As part of her technique of cognitive research on experimental animals, Michiko Saito conducted research in which samples were shown to dogs. As far as the cognitive ability of dogs is concerned, it had been thought that their eyesight rated about 0.2 to 0.3, and that they were colorblind. But through further researches, it has been found that they can distinguish among three color panels, and also that distinguish among the colors of a signal apparatus, are capable of recognizing number values up to 5.

As one step in her research into the cognitive capacity of dogs, Ms. Saito took up the issue of showing samples of basic shapes dogs to determine whether they could discriminate among them. A specialized "Skinner box" was prepared with three display windows and response apertures. The device was set to feed automatically when a correct answer was generated, and the apparatus was equipped to operate with a computer and sensor coordinated for operant conditioning.

The shapes (stimuli) used were the circle, square and triangle. Initially, as experimental pattern A, one shape was shown in the center display window. Next, the same shape was shown in either the left or right window. In pattern B, when the same shape was displayed in either the left or right window, a different shape was shown in one of the remaining windows. For pattern C, a 'x-sign' shape was displayed.

These three experimental patterns were conducted on three canine subjects to determine the degree to which they are capable of distinguishing among shapes. The results were that correct responses were attained in 88% of pattern A cases, while only 48.8% of pattern B responses were correct.

"Despite the fact that initial training was repeated sufficiently, it appears that the dogs did not sufficiently learn the procedures for this experimental method, and appeared not to be comparing the contrastive stimuli in the samples.

"Interestingly, when the stimulus was repeated in either the left or right windows, a propensity toward increased correct responses under certain conditions was evident, including a positional propensity to select the circle, triangle and square (listed in order of preference). In a significantly great proportion of instances, the correct response was made when the shape appeared in the left display window, although it cannot be said that the propensity based on shape is significant. It was confirmed that when humans were present during the experiment, a positive influence was exerted as the subject sensed the presence of the person."