Abstracts / Neuroscience Research 58S (2007) S1–S244
P2-f34 Prediction of multi-dimensional decision-making based
S161
P2-f39 A brain–machine interface for predicting both arm-
on the single-trial activities of a population of prefrontal neurons
reaching movements and postures
Narihisa Matsumoto 1 , Ryohei P. Hasegawa 1,2 1 Neuroscience Research Institute, AIST, Tsukuba, Japan; 2 National Eye Institute, NIH, Bethesda, USA
Hideaki Hirose 1,4 , Kyuwan Choi 2 , Ken-Ichiro Tsutsui 1 , Yoshio Sakurai 3,4 , Yasuharu Koike 2,4 , Toshio Iijima 1,4 1 Graduate School of Life Sciences, Tohoku University, Sendai, Japan; 2 Precision and Intelligence Lab, Tokyo Institute of Technology, Yokohama, Japan; 3 Department of Psychology, Graduate School of Letters, Kyoto University, Kyoto, Japan; 4 JST CREST, Saitama, Japan
To utilize the brain signals related to complex decisions for multidimensional sensorimotor responses, we have developed an algorithm that produces a post hoc prediction of trial types based on the singletrial activity of 323 prefrontal neurons which were recorded individually while monkeys remembered one of six cue locations either to make or not to make a saccade. We used a template matching method to classify the activity into 12 trial types (6 locations by 2 responses) in the following two ways. The single-dimensional classification of 12 trial types at once made correct predictions on about 80% of trials during the late cue period. The multi-dimensional classification, which combined separate predictions of cue locations during the early cue period and behavioral responses during the late delay period, made correct predictions on more than 90% of trials. These results suggest that the multi-dimensional classification might have advantages in developing a better cognitive brain-machine interface. Research funds: KAKENHI (17700167,NM; 18019048,RPH); NEI (RPH)
In the present study, we developed a brain–machine interface (BMI) system that can predict both arm-reaching movements and target-holding postures from neuronal activity in the primary motor area (MI) when a monkey is performing a motor task. Low-pass filtered electromyography (fEMG) signals were computed from the neuronal activity by a linear regression model. Arm-movement trajectories were calculated from the fEMG signals by combination of an artificial neural network and servo-mechanism. The estimated fEMG signals and trajectories were well correlated with actually recorded EMGs and trajectories during movements and posture maintenances. Positions not only of the arm tip (wrist joint) but also the middle joint (elbow joint) were accurately estimated. These results suggest that our BMI system could predict both the dynamics and kinematics of neuroprosthetic devices simultaneously. Research funds: JST CREST Program
P2-f35 Development of an electrodes array for a homogeneous sampling of orientation tuned cells in the visual cortex Yoshiko Maruyama, Hiroyuki Ito Department of Information & Communication Sciences, Faculty of Engineering, Kyoto Sangyo University, Kyoto, Japan We recorded multiple single units simultaneously from the primary visual cortex of an anesthetized cat. We attempted to predict an orientation of the light bar stimulus based on the multineuron data of a single trial. For a better prediction, we need homogeneous samplings of differently orientation tuned cells. However, orientation selectivities were rather heterogeneous in our previous data recorded by 4-tetrodes array. To solve the problem, we developed a 8-single electrodes array. Since new array’s channels were more densely distributed, we expected that orientation selectivities were more homogeneous. By the circular variance, we compared the distribution of unit’s orientation selectivities between 8-single electrodes array and 4-tetrodes array. Some recordings suggested that 8-single electrodes array can record units with a larger variance.
P2-f36 Modeling of extracellular multiple neuronal activities
Role of nitric oxide synthase-containing hypothalamic P2-f40 paraventricular-spinal neurons in the CRF-induced activation of the central sympatho-adrenal outflow in rats Daisuke Usui 1,2 , Naoko Yamaguchi 1 , Shoshiro Okada 1 , Takahiro Shimizu 1 , Hiroshi Wakiguchi 2 , Kunihiko Yokotani 1 1 Department of Pharmacology, Graduate School of Medicine, Kochi University, Kochi, Japan; 2 Department of Pediatrics, Graduate School of Medicine, Kochi University, Kochi, Japan Centrally administered corticotropin-releasing factor (CRF) has been shown to activate sympatho-adrenal outflow in rats, but the central mechanisms remain unclear. In this study, we examined the role of paraventricular hypothalamic nucleus (PVN) neurons, known as the autonomic center, activated by in centrally administered CRF in rats, with regard to nitric oxide synthase (NOS). After injection of retrograde tracer into the spinal cord, retrogradely labeled cells were detected in the dorsal cap, ventral and posterior part of the PVN. After administration of CRF, neurons co-expressing of inducible NOS (iNOS) and c-Fos were observed in these retrogradely labeled cells. The CRF had little effect on neuronal NOS. These data suggest that PVN-spinal neurons expressing iNOS play a role in the centrally administered CRF-induced activation of the sympatho-adrenal outflow in rats.
P2-f41 Neurokinin 1 receptor plastically changes excitability of
based on the multi-unit recording
neurons in reticular formation of rat medulla oblongata
Takashi Kubo 1 , Norihiro Katayama 1 , Hajime Mushiake 2 , Osamu Kikuchi 1 , Akihiro Karashima 1 , Mitsuyuki Nakao 1 1 Department of Applied Information Sciences, Graduate School of Information Sciences, Tohoku University, Sendai, Japan; 2 Department of Physiology, Tohoku University School of Medicine, Japan
Naoyuki Himi 1 , Tomoshige Koga 2 , Katsuhiko Tsujioka 1 1 Department of Physiology, Kawasaki Medical School, Kurashiki, Japan; 2 Department of Rehabilitation, Kawasaki University Med. Welfare, Kurashiki, Japan
To obtain good measurement of extracellular multi-unit activities, adjustment of electrode parameters such as the size and spacing of electrodes is essential. However, these parameters have been decided according to the researches’ experience. To optimize the electrode design, computer simulation system, we constructed a model of multi-unit recording in a three-dimensional neural tissue using multi-site electrodes. Computer simulations show that this model faithfully reproduces waveforms of multi-unit activities, amplitude distribution of extracellular potential and power spectral densities of the signals. These results suggest this system is useful for designing multi-site recording electrode.
Our previous experiments using rats suggested that several autonomic responses accompanying with emesis were controlled by neurons in reticular formation near the nucleus ambiguus (NA) of medulla oblongata, and induced via neurokinin 1 receptor (NK1R) activation. In this study, the plastic changes of activities in the neurons near NA were investigated. Whole-cell recording was applied to the neurons in coronal slice preparations of brainstem isolated from infant rat (6–12 days). In the several neurons, tetanus stimulation (100 Hz, 2 sec.) induced the augmentation of amplitude of EPSCs that was maintained for a few minutes. However, these augmentations were canceled by perfusion of NK1R antagonist, sendide (1 M). These results suggest that the excitability of neurons plastically changes via NK1R activation and hence participates in generating long lasting autonomic responses accompanying with emesis.