A neural network model for arm trajectory control

$29

8. Behaviour A NEURAL NETWORK MODEL FOR ARM TRAJECTORY CONTROL. MITSUO KAWATO 1, TQHRU SETOYAMA* AND RYOJI SUZUKI ~, Department of Biophysical Engineering, Faculty of Engineering Science, Osaka University, Toyonaka, Machikaneyama 1-1, Osaka 560, Japan. We propose a neural network model and feedback-error-learning rule by which an internal neural model of inverse dynamics of the motor system is acquired in a three-layer neural network by heterosynaptic plasticity using a feedback motor command as an error signal. It contains the transcortical loop (feedback loop) and the cerebrocerebellum and parvocellular part of the red nucleus (inverse-dynamics model) and the motor cortex (feedback command calculation and summation of it with feedforward command). We used a modification of the back-propagation learning rule while still using the feedback motor command as the error signal. The performance of control of a robotic manipulator improved considerably during learning. As motor learning proceeded, the inverse-dynamics model gradually took the place of the feedback loop as main controller. Furthermore, the neural network can control a quite new and faster movement pattern after learning a slow movement pattern. lCognitive Processes Department, ATR Auditory and Visual Perception Research Laboratories ~Department of Mathematical Engineering and Information Physics, Faculty of Engineering, University of Tokyo

ANTICIPATORY NEURONAL ACTIVITY IN THE LIMBIC SYSTEM DURING MALE RAT COPULATORY BEHAVIOR. TSUYOSHI SHIMURA and MINORU SHIMOKOCHI, Department of Behavioral Physiology, Faculty of Human Sciences, Osaka University , I-2 Yamadaoka, Suita, O s a k a 565, J a p a n . Behavioral studies suggest that experience is an important factor for the normal expression of male copulatory behavior. We recorded successive changes in neuronal activity in the limbic system of male rats during conditioning. Chronic recording electrodes had been previously implanted in the male rat under sodium pentobarbital anesthesia. The observation cage was divided into two compartments by a transparent shutter. An experimental male and an estrogen-progesterone treated female were separated in the compartments. The male was trained to a p p r o a c h t h e s h u t t e r i n r e s p o n s e t o a c u e t o n e (8kHz) t o c o p u l a t e w i t h t h e f e m a l e . About three training sessions were needed for males to approach on the cue tone. Two different unitary firing patterns were observed during pursuit-mounting behavior in the medial preoptic area; one increased continuously during pursuit of a female and the other increased only during thrusting. The conditional change occurred in the former. When the approach behavior was ful!y conditioned, the unit fired about three times more than the pre-tone base line level during the cue tone. The increased firing was not associated with any locomotion. It might reflect increased sexual arousal of the male.

HAND

USE

PREFERENCE

OF

JAPANESE

MACAQUE

MONKEYS

RAISED

IN

AN

OPEN

COMPOUND.

KISOU

KUBOTA~

Department o f Neurophysiology, Primate Research Institute~ Kyoto University, Inuyama City, Aichi, 484 r Japan. Preferred hand use was examined in a group of Japanese monkeys (Arashiyama R Troop) which were raised in a 20x2Om open-field compounds (area, 20 m x 20 m) of our institute. The 17 male and 27 female monkeys ranged in age from 1 to 29. They were provisioned every morning by monkey pellets (Oriental Yeast, PS-type; white, cylindrical 1.9 cm long, and 1.2 cm diam. and 0.5 gr weight) scattered over 8 m x 4 m area. Feeding behaviors were video-taped for about 25 h, beginning April i, 1988 for 3 months. Hand use behaviors were selected in which the pellet was pinched between thumb and forefinger by either left or right hand and brought to the mouth, not necessarily by the same hand, for at least 4 times successively in a sitting posture with both feet and ischial callosity touching the ground. For each monkey at least 36 pick-up behaviors were selected and hand side was determined statistically (x2 test). Of the 44 monkeys, 19 showed a left hand preference, 5 a right hand preference and 20 no preferences. In young monkeys (< 7 years of age, n=29) 9 showed a left, 2 a right and 18 no preferences, monkeys with right or no preferences being monkeys significantly less. In adult monkeys () 7 years old, n=15) I0 showed a left, 3 a right and 2 no preferences, monkeys with left hand preferece being significantly more numerous than monkeys with no preference. There were no significant correlations between hand preferences of ii mothers and those of their children (n=34). There were no differences of hand preference distribution between males and females.