In this paper, we propose a particle swarm optimization(abbr. PSO) based on chaotic spike oscillator dynamics(abbr. CSOPSO). Our method has ability to search optima without stochastic elements. Since the basic particle dynamics exhibits chaotic behavior on phase space consisting of the velocity and position, particles on the search space move with chaotic motion. Size of the chaotic attractor corresponding to search range of position can be controlled by single parameter. We focus on influence between size of the attractor and searching ability. The effectivity of CSOPSO by comparing with a previous PSO by some benchmark problems is considered.