In this paper, I tried to investigate recent psychological experiments on sensory generalization in animals. 1) A number of investigators have dealt with sensory generalization in the human psychology, particularly in the respondent conditioning situation, and some concave generalization gradients were obtained, when plotted on jnd scale. But a recent experiment on operant conditioning in pigeons for spectral stimuli has pointed out that there is no meaningfull relation between the shape of generalization gradient and discriminability (size of jnd). In my experiment similar to it, however, it was found that generarization gradients transform following changes of discriminability along the spectral continuum. There seems to be the distance equivalence which is describable as neither physical nor simple jnd distance from conditioned stimulus. 2) In recent experiments about the post discrimination gradient, it was discovered that a peak shift of generalization gradient appears away from the negative stimulus in the direction of the positive. This suggests that there are some inductions from the positive stimnlus to the negative and vice versa. More comparison of discrimination training situation with single stimulus conditioning is necessary. 3) In discrimination training experiments, it was also found that reinforcement schedules with positive and negative stimulus bring out different effects on the generalization gradient. Experimental data were not well described by the summation theory and the non summation theory provided a more adequate description of them. The need for further studies on the favorable condition to the non summation hypothesis is pressing.