNet), Osaka2, Tokyo Medical and Dental University, Tokyo3, Jikei University School of Medicine, Tokyo4 Spatial receptive fields of simple cells in V1 can be described by two-dimensional (2D) Gabor functions with an appropriate combination of spatial parameters (such as position, scale, orientation, and so on). Wide varieties of the parameters across the V1 population suggest that simple cells as a whole function as a bank of the 2D Gabor filters. Visual neurons downstream, therefore, have an opportunity to collect outputs from simple cells that resemble the 2D Gabor filters. Here, in order to investigate this type of integration over simple cells, we developed a new analytical method by extending standard white-noise analyses. We refer to this analysis as "Gabor wavelet domain reverse correlation" since spike-triggered average is computed after transforming the stimuli in the 2D space domain into those in the 2D Gabor wavelet domain. As a result, receptive field profiles of a single neuron are characterized as a collection of weights of the 2D Gabor filters. We applied this analysis to the responses of V2 neurons recorded from anesthetized monkeys. For a subset of V2 neurons, we successfully assessed orientation preference at each local position within a receptive field by evaluating the filter weights independently. For these neurons, the orientation preference was mostly homogenous across the receptive field. We did not observe an obvious inhomogeneity in this respect that is explicitly consistent with angle selectivity of V2 neurons reported in previous studies. Because our analysis does not take any interactions between the filters into consideration so far, further analyses focusing on the interactions may be needed to fully understand receptive field profiles of V2 neurons.	P2-1-131 両眼間輝度反転ステレオグラムに対する反転奥行き知覚は相対奥行き参照枠において起きる Reversed depth perception in binocularly anticorrelated stimuli occurs in a relative depth reference frame ○青木俊太郎1, 塩崎博史1, 藤田一郎1 ○Shuntaro C. Aoki1, Hiroshi M. Shiozaki1, Ichiro Fujita1 大阪大学院・生命機能1 Grad Sch Front Biosci, Osaka Univ, Toyonaka, Japan1 Cortical areas MT-MST and V4-IT encode binocular disparity using different computations. For spatially adjacent visual features, areas MT and MST calculate binocular correlation and encode absolute disparity of each feature, whereas areas V4 and IT detect matched components in two retinal images and encode relative disparity between the features. Binocularly contrast-reversed (anticorrelated) stereograms surrounded by correlated stimuli produce depth perception in which depth direction is reversed. This reversal is explained by correlation computation, suggesting involvement of MT and MST. To further examine the contributions of MT-MST and V4-IT, we investigated the reference frame where the reversal of perceived depth takes place. Dynamic random dot stereograms (RDSs) composed of a center disk and a surrounding annulus were presented to human subjects discriminating depth between the center and surround. The center disk was either a correlated or an anticorrelated RDS (cRDS or aRDS) and had absolute disparity fixed at zero. The surround was a cRDS and its absolute disparity varied across trials, resulting in variable relative disparity between the center and surround. The depth perception of five of six subjects was reversed for aRDSs when the surround disparity was deviated from zero, suggesting depth reversal in a relative reference frame. Vergence eye movement did not explain the reversal since subjects still reported the reversed depth for a brief stimulus presentation (94 ms) which is too short for occurrence of eye movement. The performance of subjects dropped to chance levels when an uncorrelated RDS was presented at the center, indicating that they did not make decisions based solely on the surround disparity. We suggest that the reversal of perceived depth in anticorrelated stimuli occurs in a relative reference frame. Disparity signals represented in various areas may contribute to the reversed depth perception.
P2-1-132 情報量最大化原理によるV2の選択性の説明 Information maximization in a multilayer network explains the properties of V2 neurons ○藤原謙三郎1, 田中琢真1, 中村清彦1 ○Kenzaburo Fujiwara1, Takuma Tanaka1, Kiyohiko Nakamura1 東工大・総理工・知能システム科学1 Dept. Computational Intelligence and Systems Science, Tokyo Tech, Tokyo1 Neurons in the primary visual cortex (V1) exhibit selectivity to retinal position, orientation, direction of motion, and color of visual stimuli. V1 neurons are classified into simple cell, complex cell, and hypercomplex cell by their response properties. These V1 neurons can be regarded as edge detectors in the visual information processing system. Information maximization principle (infomax) is a hypothesis proposed to explain the receptive field properties of sensory cortices including V1. Previous studies showed that the models based on information maximization principle explain the receptive field properties of V1 neurons. V2 neurons detect much more complex features, such as curvature and illusory contour, than edges detected by V1 neurons. The receptive field properties of V2 neurons have not been explained by information maximization principle. In this presentation, we propose a model to replicate the receptive field properties in V2 on the basis of information maximization principle. We apply a learning method based on information maximization principle to a multilayer feedforward network and examine the response properties of model neurons comparing them with previous experiments.	P2-1-133 マウス頭頂連合野から一次視覚野への抑制投射遮断による視力低下 Reduced visual acuity after blockade of feedback inhibitory projections from the parietal association area to the primary visual cortex in mice ○菱田竜一1, 堀江正男2, 塚野浩明1, 任海学1, 竹林浩秀2, 澁木克栄1 ○Ryuichi Hishida1, Masao Horie2, Hiroaki Tsukano1, Manabu Tohmi1, Hirohide Takebayashi2, Katsuei Shibuki1 新潟大・脳研・システム脳生理1, 新潟大・医歯学総合・解剖第22 Dept Neurophysiol, Brain Res Inst, Niigata Univ, Niigata, Japan1, Div Neurobiol and Anat, Grad Sch Med and Dent, Niigata Univ, Niigata, Japan2 It is thought that feedback projections from higher association areas modulate sensory functions in the neocortex. However, it remains to be elucidated how higher association areas modulate functions of the primary areas. To better understand the neural mechanisms underlying the feedback control of sensory functions, we studied properties of inhibitory projections from higher areas to the primary visual cortex (V1) in mice using transcranial flavoprotein fluorescence imaging. A decrease in flavoprotein fluorescence signals was observed in V1 immediately after electrical stimulation applied to the parietal association area. Electrophysiological unit recordings in V1 revealed that the decrease reflects the inhibition of basal neural activity. Using the decrease in flavoprotein fluorescence signals for monitoring the feedback inhibition, we found that electrical stimulation applied to parietal association cortex, posterior cingulate cortex and medial prefrontal cortex led to inhibition of V1 neurons. We investigated the functions of the feedback projections by blocking the pathways from parietal association area to V1. In acute experiments, the cortical cuttings between V1 and the parietal association area using a razor blade or the muscimol injections into the parietal association area resulted in the potentiation of visual responses in V1, indicating the presence of tonic inhibition mediated by the feedback projections. In the chronic experiments, the mice with the cortical cuttings showed a significant reduction of V1 response to grating stimuli with reduced contrasts. Tracer injections to the parietal association cortex showed the presence of direct cortico-cortical pathways and indirect pathways via reticular thalamic nucleus, both of which may mediate the feedback inhibition. These observations strongly suggest that feedback inhibitory projections from parietal association area to V1 can adjust neural responses in V1 for keeping appropriate visual acuity.
P2-1-134 錯視知覚には視床枕が関与する Pulvinar is involved with the perception of the optical illusions: an fMRI study ○田部井賢一1, 佐藤正之1, 木田博隆1, 木崎萌丹2, 佐久間絵2, 佐久間肇3, 冨本秀和1,4 ○Ken-ichi Tabei1, Masayuki Satoh1, Hirotaka Kida1, Moeni Kizaki2, Haruno Sakuma2, Hajime Sakuma3, Hidekazu Tomimoto1,4 三重大院・医・認知症医療学1, 三重大・医2, 三重大院・医・放射線医学3, 三重大院・医・神経病態内科学4 Dept Dementia Prevention and Therapeutics, Mie Univ, Mie, Japan1, Fac Medicine, Mie Univ, Mie, Japan2, Dept Radiology, Mie Univ, Mie, Japan3, Dept Neurology, Mie Univ, Mie, Japan4 The findings of neural processing of optical illusion provide unique clues for exploring the neural mechanisms of visual perception. Previous studies showed that some visual areas contributed to optical illusions such as illusory contours and Muller-Lyer illusion, however, the neural mechanisms underlying optical illusion have not been clearly identified. Using fMRI, we identified the activated brain regions during the perception of optical illusions. Eighteen young healthy subjects were engaged. The optical illusion stimuli were made of kana: the Japanese phonograms. In the shape task, participants replied aloud whether they perceived the shape of two optical illusions as the same one or not. In the word task, participants read aloud the word kana letters in stimuli. Using SPM8, the direct comparison between shape and word tasks showed activations of the left inferior frontal gyrus, the medial frontal gyrus, and the right pulvinar. The direct comparison between word and shape tasks did not show any significant activation. It is well known that there are two visual pathways: geniculate and extrageniculate systems. The pulvinar belongs to the latter system and the present results suggest that extrageniculate system may be involved in cognitive processing of optical illusions.	P2-1-135 二光子顕微鏡を用いた麻酔、覚醒状態での視覚応答特性の比較 Changes in visual responses of rat cortical neurons from the anesthetized to awake states, revealed by two-photon functional imaging ○木村塁1, 惣谷和広1, 尾関宏文1, 蝦名鉄平1, 柳川右千夫2, 津本忠治1 ○Rui Kimura1, Kazuhiro Sohya1, Hirofumi Ozeki1, Teppei Ebina1, Yuchio Yanagawa2, Tadaharu Tsumoto1 理化学研究所 脳科学総合研究センター1, 群馬大院・医2 Brain Science Institute, RIKEN, Wako1, Graduate School of Medicine, Gunma University, Gunma2 To understand neural mechanisms underlying perception and behavior, it is necessary to compare activities of the same population of neurons between anesthetized and awake states. To address this issue, we used in vivo two-photon functional calcium imaging and examined how changes in the brain state modulate visual responses of excitatory and inhibitory neurons in the primary visual cortex. Functional imaging of cortical neurons was carried out by injecting calcium indicator (fura 2-AM) and astrocyte marker (SR101) into layer 2/3 of the primary visual cortex of adult VGAT-Venus transgenic rats, in which inhibitory neurons were identified with Venus fluorescence. Comparing the responses between the two states (monitored by EEG activity), we found that some neurons with no visual responses in the anesthetized state became responsive in the awake state. For the neurons which responded in both conditions, the response amplitude of excitatory neurons did not change, but that of inhibitory neurons increased in the awake state. In addition, we found no change in orientation tuning bandwidth between both states. We also calculated the correlation of responses among simultaneously recorded neurons, and found that the correlated activities increased in excitatory to excitatory neurons pairs in the awake. Functional significance of these findings will be discussed.
P2-1-136 ラットの視神経の有髄部：免疫組織化学法を用いて解析した化学構築による新たな区分 The myelinated part of the rat optic nerve: A new sub-division scheme based on chemoarchitecture analyzed by immunohistochemistry ○河野純1 ○June Kawano1 鹿児島大院・医歯・神経解剖1 Lab for Neuroanat, Kagoshima Univ, Kagoshima, Japan1 Background Our previous study showed that the myelinated (m) part of the rat optic nerve was divided into 2 subparts: myelinated parts 1 (m1) and 2 (m2; Neuroscience 2012, P2-a26). However, it is unclear whether chemoarchitecture is homogeneous in each subpart. Method Distribution of GFAP (glial fibrillary acidic protein), oligodendrocyte-marker protein (RIP), GS (glutamine synthetase), and of NFH (neurofilament heavy chain) was investigated in the m part by using fluorescent double-labeling immunohistochemistry. Results The m part was divided into 2 subparts: m1 and m2. The m2 part occupied a major portion of the optic nerve and chemoarchitecture of the m2 part was almost homogeneously organized. Different from the m2 part, chemoarchitecture of the m1 part was not homogeneously organized, and this subpart was divided equally spaced at one-third length into 3 sub-subparts: myelinated parts 1a (m1a), 1b (m1b), and 1c (m1c). Chemoarchitecture in the m1a part was significantly different from that in the m2 part. Between the m1a and m2 parts, gradual changes of chemoarchitecture were seen. Astrocytic filaments little by little decreased in density from a high level in the m1a part to a moderate level in the m2 part. Myelinated fiber-bundles gradually increased in tightness from loose in the m1a part to tight in the m2 part. GS-immunoreactivity (GS-IR) little by little decreased in strength from a moderate level in the m1a part to a weak level in the m2 part. NFH-immunoreactive axons gradually increased in tightness from loose in the m1a part to moderate in the m2 part. In addition, most of moderately GS-immunoreactive glial-cells were oligodendrocytes in the m part. Conclusion We created a new sub-division scheme in the m part which was divided into 4 subparts: m1a, m1b, m1c, and m2. The scheme may work as a new system for performing pathological analysis of rodent models of human optic nerve diseases such as glaucoma, and multiple sclerosis (neuromyelitis optica).	P2-1-137 ガボールウェーブレット領域におけるサブスペースマッピング Subspace mapping in Gabor wavelet domain ○佐々木耕太1,2, 大澤五住1,2 ○Kota S. Sasaki1,2, Izumi Ohzawa1,2 大阪大院・生命機能・脳神経工学1, 脳情報通信融合研究センター2 Grad Sch of Frontier Biosci, Osaka Univ, Osaka, Japan1, Center for Information and Neural Networks (CiNet), Osaka, Japan2 What stimulus is suitable for examining the functional properties of visual neurons? Two extreme examples of visual stimulus are drifting grating varied only in one parameter and dynamic noise. The former stimulus is specialized for measuring tuning to one stimulus parameter. One cannot investigate how other parameters affect the activity of neurons. Moreover, once the stimulus parameters were optimized for a particular neuron, other neurons that are recorded simultaneously might not be driven effectively, which is undesirable in a multi-electrode experiment. On the other hand, the latter stimulus covers vast stimulus space sparsely. It is generally time-consuming because individual neurons responds to small fractions of stimulus space. Ringach et al. (1997) proposed a compromise between the two extremes: subspace mapping. Based on its idea, we devised visual stimulus to examine single neurons' tuning to various parameters. V1 simple cells are thought to be elementary building blocks for visual information processing in the cerebral cortex. We devised visual stimulus combining visual features that resemble the receptive fields of simple cells. More specifically, first, a rich repertoire of drifting Gabor elements (space-time wavelets) were prepared ((27 + 67 + 115) positions, 24 directions of motion, 4 initial spatial phases, 3 temporal frequencies (TF); 27 positions for the lowest spatial frequency (SF), 67 for middle, 115 for the highest, in a half octave step). Among these 60,192 elements, several elements were picked up in every stimulus frame successively. Once a Gabor element was picked up, it was activated for 500 ms, and then extinguished. Using the novel subspace mapping stimulus, we recorded the responses of single neurons in cat V1. When the data were analyzed, the responses were averaged for spatial phase and TF. We obtained the position of the receptive fields and tuning to the direction of motion and SF of single neurons in < 30 min.
P2-1-138 Posterior cortical atrophyの診断のため劣化ガボールウェーブレットを用いたシミュレーション A degraded Gabor wavelet simulation for diagnoses of Posterior Cortical Atrophy ○加藤大典1, 佐々木耕太1,2, 大澤五住1,2 ○Daisuke Kato1, Kota S. Sasaki1,2, Izumi Ohzawa1,2 阪大・生命機能1, 脳情報通信融合研究センター2 Graduate School of Frontier Biosciences, Osaka University, Osaka1, Center for Information and Neural Networks (CiNet), Osaka2 Posterior cortical atrophy (PCA) is a neurodegenerative syndrome characterized by a progressive, often striking, and fairly selective decline in visual-processing skills and other functions that depend on parietal, occipital, and occipital-temporal regions of the brain. Patients with PCA often face considerable delays in diagnosis. The purpose of this study is to develop a simple method to find a sign of PCA in the early stage. As the first step toward this goal, we examined what primary visual cortex can inform higher visual cortex when primary visual cortex is affected by PCA. A simple cell in primary visual cortex has a receptive field which resembles a Gabor function. When the receptive fields of simple cells are compared across neurons, they are similar to each other except for its orientation and size. These features are modeled as a Gabor wavelet pyramid on a computer. In this study, the characteristics of Gabor wavelets were tailored to mimic those of actual simple cell receptive fields (tuning bandwidth in the frequency domain = 1.5 octaves, aspect ratio = 1:1). Preferred orientation and spatial phase were discretized into eight (22.5° steps) and two values (0° and 90°), respectively. Typically, seven scales (or sizes) of Gabor wavelets were used in one octave step. We tested several patterns including a Landolt-C chart. First, these patterns were decomposed using the Gabor wavelet pyramid described above. To simulate PCA-affected primary visual cortex where some simple cells do not function, we eliminated a fraction of Gabor wavelets on purpose when images were reconstructed. The resultant images were degraded similarly across all scales. This characteristic is substantially different from usual eye diseases (e.g. cataract, near-sightedness, etc.), which affect contrast sensitivity and visual acuity.	P2-1-139 判断とニューロン活動の相関はボトムアップ信号によるかトップダウン信号によるか？MT野ニューロンの発火頻度変動の解析 Is decision related response modulation of sensory neurons due to bottom-up or top-down signal?: Analysis of trial-to-trial spike count variability in MT neurons ○熊野弘紀1, 宇賀貴紀1 ○Hironori Kumano1, Takanori Uka1 順天堂大院・医・神経生理1 Dept Neurophysiol, Grad Sch of Med, Juntendo Univ, Tokyo, Japan1 Decision related response modulation (choice probability: CP) is observed in many sensory cortex and is believed to represent the association between perceptual choice and neuronal activity. Various explanations for this association have been proposed including bottom-up and top-down accounts of CP. It has, however, been difficult to distinguish between these possibilities. Here we report a new method for distinguishing between a bottom-up and top-down account of CP. We first predicted the amount of trial-to-trial spike count variability based on bottom-up and top-down models. In the bottom-up model, simulated sensory neurons fired with a spike count variability that was equal to the mean. Decisions were made via a circuit consisting of multiple independently firing neurons. On each simulated trial, responses of a simulated sensory neurons and the decision outcome was determined, and CP was calculated across trials. In this case, the overall variability in response to a visual stimulus did not change depending on CP, but variability measured separately for each behavioral choice decreased with higher CP. In the top-down model, a decision dependent signal was added to the neuron's response. In this case, variability measured separately for each behavioral choice did not depend on CP, but the overall variability increased with higher CP. To determine which model better described the data, we analyzed CP measured in area MT during a task switching paradigm, where two tasks (direction discrimination and depth discrimination task) were randomly interleaved. Measurement of CP in two tasks allowed us to analyze the relative change in neuronal variability independent of the large fluctuations across individual neurons. The overall variability in response to visual stimulus did not change depending on CP. Conversely, variability measured separately for each behavioral choice decreased with higher CP. Overall, our results are consistent with the bottom-up account of CP.
P2-1-140 初期視覚野細胞の位相空間による運動選択性解析 Phase-domain analysis of motion selectivity of visual cortical neurons ○中園貴之1,2, 大澤五住1,2 ○Takayuki Nakazono1,2, Izumi Ohzawa1,2 大阪大院・生命機能1, 脳情報通信融合研究センター2 Grad School of Frontier Biosciences, Osaka University1, Center for Information and Neural Networks(CiNet), Osaka2 In most previous neurophysiological studies of the neural basis of motion perception, direction and speed selectivities of visual cortical neurons were conducted in the space-time domain. However, as Simoncelli & Heeger model (1998) indicates, there is evidence that integration of motion information in the high-order visual areas may be better understood in the space-time frequency domain. It is therefore important to characterize motion and direction selectivity properties of the early visual areas in the frequency domain. For this purpose, we analyzed response properties of neurons in terms of interactions of two spatial frequency components contained in random-dot stimuli. Specifically, for a given spatial frequency component, motion information is predominantly encoded as changes of its spatial phase over time. We have therefore examined interaction of the optimal spatial frequency component across two stimulus frames separated in time. Responses of neurons were recorded from single isolated neurons in the primary visual cortex(A17,18) in anesthetized cat during presentations of ternary dynamic random-dot stimuli. Spike-triggered analyses (STA) were performed for a pair of phases for a selected spatial frequency components. Simple cells responded well when two specific phases were contained in stimuli as expected from their spatial receptive fields. And for direction-selective neurons, the two specific phases were dependent on the interval of two stimuli pair. On the other hand, complex cells showed responses as long as the phase difference of the frequency component was at a specific value for a given temporal separation. Our results suggest that the frequency-domain analysis method provides a robust basis on which next integration stage of visual neural processing may be undertaken. Analyses of integrations over multiple frequency components and over multiple neurons are two of the future directions in this approach.	P2-1-141 様々な色相、彩度、輝度の組み合わせからなる色刺激に対するV1, V2, V4細胞の応答 Visual responses to color stimuli with a variety of combinations of hue, saturation and luminance in Macaque V1, V2, and V4 ○田村弘1, 高田悠史1 ○Hiroshi Tamura1, Hisashi Takada1 大阪大院・生命機能・認知脳1 Lab Cog Neuro, Front Bio, Osaka Univ, Osaka1 Colors of objects plays an important role in visual recognition of objects. Colors can be described by means of three dimensions, i.e., hue, saturation and luminance. In this study, we aim to reveal how the colors having a variety of combinations of hue, saturation and luminance are represented and processed in visual cortical areas. For this purpose, we devised a set of 238 stimulus colors, which were sampled at regular intervals [12 hue directions, 5 saturation levels, 11 luminance levels, not all the combinations were presented because of limitations of the display (organic electroluminescent display, PVM-1741, SONY)] in the DKL (Derrington Krauskopf Lennie) color space that represents the color in three dimensions based on the response ratio of three cones in the retina. We measured firing rate of action potentials of single neurons to the color stimulus set. Recordings were conducted from V1, V2, and V4 of two analgesized (fentanyl citrate) and immobilized (vecronium bromide) macaque monkeys (Macaca fuscata). Neurons that responded selectively to the color stimuli were found in all cortical areas [V1, 53.6% (98/183); V2, 52.3% (46/88); V4, 41.0% (75/183)]. Responses to the color stimulus set of V4 neurons were different from those of V1 and V2 neurons. Selectivity to the color stimuli of neurons in V4 was sharp compared to those in V1 and V2 (sparseness and aggregation indices, p < 0.001, K-W test). Preferred colors of V4 neurons were closer to the origin of the DKL color space than those of V1 and V2 neurons (p = 0.006, K-W test), indicating that V4 neurons preferred low contrast and low saturation colors. Furthermore, the preferred colors of V4 neurons distributed more uniformly in the DKL color space, whereas those of V1 and V2 neurons were distributed at the edge of the color space. These results suggest that V4 neurons represent wide variety of colors that have various combinations of hues, saturation and luminance.
P2-1-142 物体表面画像に対するサル腹側視覚経路神経細胞の応答 Visual responses of neurons to colored texture images of natural objects ○石田秀太1, 大塚晴輝1, 高田悠史1, 井上祐哉1, 山根ゆか子1, 田村弘1 ○Shuta Ishida1, Haruki Otsuka1, Hisashi Takada1, Yuya Inoue1, Yukako Yamane1, Hiroshi Tamura1 大阪大院・生命機能・認知脳1 Lab Cog Neuro, Front Bio, Osaka Univ, Osaka1 Object colors and textures are useful cues for visual object recognition. In the present study, we examined responses of neurons in the ventral visual pathways to understand the representation of surface properties of objects. We prepared a set of 64 object-surface images derived from 8 object categories, such as stone, bark, leaf, flower, fruit, butterfly wing, feather and skin/fur. We cut out square region (6 deg in visual angle) of object images to remove the outer contour. Action potentials (spikes) were recorded from the primary visual cortex (V1), V4, and the inferior temporal cortex (IT) of analgesized (fentanyl citrate) and immobilized (vecronium bromide) monkeys (Macaca fuscata, n = 4). Visually responsive neurons were observed in all the 3 areas [V1, 55% (738/1333); V4, 78% (704/903); IT, 46% (305/665)]. Sharpness of stimulus selectivity evaluated with a sparseness index was not different among the 3 areas. To analyze the properties of visual responses to the images, we examined the relation between responses and image features such as orientation, spatial frequency, luminance and colors. Reponses of most neurons in V1 (92%, 677/738) correlated with at least one of the image features (r2 > 0.06, p < 0.05), while in V4 and IT the incidences of correlated responses were lower than that of V1 [V4, 69% (487/704); IT, 74% (227/305); p < 0.001]. The coefficients of determination (r2) between responses and the image feature values of V1 neurons were higher than those of neurons in V4 and IT (V1, 0.19; V4, 0.09; IT, 0.10, median; K-W test, p < 0.001). These results suggest that responses of most V1 neurons to the object-surface images are explained with the orientation, spatial frequency, and/or color selectivity, while the responses of some of the neurons in V4 and IT were constructed in a manner different from that of V1.	P2-1-143 マカカ属サル下側頭葉皮質神経細胞の物体選択性に対する背景自然画像の影響 Natural scene backgrounds affect object-image selectivity of neurons in the macaque inferior temporal cortex ○向将充1, 山根ゆか子1,2, 伊藤淳司3藤田一郎1,2, 田村弘1,2 ○Masamitsu Mukai1, Yukako Yamane1,2, Junji Ito3, Serge Strokov3, Sonja Gruen3,4, Ichiro Fujita1,2, Hiroshi Tamura1,2 大阪大院・生命機能・認知脳1, 脳情報通信融合研究センター2, INM-6, Forschungszentrum Jülich3, Theoretical Systems Neurobiology4 Lab Cog Neuro, Grad Sch Front Bio, Osaka Univ, Osaka1, CiNET, Osaka2, INM-6, Forschungszentrum Jülich, Germany3, Theoretical Systems Neurobiology, RWTH Aachen Univ, Germany4 The inferior temporal (IT) cortex is a higher visual area in the ventral visual pathway that is crucial for the visual object recognition. Studies of IT neurons have been focused on responses to isolated objects presented on a plain background. However, because objects in realistic conditions are embedded in complex backgrounds and receptive fields of IT neurons are large, object representation of IT neurons may involve interactions between objects and their background. We investigated whether and how the presence of a complex background scene affects the response selectivity of IT neurons to object images. We prepared 448 images (64 objects on 7 natural-scene backgrounds). The spiking activities of IT neurons were recorded extracellularly from analgesized (with fentanyl citrate) and immobilized (with vecuronium bromide) monkeys with a linear array electrode (Neuronexus). We recorded 110 neurons in 5 recording sessions, of which 73 (66%) were visually responsive. The responses of 86% (63/73) of these neurons were responsive on the object images, those of 75% (55/73) on the background images, and/or those of 58% (42/73) on the combination of the two (p<0.01, two-way ANOVA). To quantify the invariance of stimulus preferences to changes of the background images, we calculated the rank-order correlation between the responses to the stimulus set on two different backgrounds, for all possible pairs of the backgrounds. In most of the neurons (86%, 64/73), the responses were not correlated from each other (r≤0.31, p ≥ 0.01, test for independence), at least for one pair of backgrounds. Selectivity to foveated object images of most IT neurons was thus not invariant to changes in backgrounds. We suggest that the representation of visual objects changes depending on the background scenes.
P2-1-144 ネコ一次視覚野ニューロンは物体表面の質感に選択性を持つ Neurons in the cat primary visual cortex is selective for skewed-statistics-related surface properties of visual images ○内藤智之1, 末松尚史2, 三好智満3, 澤井元3, 佐藤宏道1,2 ○Tomoyuki Naito1, Naofumi Suematsu2, Tomomitsu Miyoshi3, Hajime Sawai3, Hiromichi Sato1,2 阪大院・医・認知行動1, 阪大院・生命・認知行動2, 阪大院・医・統合生理3 Grad Sch Med, Osaka Univ, Toyonaka1, Grad Sch Front Bio Sci, Osaka Univ, Toyonaka2, Grad Sch Med, Osaka Univ, Suita3 Natural images contain substantial variation in first-order statistics (mean) of luminance histogram that corresponds to local luminance of stimulus, and second-order statistics (variance) corresponding to local contrast of stimulus. An elucidation of neural selectivity for these variations is important for understanding the functional significance of visual neurons under natural viewing conditions. Neuronal sensitivity to local luminance and local contrast is well investigated in neurons in the early visual pathway. However, little is known about neuronal selectivity for variations in third- or more higher-order statistics of luminance histogram of visual neurons. We investigated neuronal selectivity of neurons in the lateral geniculate nucleus (LGN) and the primary visual cortex (V1) of anesthetized cat for third-order statistics (skewness) of luminance histogram by using dynamic dense noise stimuli of which mean and variance of luminance histogram were kept constant while the skewness was changed systematically. We found that most of V1 neurons exhibited clear sensitivity to the skewness, while neurons in the LGN exhibited little or no selectivity. There were three types of skewness preference of V1 neurons: 1) negative skewness preference, 2) near zero skewness preference, and 3) positive skewness preference. Many psychophysical studies have so far suggested that there are neural mechanisms sensitive to skewed statistics of visual images, and that their outputs can be used in estimating surface properties. Our results suggest that the primary visual cortex is the first stage that shows the selectivity for the skewed-statistics-related surface properties.	P2-1-145 周波数-眼間位相差領域におけるV1視差選択性細胞の応答特性 Analysis of V1 disparity-selective neurons in the spatial frequency-phase difference domain ○馬場美香1, 佐々木耕太1,2, 大澤五住1,2 ○Mika Baba1, Kota S. Sasaki1,2, Izumi Ohzawa1,2 大阪大院・生命機能・脳神経工学1, 脳情報通信融合研究センター2 Grad Sch Frontier Biosciences, Osaka Univ, Osaka, Japan1, Center for Information and Neural Networks (CiNet), Osaka, Japan2 Visual information from the two eyes is converged onto single neurons in the brain to perceive stereoscopic depth. At its initial stages, many neurons in the primary visual cortex show selectivity to binocular disparity. What rule is applied for these neurons when they combine binocular information from multiple sources to encode binocular disparity? Binocular disparity can be represented as difference in spatial phase between the two eyes. When binocular disparity is constant, difference in spatial phase is proportional to spatial frequency (SF) (assuming that there is no offset in receptive field positions). We asked whether neurons in the primary visual cortex are built on this relationship. The responses were recorded from single neurons in the cat primary visual cortex, presenting dynamic 1D dense line stimuli to both eyes. The patterns were uncorrelated between the two eyes. The orientation of lines was matched to the preferred orientation of a recorded cell. These stimuli allow one to examine tuning to difference in spatial phase between the two eyes for various SFs simultaneously. Specifically, we first selected spike-triggered stimuli using a reverse correlation technique. The responses to difference in spatial phase between the two eyes were then calculated for each SF component in these stimuli after the spike-triggered stimuli were transformed into the SF domain separately for each eye. These responses were accumulated for all spike-triggered stimuli to obtain tuning to difference in spatial phase between the two eyes. We found that, in some complex cells, the preferred phase difference was proportional to SF. This characteristic was basically absent in simple cells. Thus, these results suggest that a particular set of subunits tuned to different SF are combined in primary visual cortex so that a complex cell can encode a constant binocular disparity across a range of SF.

上部に戻る

前に戻る