Event Abstract Back to Event Estimation of the information pathway in the neural network in the premotor center of Bombyx mori to generate the flip-flop activity and its validation by the simulation Ikuko Nishikawa1*, Akira Takashima2, Shigehiro Namiki2, Tomoki Kazawa2, Stephan Shuichi Haupt2, Hidetoshi Ikeno3 and Ryohei Kanzaki2 1 Ritsumeikan University, Human and Computer Intelligence, Japan 2 The University of Tokyo, Japan 3 University of Hyogo, Japan The lateral accessory lobe (LAL) and the ventral protocerebrum (VPC) are known to form a premotor center of brain in insects. The present report focuses on the neural network of the LAL-VPC of Bombyx mori. The neuroethological experiments elucidate LAL-VPC generates a motor command of the pheromone oriented behavior, which is composed of a sequential activity of surge, zigzag-turn and loop [1,2]. The LAL-VPC region has been characterized by its temporal activity with alternating flip-flop patterns between both hemispheres, through the records from the descending neurons. Moreover, electrophysiological and immunohistochemical studies enable the physiological and morphological classification of LAL-VPC neurons, and thereby suggest possible anatomical connections among the neurons.The goal of this study is to estimate the functional connection strengths of the above anatomically found synaptic connections to elucidate the main information pathway for generating the flip-flop activity in LAL-VPC. The available data to challenge this problem is the stimulus-evoked activities of LAL-VPC neurons. Another key is the experimental finding that LAL-VPC consists of multiple neuropile regions [3], and each neuron possesses input branches from one or more regions and output branches to other regions. Based on these experimental findings, first, we define the following self-consistent formula to derive the connection strengths. That is, the activity of each neuropile region is given by the weighted summation of the outputs from the connecting neurons, and at the same time, the activity of each neuron is given by the weighted summation of the inputs from the connecting regions. Then, the formula is solved as an optimization problem, whose evaluation function is the reproducibility of the observed neuron activities, and the constraints are the observed anatomical connections. Finally, the connection strengths obtained as optimal solutions of the above static model are evaluated by a network simulation, in which each neuron has a spiking dynamics.A simulator of LAL-VPC network is described by Integrate-and-Fire model for each neuron and two-state kinetic model for each synapse. All of experimentally observed LAL-VPC neurons are connected by the strengths obtained by the above method. Most bilateral neurons are considered as inhibitory, and all local interneurons are set excitatory for the simplicity. Poisson noise is given to each neuron for the spontaneous activity, and the post inhibitory rebound firing is considered. The obtained connection is evaluated by how the temporal activity of each neuron reproduces the physiological data, especially the alternating flip-flop in both hemispheres.
Read full abstract