Piaget proposed that the change from one stage to another involved a ``massive structural change''. In other words, large change in behaviour must be caused by large change in structure. Unfortunately, he didn't say much about possible mechanisms. Young wanted to explore the idea that, given a suitable computational system, you can get large changes in behaviour from small changes in structure - specifically, by adding small pieces of rule-like knowledge in a way that doesn't interfere with knowledge that's been added before.
To do this, he constructed a set of rules which produced stage 1 behaviour, and then showed that you could add a few more rules to get stage 2, and a few more to get stage 3. So this is a model of the development of seriation skills in children. When a new skill is acquired, a new rule which describes it gets added to the existing productions in LTM. This does not explain how such new rules are created, but it's an important step towards such an explanation.