Even when you were very young, if someone had told you that most Snarks are green — and, also, that every Boojum is a Snark — you would have been able to conclude that most Boojums are green. What could have led you to that conclusion? Presumably, you answer questions about the properties of Boojums by attaching your polyneme for Snark to whichever of your memory-units currently represents a Boojum. Accordingly, you assume that the color property of a Boojum is green by using your usual way of recalling the properties of things you know — activating their polynemes to set your various agencies into the corresponding states. In other words, we do this kind of reasoning by manipulating our memories to replace particular things by typical things. I mention all this because it is often assumed that adults are better than children are at what is often called abstract or logical reasoning. This idea is unfair both to adult and child because logical thinking is so much simpler — and less effective — than common-sense thinking. Actually, what appears to be a matter of logic is usually not logical at all and frequently turns out to be wrong. In the case above, you would have been wrong because Boojums are albino Snarks.
The situation is different when you happen to know more about a particular example. For example, suppose you first had learned that penguins cannot fly and then learned that penguins are a kind of bird. When you discover that, should you replace all of your penguin properties with those of your generic bird? Clearly not, since then you'd lose your hard-earned penguin facts. To deal with this effectively, children must develop complex skills, not merely to replace one representation with another, but to compare two representations and then move around inside them, making different changes at different levels. These intricate skills involve the use of isonomes that control the level-band of the activities inside our agencies.
In any case, to reason well, our memory-control agencies must learn to move our memories around as though those memories were building-blocks. Conceivably, those agencies have to learn such skills before we can learn to build with blocks in the outside world of object-things. Unfortunately, we know very little about how such processes work. Indeed, we're virtually unaware that they even exist, because these kinds of commonsense inferences and assumptions come to mind without the slightest conscious effort or activity. Perhaps this unawareness is a consequence of the speed with which those skills employ the very same short-term memory-units that might otherwise be used to record those agents' own recent activities.