The goal of this research is to investigate the sensation produced with ultimate dynamic pin-matrix type tactile display. It has been a lot of research for tactile display which has many pin-rods moving only in the vertical direction. These systems are typically divided into two types; one type is designed for presenting static shape (such as geometrical or geographical features, characters or icons on the computer screen, etc.) and another type is for displaying patterns changing dynamically. The advantage of passive type is that we can explore the presented texture at our own paces. On the other hand, an advantage of dynamic type is that it can provide richer information than static displays with small presenting area; that is because we need only put our fingers on the top of the display and feel tactile patterns changing automatically. We think this advantage will be important for applying tactile display to usual computer interface or cell phone. Although some previous studies proposed dynamic tactile display with various actuators, these prototypes had their disadvantages such as low response time or low density of pin matrix, and so on. Therefore, it is thought that these tactile displays are still in the early stage. However, we haven't well understood what kind of sensation can be presented with dynamic pin-matrix type tactile display yet. We think that it is not necessarily appropriate approach to evaluate this type of displays after developing them. We believe that we need a novel methodology to design or evaluate dynamic pin-matrix tactile displays before creating them. To investigate the tactile sensation with a high-density Pin-Matrix (PM) type tactile interface, we generated dynamic movement of a non-actuated prototype by moving it over textures etched onto a flat surface. With this method, a human works as a part of actuator, therefore it is unnecessary to add actuator for each pin-rod. Figure 1 shows three types of pin-matrix we created; PM1 has the smallest interval 1.0mm from center to center of the pin-rod, PM2 has 1.5mm pin interval which is approximately as same as the two-point threshold on the fingertip, and PM3 has 2.0mm interval. We made supporting plates with thin bakelite board (thickness: 1.0mm), which has 0.9mm holes for penetrating pin-rods. The bottom of the pin-rod has 0.4mm hemisphere shape, which enables to reduce the effect of the friction dramatically and to go over the texture without tumbling. With this specification, PM can move freely corresponding to the texture surface. We also took the length of spaces placed between two bakelite boards into account in order to keep the movement direction of pin-rods only in the vertical. To investigate how small interval is enough for human to recognize the shapes engraved on the texture, we conducted a psychophysical experiment with touching textures using these PMs. The result showed that the smaller the pin interval becomes, the easier subjects understood what shape was engraved on the textures. In another experiment, we asked subject that the engraved texture is convex or concave. Subjects could easily tell convex from concave, therefore this result might suggest that not only the tangential force but also normal force on finger tip is used for shape recognition. And we also found a kind of edge enhancement effect in using PM, especially in small pin interval. We think this effect might result from an aliasing effect as mentioned in previous study [Kikuuwe et al. 2005], which is caused by discrete sampling of the target texture. We think that our devised method can be broadly used for deciding parameters of pin-matrix type tactile displays for particular application. In our preliminary result we confirmed that subject could easily recognize the shape with 1.0mm interval pin-matrix. Although we need to perform further experiment with smaller pin-rod or smaller pin interval, we can say that current, dynamic pin-matrix type tactile display is not sufficient enough for displaying shapes to finger tip. With the same methodology we proposed, it will be possible to find which pin interval is the most appropriate for a dynamic Braille or a computer interface.