The basic idea is simple: we learn by attaching agents to K-lines, but we don't attach them all with equal firmness. Instead, we make strong connections at a certain level of detail, but we make weaker connections at higher and lower levels. A K-line for a kite might include some properties like these:
Whenever we turn on this K-line, it tries to activate all these agents, but those near the fringes are attached as though by twice used tape and tend to retreat when other agents challenge them. If most of the kites you've seen before were red and diamond-shaped, then when you hear about Jack's kite, those weak connections will lead you to assume that Jack's kite, too, is red and diamond-shaped. But if you should hear that Jack's kite is green, your weakly activated red-color agent memories will be suppressed by your strongly activated green-color agents. Let's call these kinds of weakly activated memories assumptions by default. Default assumptions, once aroused, stay active only when there are no conflicts. In psychological terms, they are things we assume when we have no particular reason to think otherwise. Later we'll see that default assumptions embody some of our most valuable kinds of commonsense knowledge: knowing what is usual or typical. For example, they're why we all assume that Jack has hands and feet. If such assumptions turn out to be wrong, their weak connections allow them to be easily displaced when better information comes to mind.