Recent investigations on perceptual learning were surveyed from the operational point of view in perceptual experiment. It was suggested that there were three types of approach: (1) We can assume, as the first type of approach, a certain possibility of measuring directly the factors of the perceptual learning in the experiments if a discrimination in an experiment goes through the following three steps, viz., to know environment, to make judgmental base, and discrimination. Besides, it may safely be said that the approach thus assumed, either the first step or the second being decided under a specific condition, and equal to other conditions, enables us to analyse the relationship between learning and perception, and equal to other conditions, enables us to analyse the relationship between learning and perception by means of the quantitative difference of discrimination under the each condition. (2) The second type of approach is the experiment trying to show how influences of past perceptual learning are working on the present discrimination. As to the experimental operation, it was indistinct about what points in discriminational steps are influenced, though what is the past experience is clear. Therefore this type of approach becomes a indirect inference by the experiment. (3) A general discussion or theory and some kinds of experiments concerning the whole connections of repeating perceptual learning are included in the third type of approach. Those experiments are on the probability involved in the perceptual events, training effect, and repeating experiments for the same subject. The first type of experiment wears an aspect of the operational certainty and seems to be reliable. As the second type, on the other hand, the experimental conditions are not in a perfect controlling and there lies something leap in arguments. The second one, however, has very interesting ideas, and it is certain that they would offer many suggestive and advantageous problems. Therefore, it is naturally required to make it sure experimentally one by one, by the same way with the first type of experiment limiting the conditions operationally. If so, not only the ideas will be supported, but many other new facts should occur as well. In the experimental approach, cited as the third type, we find no fault with experiments themselves, though it is not suitable to discuss about the long repeated general learning processes on the basis of those experimental results. For that purpose it is thought to be necessary to consult with the results from another kinds of perceptual experiments as well as with the fruits in the field of learning. The learning experiments show how the longitudinal learning processes are modifying. The perceptual experiments make the states of the lateral cross sections of the learning processes clear. We can get many informations about perceptual learning from the both fields, learning and perception. A perceptual learning theory in an objective psychology can not be established without supporting of many experimental data from the both fields.