In order to overcome prematurity of ant colony algorithm, the conception of belief space originated in cultural algorithm is introduced, and a new cultural ant algorithm is proposed for continuous optimization problems. Firstly, the coding scheme for ant colony algorithm to solve continuous optimization problems is discussed. Then belief space is brought in, and designed as the form of two parts: individual belief space and population belief space. The former is used to conduct individuals' deep search for better solutions, and the other to help worse individuals drop their current bad solution space for broad search. The update rules of both population space and belief space are given subsequently. Eight common standard functions are used to test the new algorithm, which is compared with four other algorithms at the same time. The results show effectiveness and superiority of the new algorithm. Finally the effect of the parameter used in the algorithm is discussed, and so does the both two belief space.