We demonstrate here the experimental results indicating the existence of interdroplet interactions between droplets even in the late stage of droplet spinodal decomposition. On the basis of these results, we discuss a few possible coarsening mechanisms of droplet spinodal decomposition. A droplet of finite curvature has long-range concentration fields around it to satisfy the boundary condition given by the Gibbs-Thomson relation. For a pair of neighboring droplets, thus, the closed isoconcentration lines including the two droplets are generally formed in the matrix phase in the late stage of spinodal decomposition. Thus, the droplet can feel its neighboring droplet via the concentration gradient produced by the latter. The excess bulk and gradient energy associated with the diffuse concentration changes around droplets may cause a lon,g-range attractive interaction between the droplets, which induces the interdroplet collisions. We call this mechanism "gradient-induced-coupling mechanism." The gradient force associated with an interface is likely responsible for the motion of droplets under the incompressibility condition. In fluid mixtures, this coupling of concentration fields can further be enhanced by the violation of the local equilibrium due to the quick hydrodynamic coalescence of colliding droplets. In such a case, there is a possibility that the concentration diffusion cannot follow this quick geometrical coarsening. This interface quench effect probably leads to the phenomenon of "collision-induced collision via diffusion" : A droplet experiencing collision, which accompanies strong diffusion fields, has a higher probability of the subsequent collisions. The hydrodynamic flow induced by droplet coalescence also leads to the subsequent collisions, which we call "collision-induced collision via flow." Further, the geometrical configuration of droplets also leads to "geometrical collision-induced collision." All the above mechanisms likely make the coarsening of droplet spinodal decomposition faster than the prediction of Brownian-coagulation mechanism.