Glass looks transparent and cold. Like this example, material perception has multimodal properties, and when we see an object we generally recognize its visual as well as tactile properties. Reaching arm to visible objects and touching them are a natural behavioral repertoire to acquire or to confirm multimodal impression in material perception. Therefore, by analyzing touching behaviors to real objects, we should be able to obtain important clues on how the perceptual space of various materials is organized in the brain. In this experiment, monkeys were trained with a simple behavioral task in which the monkey was required to reach and grasp a rod to obtain a reward. The monkeys were trained in a condition where a resin rods with a color (white, grey or black) or a pattern (polka dots or grating; 3 types for each) was presented until they successfully performed in >90% trials. Then, the monkeys were tested in a condition in which a rod chosen from 36 samples (9 material categories; metal, glass, ceramic, stone, wood, bark, leather, fabric and fur; 4 samples for each) was presented. We found that the monkeys' performance depended on material categories; the monkeys correctly grasped some materials but failed to grasp other materials. Especially when fur objects were presented, the monkeys avoided to touch them. Although both metal and fur should be familiar materials for monkeys in laboratory environment, the monkeys' responses to these two categories of materials were quite distinct. These results suggest that material perception is organized by visuo-tactile integration taking into account the biological significance of materials, and thereby visual perception of material can influence decisions about whether to touch it or not.