A congnitive shape decomposition method that agrees with human intuition is proposed for the conceptual recognition from sillouettes of objects. Descriptions specifying the structure of shape in terms of meaningful parts and relations have cognitive power and anthropomorphism. In general, man-made objects have a lot of collinear lines and regularity. For the cognitive decomposition of man-made objects, many heuristic rules based on the cognitive experimentation are applied on the context of collinerarity and regularity. The cognitive shape decomposition for the natural shape is carried out by analyzing the possible configuraitions of vertices and line segments for one concave vertex. A cost function for the configuation is designed by weighted sum of five criteria such as, the length of split line segment, the number of split line segments at concave vertex, the proximity of concave vertex, and the correspondence of vertices. These criteria are vased on the property of human perception such as proximtiy, symmetry, and simplicity. The most promising vertex os selected among three set of visible vertices by evaluating the cost function. A number of experiments conducted on the different types of shapes shows that the results correspond with human intuition.