(Not a literal translation, paraphrased)
If you learn it a little bit you are motivated to get better and better. That is why you keep on making the correct movements fitting to the perceptions. This way you produce robust connections between so-called perception cells and action cells. He makes a reference to Hebb: "Cells that fire together wire together". The more you do it, the more ingrained it becomes. Because every bike works (almost) the same, the relation between perception and action is stable. You never experience something completely different on a bike, that's why you never unlearn it.
So, you might think, that's not such a bad answer. And no, it's not. It's a bit unclear on which parts of the brain are involved. What are perception cells and action cells, but other that that it sort of makes sense. But in the last section it goes off a bit:
Less consequent experiences are more susceptible to forgetting. That's why we have problems forgetting what we ate last week for every day. This is because we eat something different almost every day, the memories interfere and we are not able to form strong memorytraces.
Right, so riding a bike is part of declarative memory? It seems that the answer here is solely built on the principle that if a trace is enforced every time in a constant way it becomes stronger. Though true, it doesn't tell you why you never unlearn to ride a bike, even though you haven't done it for years, while you do forget what you ate yesterday!
What about those pesky Dutch spelling rules? We got drilled in school to know where to put d, t or dt. But sometimes I forget.... Why do we forget numbers so easily, but are always able to thump in our pin at an ATM machine? (Honestly, sometimes I can't remember the digits, but when I see a keyboard I can type it!). The same goes with passwords at your computer, sometimes you can't remember, but your fingers can!
The answer of course lies within the cerebellum. If you say coordinated movements, such as riding a bike, you say cerebellar involvement. On the other hand, riding a bike is also very much a sequence that you have learned, so you would expect striatal involvement. It's probably a bit of both; maybe first cerebellar coordination to get the movements right, then striatal consolidation? Why then don't you unlearn how to ride a bike, while you do forget other things? Although we do not know the cellular and molecular mechanisms, this question can be answered via a different route.
Imagine what would happen if you would learn and unlearn movements in the same vivid and thorough way as you remember your lunch. You would step onto the greens of a golfcourse and you would be able to swing the perfect ball within a few trials. Sounds perfect, doesn't it. But a quick learning implies a quick unlearning as well. You would not be able to make that perfect swing again in one go. And that's what you don't want for your movements. You would forget how to walk after sitting for an hour. So, movements are learned very slowly, so they are also unlearned or forgotten very slowly. Why don't you forget how to learn to ride a bike? Because it took long to learn.