Fable
The apprentice learned classification by identifying similarities and opposites
The painting hall took in two apprentices. The master did not first lecture on the principles of "good painting," but only placed a basket of apples, a basket of pears, and several similar fruit pictures on the table, and had them sort every day.
On the first day, the apprentice sorted by color. Red went to one side, yellow to the other. The master brought a green apple and a yellow pear, and they were immediately confused. It turned out that color could help a little, but it was not a reliable way to identify things.
The next day, they sorted by size. Small apples were put in the pear pile, and large pears were put in the apple pile. The master did not scold them, but had them place the wrongly sorted fruit together and look carefully at which parts were similar and which were actually different.
The apprentices began making pairwise comparisons. The same apple, seen from the front and from the side, should be recognized as the same kind; an apple and a pear, even if similar in color, should be pulled apart. Each time they misclassified, the master had them point out "what should be brought closer and what should be pushed apart."
At first they remembered only a few examples. When they encountered a spotted fruit, they hesitated again. So the master added more paired exercises: the same fruit under light and shadow must still stay close, while different fruits must not be mixed even if they share a plate and color.
A few weeks later, the apprentices no longer fixated on color or size. In their minds they had formed an invisible distance map: similar items should sit close together, easily confused ones should be pulled apart, and the harder a pair was to distinguish, the more it sharpened their eye.
The master also deliberately introduced difficult cases: apples bitten by insects, pears that grew round, fruit drawn in shadow. Easily distinguishable appearances cannot teach true discernment; only pairs that look most alike but are different, or are the same kind but deformed, force the apprentice to grasp stable features. After a while, the master drew the same apple as a distant view, a close view, and partially occluded. If the apprentice classified them into three types, it showed he only looked at the surface; if he also pulled a round pear into the apple pile, it showed he had gotten too close. The difficulty of training lies precisely in finding the distance between "variation within the same class" and "similarity across different classes." The painting hall later divided these pairs into difficulty levels. First learn obvious differences, then subtle differences, so that the apprentice's discernment does not stop at the surface.
Only at the end did the master let them draw a new fruit. Though neither had memorized definitions, they could grasp form and texture. The people in the painting studio understood that recognition ability does not grow from isolated samples, but is honed through repeated comparisons of "these two should be close, those two should be far."