- Email: [email protected]
Multinomial Bayesian network model reproducing receptive field properties of V1 simple cells and V2
Abstracts / Neuroscience Research 71S (2011) e108–e415
e349
gated into eye specific domains. Functional CG responses are evident during the early phase of innervation (P7), however paired pulse facilitation does not fully mature until the end of the second postnatal week. To test whether CG innervation relies on retinal input, we did enucleation and genetic deafferentation. In both cases, CG axons continue to target and innervate dLGN but do so in a highly accelerated manner. Thus, these results suggest that RG and CG innervation of dLGN occurs in a coordinated fashion and that retinal axons are involved in regulating the timing of CG innervation. Moreover preliminary experiments in other transgenic strains that allow for the visualization of projections that arise from cholinergic brainstem nuclei and thalamic reticular nucleus show a similar delay in dLGN innervation that is also disrupted by the loss of retinal input. Taken together, these results suggest that the development plan regulating nonretinal innervation of visual thalamus is highly conserved. Research fund: EY12716 (WG).
effect of static standing contrast and consequences of its fading, we added a pedestal contrast offset to the base sequence. The resulting stimuli had 7 contrasts within the 0–50% range and was without phase inversion. The orientation and spatial frequency of gratings were fixed at the optimal value of each cell. We found that V1 neurons showed a shift of contrast response function during pedestal conditions such that a physically blank screen (i.e. 0% contrast stimulus) generated clearly detectable responses. The amount of shift was reasonably predicted by the arithmetic mean of contrasts presented in each experimental condition. These results support the hypothesis that a negative image is generated during visual fading, and it behaves as if it were a physical entity. Additionally, we have analyzed time course of shifts of contrast response functions using a sliding analysis window over time. Time course of the shift was roughly comparable to that we observe perceptually. Research fund: Kakenhi 22135006, 22300110, and Global COE.
doi:10.1016/j.neures.2011.07.1528
doi:10.1016/j.neures.2011.07.1530
P4-h16 Object-selective responses evoked by phase-shift motion stimuli in macaque inferior temporal cortex: A case study using chronically implanted multielectrode arrays
P4-h18 Vertical and horizontal correlations of neuronal activities in the cat primary visual cortex
Ryusuke Hayashi 1,2 1
System Neuroscience Group, AIST, Tsukuba, Japan 2 PRESTO, JST, Kawaguchi, Japan Introduction: It has been shown that neurons in the inferior temporal cortex (IT) have shape selectivity that is relatively invariant with visual cue (Sary et al., 1993). Here, I revisited the motion invariance of shape selectivity in the IT cortex and investigated neural responses elicited by object images with phase-shift motion: images are perceived to move in one direction, while, unlike ordinary position-shift motion, their positions remain stationary. Methods: Multi-unit activities were recorded using two multielectrode arrays (32 electrodes each) chronically implanted in the right IT cortex of a macaque monkey. The animal was trained to do a fixation task during the recording: one of 25 object images, moving either leftward or rightward, was randomly chosen and presented for 2 s. The cycle of the phase shift was set either at 100, 133, 200, 267, or 400 ms. Results: The phase-shift operation alternated the polarity of image contrast every half cycle, which caused the oscillation in IT neurons’ responses when the cycle was longer than 200 ms. The oscillation, however, vanished when the cycle was shorter than 133 ms, i.e. the time equivalent as the mean onset latency of IT neurons. In the case of shorter cycle, the responses were activated only by the specific object images, regardless of motion direction. In addition to this motion invariance of shape selectivity, I observed that phase-shift motion elicited optokinetic nystagmus (OKN, involuntarily evoked oscillatory eye movement) in the monkey. Discussion: The author have recently developed a new binocular rivalry display that consists of two different object images with opponent phase-shift motions and showed that the perceptual alternation in human observer is accompanied with the switch of OKN. The present study suggests that the same binocular rivalry experiment is applicable to monkey to know its perceptual state from the evoked OKN and phase-shift motion of short cycle does not affect IT selectivity. Research fund: JST PRESTO program. doi:10.1016/j.neures.2011.07.1529
Hiroki Tanaka , Izumi Ohzawa Grad. Sch. of Front. Biosci., Osaka Univ., Toyonaka, Japan Primary visual cortex (V1) consists of vertically arranged functional columns, which are divided into 6 tangential layers. Cells in different layers and columns substantially interact by vertical and horizontal connections. In spite of many studies, our knowledge of these interactions is limited. Specifically, while correlation of neuronal activities is classified into temporally precise one (due to monosynaptic connection or strong common inputs) and loose one (due to polysynaptic connection or weak common inputs), how do 2-d structures of these correlations (relationship of correlation with laminar structures or vertical and horizontal distance between neurons) differ? To address these questions, we penetrated multi-array electrodes into cat V1 and recorded activities simultaneously from many neurons located in different columns and layers. We examined correlations of activities of isolated single neurons under visual stimulation or in spontaneous conditions. We found that the loose correlation extends widely across columns and layers. On this temporal scale, the vertically separated pairs showed consistent offset of correlated discharges that is linearly related with vertical distances of the pair (neurons in the deeper layers tend to fire prior to neurons in the superficial layers), but horizontally separated pairs did not show such offset. On the other hand, precise correlation was highly localized spatially and consistent offset was not observed. We also found that both type of correlations, especially the loose one, was stronger for superficial layers. These results suggest that the loose correlation reflects propagation of inputs from deep layers to upper layers along vertical connections and these inputs are widely synchronized across columns through horizontal connections. On the other hand, precise correlation reflects much shorter range of connections both in vertical and horizontal directions. Research fund: HT was supported by Kakenhi 21700433. IO was supported by Kakenhi 22135006, 22300110, and Global COE. doi:10.1016/j.neures.2011.07.1531
P4-h19 Multinomial Bayesian network model reproducing receptive field properties of V1 simple cells and V2 Haruo Hosoya 1,2
P4-h17 Neural correlates of fading illusion revealed in responses of V1 neurons Kohei Kurihara , Kota S. Sasaki, Izumi Ohzawa Grad. Sch. of Frontier Biosciences, Osaka Univ., Osaka, Japan When a static image without sharp edges is fixated somewhat peripherally, the image is perceived to fade away gradually. This phenomenon is known as Troxler fading or fading illusion. Traditionally, at least an aspect of this illusion is explained by a hypothesis that physical stimulus is cancelled by a negative image generated in the visual pathway, because the negative image is perceived momentarily when the physical stimulus is suddenly removed. One prediction that may be tested neurophysiologically is that, in the faded or adapted state, visual neurons actually respond to a blank screen, because the negative image is there even when no physical stimulus is present. To test this prediction in cat V1, we measured the base contrast response function of neurons by presenting flash grating stimuli of various contrasts (7 contrasts in +/−25% range; negative value denotes phase reversal) in a rapid succession. Data were analyzed using reverse correlation. To examine the
1
RIKEN BSI 2 JST Presto
This work proposes a model of visual cortex using a multinomial Bayesian network and shows that this can reproduce important receptive field properties of V1 simple cells and V2. Multinomial variables are exploited for (1) modeling the nested feature representation in visual cortex by hypercolumns (feature spaces) composed of dozens of minicolumns (individual features), (2) enforcing sparsity of activation for efficient representations, and (3) reproducing the physiological phenomena called divisive normalization. In our simulation, a three-layer multinomial Bayesian network was constructed and trained by a sampling-based maximal-likelihood learning method for extracting features from natural image patches in a hierarchical manner. Then, the second-layer units had properties similar to V1 simple cells in the shape of receptive fields like Gabor filters, the distributions of spatial frequencies, bandwidths, and aspect ratios, and the response property of cross-orientation suppression. Moreover, the third-layer units were similar to V2 cells with receptive fields combining similar orientations and had the
e350
Abstracts / Neuroscience Research 71S (2011) e108–e415
distributions of orientation differences compatible with known experimental data by Anzai et al. (2004). Research fund: PRESTO. doi:10.1016/j.neures.2011.07.1532
P4-h20 Intra- and inter-visual area differences in optimal spatial frequency: A human fMRI study Tetsuya Yamamoto 1 , Hiroki Yamamoto 2 , Kenichiro Miura 3 , Nobukatsu Sawamoto 4 , Hidenao Fukuyama 4 , Kenji Kawano 3 1
Kokoro Res. Ctr., Kyoto Univ., Kyoto, Japan 2 Grad. Sch. of Human & Environ. Stud., Kyoto Univ., Kyoto, Japan 3 Dept. of Integrative Brain Sci., Grad. Sch. of Med., Kyoto Univ., Kyoto, Japan 4 Human Brain Res. Ctr., Grad. Sch. of Med., Kyoto Univ., Kyoto, Japan In functional magnetic resonance imaging studies, the phase-encoding method is a powerful technique to obtain the visual field map in the human visual cortex. In the present study, we applied this method to obtaining optimal spatial frequency maps in the human visual cortex. Two types of visual stimuli were used in the experiment: an achromatic luminance grating and an isoluminant chromatic grating. The spatial frequency of these stimuli increased continuously from 0.02 to 2 cycles/deg or decreased inversely six times in each run. We found robust activation synchronized with the stimulus cycle in most of the occipital visual cortex and parts of the posterior parietal and temporal cortex. In both maps for the luminance grating and chromatic grating, optimal spatial frequency decreased with increasing eccentricity in each visual area and with progression to higher-order visual areas. When both the maps were compared, optimal spatial frequency for the chromatic grating was generally lower than that for the luminance grating at corresponding cortical location. These results suggest that spatial frequency tuning depends on visual field eccentricity and the hierarchy of visual areas in humans. As the present optimal spatial frequency maps were clear even in higher-order visual cortex, they will be valuable for identifying individual visual areas in cortical regions having difficulty in obtaining visual field maps. Research fund: KAKENHI (21240037). doi:10.1016/j.neures.2011.07.1533
P4-h21 A theoretical study of orientation map reorganization induced by single-orientation exposure Masanobu Miyashita 1 , Nodoka Wakabayashi 1 , Shigeru Tanaka 2 1
Toyohashi University of Technology, Toyohashi, Japan 2 The University of Electro-Communications, Tokyo, Japan Recently, it has been found that the exposure of kittens to a single orientation through cylindrical-lens-fitted goggles for 1 or 2 weeks markedly altered orientation maps in the visual cortex. The over-representation of the exposed orientation was the most conspicuous in the case where the goggle rearing started around postnatal 30 days after normal rearing. The extent of orientation map alteration by single-orientation exposure diminished as the onset of goggle rearing was delayed. In this study, to find a possible mechanism of the orientation map alteration, we carried out computer simulations based on our model for the activity-dependent self-organization of geniculocortical inputs. First, we conducted simulation assuming that correlated activities are evoked spontaneously in the LGN before eye opening. The arrangement of preferred orientations in the simulated map was somewhat irregular and orientation selectivity was low. Next, using thus obtained orientation map as an initial map, we performed simulations under the random presentation of drifting oriented gratings in various orientations at an equal probability, which was assumed to correspond to a normal visual condition. The arrangement of preferred orientations became more regular and orientation selectivity was enhanced as the simulation step. Then, we resumed simulations under the presentation of only a vertically oriented grating moving leftward or rightward for a fixed period, which mimicked single-orientation exposure in cat experiments. The simulated maps showed a marked overrepresentation of vertical orientation, and the expansion of cortical domains for the exposed orientation decreased with the duration of the simulated normal visual experience. Thus, our model successfully reproduced the qualitative features of orientation map alteration in goggle-reared kittens as well as in normally reared kittens. Research fund: KAKENHI21500268. doi:10.1016/j.neures.2011.07.1534
P4-i01 Higher visual cortical responses mediated via the superior colliculus in mice Manavu Tohmi 1 , Reiko Meguro 2 , Ryuichi Hishida 1 , Masao Norita 2 , Katsuei Shibuki 1 1
Brain Res. Inst., Niigata Univ 2 Sch Med., Niigata Univ
Injury to the primary visual cortex (V1) produces so-called blindsight. There is a hypothesis that the mechanism of the blindsight derived from the superior colliculus (SC) pathway. The visual cortex of mice is composed of V1 and surrounding higher areas. We developed a differential imaging technique based on the differential properties of neurons in V1 and higher areas. When the speed of drifting grating patterns was changed from 10 to 50 degree/s in a step, flavoprotein fluorescence signals in area LM, AL, AM and RL were increased. In contrast, a step-like change from 50 to 10 degree/s produced an increase of signals in V1 and LM. To investigate the effects of visual experience on the visual responses in higher areas, mice were reared under strobe light, under which motion of objects could not be seen. While the preferred speeds of V1 was unchanged, that of the higher visual areas were reduced in the mice reared under strobe light. These experience-dependent changes were reversible even in adult mice. Histological studies demonstrated that the higher visual areas received thalamic inputs from the nuclei surrounding LGN. These thalamic nuclei do not receive direct inputs from the retina, and may be a part of the pulvinar that are known to receive visual inputs via SC. To confirm this possibility, effects of SC lesioning were investigated. While responses in V1 were unchanged, visual responses of the higher visual areas became similar to those of V1 after SC lesioning. These results strongly suggest that visual responses in higher areas are partly mediated via SC and may be required for blindsight. doi:10.1016/j.neures.2011.07.1535
P4-i02 Spatiotemporal property of event-related steadystate visual-evoked responses Takashi Shinozaki , Tsutomu Murata NICT, Kobe, Japan Steady-state visual-evoked potential/field (SSVEP/SSVEF) are oscillatory brain activity evoked by flickering visual stimuli modulated with a frequency. SSVEP/SSVEF have been used as a tag of perceptual state, and applied for many psychological studies: binocular rivalry (Tononi et al., 1998; Srinivasan et al., 1999), perception of ambiguous visual images (Parkkonen et al., 2008), attention (Muller et al., 1998; Ding et al., 2005; Kim et al., 2007). The basic property of SSVEP/SSVEF was studied mainly in the frequency characteristics (Hari & Salmelin, 1997; Herrmann, 2001). However, the temporal property, especially in the transition state, has not been clarified enough. In this study, we investigate the spatiotemporal property of SSVEF using magnetoencephalography (MEG). The visual stimulus is checkerboard pattern reversal flickering 15 Hz or 20 Hz. The event-related brain responses synchronized with the transition between frequencies were measured using whole head MEG system (148ch, BTi). The time course of SSVEF for each frequency band was calculated using continuous wavelet transform, and averaged among trials. Spatial pattern of SSVEF for each frequency is obtained from the time period specified by the results of wavelet analysis. The results showed that steady state responses required about 200 ms delay for the transition between frequencies. Interestingly, the spatial pattern of SSVEF was differed for each frequency including harmonics. These results clarify the temporal property of steady state visual evoked responses in detail, and may improve to specify the state of perceptions with more accurate temporal resolution in psychological studies. Recently, SSVEP/SSVEF are one of the most potent candidate of brain responses for Brain Machine Interface (BMI) (Martinez et al., 2007; Allison et al., 2008). Our results could be possible to contribute to improve the performance of BMI based on SSVEP/SSVEF. doi:10.1016/j.neures.2011.07.1536
P4-i03 Two distinct tecto-thalamo-cortical pathways to visual and auditory association cortex via the lateral posterior nucleus and the suprageniculate nucleus Masao Horie , Reiko Meguro Div. of Neurobiol. and Anat., Grad. Sch. of Med. and Dent., Niigata Univ., Niigata, Japan In the present study, we examined whether neurons of the superior colliculus (SC) make synaptic contacts with neurons of the thalamic lateral posterior nucleus (LP) as a visual relay nucleus and the suprageniculate nucleus (SG) as a multimodal relay nucleus projecting to the visual and visual-auditory
e349
gated into eye specific domains. Functional CG responses are evident during the early phase of innervation (P7), however paired pulse facilitation does not fully mature until the end of the second postnatal week. To test whether CG innervation relies on retinal input, we did enucleation and genetic deafferentation. In both cases, CG axons continue to target and innervate dLGN but do so in a highly accelerated manner. Thus, these results suggest that RG and CG innervation of dLGN occurs in a coordinated fashion and that retinal axons are involved in regulating the timing of CG innervation. Moreover preliminary experiments in other transgenic strains that allow for the visualization of projections that arise from cholinergic brainstem nuclei and thalamic reticular nucleus show a similar delay in dLGN innervation that is also disrupted by the loss of retinal input. Taken together, these results suggest that the development plan regulating nonretinal innervation of visual thalamus is highly conserved. Research fund: EY12716 (WG).
effect of static standing contrast and consequences of its fading, we added a pedestal contrast offset to the base sequence. The resulting stimuli had 7 contrasts within the 0–50% range and was without phase inversion. The orientation and spatial frequency of gratings were fixed at the optimal value of each cell. We found that V1 neurons showed a shift of contrast response function during pedestal conditions such that a physically blank screen (i.e. 0% contrast stimulus) generated clearly detectable responses. The amount of shift was reasonably predicted by the arithmetic mean of contrasts presented in each experimental condition. These results support the hypothesis that a negative image is generated during visual fading, and it behaves as if it were a physical entity. Additionally, we have analyzed time course of shifts of contrast response functions using a sliding analysis window over time. Time course of the shift was roughly comparable to that we observe perceptually. Research fund: Kakenhi 22135006, 22300110, and Global COE.
doi:10.1016/j.neures.2011.07.1528
doi:10.1016/j.neures.2011.07.1530
P4-h16 Object-selective responses evoked by phase-shift motion stimuli in macaque inferior temporal cortex: A case study using chronically implanted multielectrode arrays
P4-h18 Vertical and horizontal correlations of neuronal activities in the cat primary visual cortex
Ryusuke Hayashi 1,2 1
System Neuroscience Group, AIST, Tsukuba, Japan 2 PRESTO, JST, Kawaguchi, Japan Introduction: It has been shown that neurons in the inferior temporal cortex (IT) have shape selectivity that is relatively invariant with visual cue (Sary et al., 1993). Here, I revisited the motion invariance of shape selectivity in the IT cortex and investigated neural responses elicited by object images with phase-shift motion: images are perceived to move in one direction, while, unlike ordinary position-shift motion, their positions remain stationary. Methods: Multi-unit activities were recorded using two multielectrode arrays (32 electrodes each) chronically implanted in the right IT cortex of a macaque monkey. The animal was trained to do a fixation task during the recording: one of 25 object images, moving either leftward or rightward, was randomly chosen and presented for 2 s. The cycle of the phase shift was set either at 100, 133, 200, 267, or 400 ms. Results: The phase-shift operation alternated the polarity of image contrast every half cycle, which caused the oscillation in IT neurons’ responses when the cycle was longer than 200 ms. The oscillation, however, vanished when the cycle was shorter than 133 ms, i.e. the time equivalent as the mean onset latency of IT neurons. In the case of shorter cycle, the responses were activated only by the specific object images, regardless of motion direction. In addition to this motion invariance of shape selectivity, I observed that phase-shift motion elicited optokinetic nystagmus (OKN, involuntarily evoked oscillatory eye movement) in the monkey. Discussion: The author have recently developed a new binocular rivalry display that consists of two different object images with opponent phase-shift motions and showed that the perceptual alternation in human observer is accompanied with the switch of OKN. The present study suggests that the same binocular rivalry experiment is applicable to monkey to know its perceptual state from the evoked OKN and phase-shift motion of short cycle does not affect IT selectivity. Research fund: JST PRESTO program. doi:10.1016/j.neures.2011.07.1529
Hiroki Tanaka , Izumi Ohzawa Grad. Sch. of Front. Biosci., Osaka Univ., Toyonaka, Japan Primary visual cortex (V1) consists of vertically arranged functional columns, which are divided into 6 tangential layers. Cells in different layers and columns substantially interact by vertical and horizontal connections. In spite of many studies, our knowledge of these interactions is limited. Specifically, while correlation of neuronal activities is classified into temporally precise one (due to monosynaptic connection or strong common inputs) and loose one (due to polysynaptic connection or weak common inputs), how do 2-d structures of these correlations (relationship of correlation with laminar structures or vertical and horizontal distance between neurons) differ? To address these questions, we penetrated multi-array electrodes into cat V1 and recorded activities simultaneously from many neurons located in different columns and layers. We examined correlations of activities of isolated single neurons under visual stimulation or in spontaneous conditions. We found that the loose correlation extends widely across columns and layers. On this temporal scale, the vertically separated pairs showed consistent offset of correlated discharges that is linearly related with vertical distances of the pair (neurons in the deeper layers tend to fire prior to neurons in the superficial layers), but horizontally separated pairs did not show such offset. On the other hand, precise correlation was highly localized spatially and consistent offset was not observed. We also found that both type of correlations, especially the loose one, was stronger for superficial layers. These results suggest that the loose correlation reflects propagation of inputs from deep layers to upper layers along vertical connections and these inputs are widely synchronized across columns through horizontal connections. On the other hand, precise correlation reflects much shorter range of connections both in vertical and horizontal directions. Research fund: HT was supported by Kakenhi 21700433. IO was supported by Kakenhi 22135006, 22300110, and Global COE. doi:10.1016/j.neures.2011.07.1531
P4-h19 Multinomial Bayesian network model reproducing receptive field properties of V1 simple cells and V2 Haruo Hosoya 1,2
P4-h17 Neural correlates of fading illusion revealed in responses of V1 neurons Kohei Kurihara , Kota S. Sasaki, Izumi Ohzawa Grad. Sch. of Frontier Biosciences, Osaka Univ., Osaka, Japan When a static image without sharp edges is fixated somewhat peripherally, the image is perceived to fade away gradually. This phenomenon is known as Troxler fading or fading illusion. Traditionally, at least an aspect of this illusion is explained by a hypothesis that physical stimulus is cancelled by a negative image generated in the visual pathway, because the negative image is perceived momentarily when the physical stimulus is suddenly removed. One prediction that may be tested neurophysiologically is that, in the faded or adapted state, visual neurons actually respond to a blank screen, because the negative image is there even when no physical stimulus is present. To test this prediction in cat V1, we measured the base contrast response function of neurons by presenting flash grating stimuli of various contrasts (7 contrasts in +/−25% range; negative value denotes phase reversal) in a rapid succession. Data were analyzed using reverse correlation. To examine the
1
RIKEN BSI 2 JST Presto
This work proposes a model of visual cortex using a multinomial Bayesian network and shows that this can reproduce important receptive field properties of V1 simple cells and V2. Multinomial variables are exploited for (1) modeling the nested feature representation in visual cortex by hypercolumns (feature spaces) composed of dozens of minicolumns (individual features), (2) enforcing sparsity of activation for efficient representations, and (3) reproducing the physiological phenomena called divisive normalization. In our simulation, a three-layer multinomial Bayesian network was constructed and trained by a sampling-based maximal-likelihood learning method for extracting features from natural image patches in a hierarchical manner. Then, the second-layer units had properties similar to V1 simple cells in the shape of receptive fields like Gabor filters, the distributions of spatial frequencies, bandwidths, and aspect ratios, and the response property of cross-orientation suppression. Moreover, the third-layer units were similar to V2 cells with receptive fields combining similar orientations and had the
e350
Abstracts / Neuroscience Research 71S (2011) e108–e415
distributions of orientation differences compatible with known experimental data by Anzai et al. (2004). Research fund: PRESTO. doi:10.1016/j.neures.2011.07.1532
P4-h20 Intra- and inter-visual area differences in optimal spatial frequency: A human fMRI study Tetsuya Yamamoto 1 , Hiroki Yamamoto 2 , Kenichiro Miura 3 , Nobukatsu Sawamoto 4 , Hidenao Fukuyama 4 , Kenji Kawano 3 1
Kokoro Res. Ctr., Kyoto Univ., Kyoto, Japan 2 Grad. Sch. of Human & Environ. Stud., Kyoto Univ., Kyoto, Japan 3 Dept. of Integrative Brain Sci., Grad. Sch. of Med., Kyoto Univ., Kyoto, Japan 4 Human Brain Res. Ctr., Grad. Sch. of Med., Kyoto Univ., Kyoto, Japan In functional magnetic resonance imaging studies, the phase-encoding method is a powerful technique to obtain the visual field map in the human visual cortex. In the present study, we applied this method to obtaining optimal spatial frequency maps in the human visual cortex. Two types of visual stimuli were used in the experiment: an achromatic luminance grating and an isoluminant chromatic grating. The spatial frequency of these stimuli increased continuously from 0.02 to 2 cycles/deg or decreased inversely six times in each run. We found robust activation synchronized with the stimulus cycle in most of the occipital visual cortex and parts of the posterior parietal and temporal cortex. In both maps for the luminance grating and chromatic grating, optimal spatial frequency decreased with increasing eccentricity in each visual area and with progression to higher-order visual areas. When both the maps were compared, optimal spatial frequency for the chromatic grating was generally lower than that for the luminance grating at corresponding cortical location. These results suggest that spatial frequency tuning depends on visual field eccentricity and the hierarchy of visual areas in humans. As the present optimal spatial frequency maps were clear even in higher-order visual cortex, they will be valuable for identifying individual visual areas in cortical regions having difficulty in obtaining visual field maps. Research fund: KAKENHI (21240037). doi:10.1016/j.neures.2011.07.1533
P4-h21 A theoretical study of orientation map reorganization induced by single-orientation exposure Masanobu Miyashita 1 , Nodoka Wakabayashi 1 , Shigeru Tanaka 2 1
Toyohashi University of Technology, Toyohashi, Japan 2 The University of Electro-Communications, Tokyo, Japan Recently, it has been found that the exposure of kittens to a single orientation through cylindrical-lens-fitted goggles for 1 or 2 weeks markedly altered orientation maps in the visual cortex. The over-representation of the exposed orientation was the most conspicuous in the case where the goggle rearing started around postnatal 30 days after normal rearing. The extent of orientation map alteration by single-orientation exposure diminished as the onset of goggle rearing was delayed. In this study, to find a possible mechanism of the orientation map alteration, we carried out computer simulations based on our model for the activity-dependent self-organization of geniculocortical inputs. First, we conducted simulation assuming that correlated activities are evoked spontaneously in the LGN before eye opening. The arrangement of preferred orientations in the simulated map was somewhat irregular and orientation selectivity was low. Next, using thus obtained orientation map as an initial map, we performed simulations under the random presentation of drifting oriented gratings in various orientations at an equal probability, which was assumed to correspond to a normal visual condition. The arrangement of preferred orientations became more regular and orientation selectivity was enhanced as the simulation step. Then, we resumed simulations under the presentation of only a vertically oriented grating moving leftward or rightward for a fixed period, which mimicked single-orientation exposure in cat experiments. The simulated maps showed a marked overrepresentation of vertical orientation, and the expansion of cortical domains for the exposed orientation decreased with the duration of the simulated normal visual experience. Thus, our model successfully reproduced the qualitative features of orientation map alteration in goggle-reared kittens as well as in normally reared kittens. Research fund: KAKENHI21500268. doi:10.1016/j.neures.2011.07.1534
P4-i01 Higher visual cortical responses mediated via the superior colliculus in mice Manavu Tohmi 1 , Reiko Meguro 2 , Ryuichi Hishida 1 , Masao Norita 2 , Katsuei Shibuki 1 1
Brain Res. Inst., Niigata Univ 2 Sch Med., Niigata Univ
Injury to the primary visual cortex (V1) produces so-called blindsight. There is a hypothesis that the mechanism of the blindsight derived from the superior colliculus (SC) pathway. The visual cortex of mice is composed of V1 and surrounding higher areas. We developed a differential imaging technique based on the differential properties of neurons in V1 and higher areas. When the speed of drifting grating patterns was changed from 10 to 50 degree/s in a step, flavoprotein fluorescence signals in area LM, AL, AM and RL were increased. In contrast, a step-like change from 50 to 10 degree/s produced an increase of signals in V1 and LM. To investigate the effects of visual experience on the visual responses in higher areas, mice were reared under strobe light, under which motion of objects could not be seen. While the preferred speeds of V1 was unchanged, that of the higher visual areas were reduced in the mice reared under strobe light. These experience-dependent changes were reversible even in adult mice. Histological studies demonstrated that the higher visual areas received thalamic inputs from the nuclei surrounding LGN. These thalamic nuclei do not receive direct inputs from the retina, and may be a part of the pulvinar that are known to receive visual inputs via SC. To confirm this possibility, effects of SC lesioning were investigated. While responses in V1 were unchanged, visual responses of the higher visual areas became similar to those of V1 after SC lesioning. These results strongly suggest that visual responses in higher areas are partly mediated via SC and may be required for blindsight. doi:10.1016/j.neures.2011.07.1535
P4-i02 Spatiotemporal property of event-related steadystate visual-evoked responses Takashi Shinozaki , Tsutomu Murata NICT, Kobe, Japan Steady-state visual-evoked potential/field (SSVEP/SSVEF) are oscillatory brain activity evoked by flickering visual stimuli modulated with a frequency. SSVEP/SSVEF have been used as a tag of perceptual state, and applied for many psychological studies: binocular rivalry (Tononi et al., 1998; Srinivasan et al., 1999), perception of ambiguous visual images (Parkkonen et al., 2008), attention (Muller et al., 1998; Ding et al., 2005; Kim et al., 2007). The basic property of SSVEP/SSVEF was studied mainly in the frequency characteristics (Hari & Salmelin, 1997; Herrmann, 2001). However, the temporal property, especially in the transition state, has not been clarified enough. In this study, we investigate the spatiotemporal property of SSVEF using magnetoencephalography (MEG). The visual stimulus is checkerboard pattern reversal flickering 15 Hz or 20 Hz. The event-related brain responses synchronized with the transition between frequencies were measured using whole head MEG system (148ch, BTi). The time course of SSVEF for each frequency band was calculated using continuous wavelet transform, and averaged among trials. Spatial pattern of SSVEF for each frequency is obtained from the time period specified by the results of wavelet analysis. The results showed that steady state responses required about 200 ms delay for the transition between frequencies. Interestingly, the spatial pattern of SSVEF was differed for each frequency including harmonics. These results clarify the temporal property of steady state visual evoked responses in detail, and may improve to specify the state of perceptions with more accurate temporal resolution in psychological studies. Recently, SSVEP/SSVEF are one of the most potent candidate of brain responses for Brain Machine Interface (BMI) (Martinez et al., 2007; Allison et al., 2008). Our results could be possible to contribute to improve the performance of BMI based on SSVEP/SSVEF. doi:10.1016/j.neures.2011.07.1536
P4-i03 Two distinct tecto-thalamo-cortical pathways to visual and auditory association cortex via the lateral posterior nucleus and the suprageniculate nucleus Masao Horie , Reiko Meguro Div. of Neurobiol. and Anat., Grad. Sch. of Med. and Dent., Niigata Univ., Niigata, Japan In the present study, we examined whether neurons of the superior colliculus (SC) make synaptic contacts with neurons of the thalamic lateral posterior nucleus (LP) as a visual relay nucleus and the suprageniculate nucleus (SG) as a multimodal relay nucleus projecting to the visual and visual-auditory