Selective breeding is possible, but it certainly won't be easy.
Consider the border collie dog. Some would claim that they have an instinctive desire to herd things. Apparently, even without training, they will try to herd any moving object of about the right size. Groups of children, geese, soccer balls, other dogs, the border collie gets out there and tries to push them together.
Dogs have been domesticated since about 23K years ago. Sheep about 10K years ago. So it is reasonable that dogs began to be selected for doing things around sheep that were helpful about 10K years ago. In that time we have a few varieties of dog that show helpful activity with regard to sheep.
So it would seem that it is possible to produce at least some measure of instinctual behavior through selective breeding.
Rabbits are not currently migratory. Their movements are not adapted to long distances, but rather short bursts of extreme speed and agility. Hippity-hop away from something trying to eat them, and down their burrow to hide.
So to produce a migratory rabbit would be a challenge. You would have to selectively breed both for long distance running and for the proclivity to want to go south in the winter. It means not only do you need to make changes in brain structure but also in their legs and metabolism. You need a critter than can operate at a medium exertion level for many hours a day for some weeks.
So you could apply some scientific measurements. Select rabbits that had the closest to the correct metabolism. Select rabbits with legs a little less adapted to hopping and sprinting and more adapted to "jogging." Run them on test runs and find the ones that make it, say, 5 km. Then 10 km. Then 20 km. And so on.
Then start selecting for rabbits that show up at a given location at a given time of year. That's going to be a challenge because you need to have the rabbits run loose through whatever terrain happens to be around. Let them loose here and go over there and catch them. Maybe you get a truck load of radio pingers to track them. The ones that get closest to the destination get picked up and permitted to have babies. The rest go "to live on a farm somewhere." (Heh heh.) Lather-rinse-repeat, gradually increasing the distances between the start and end, and also including the return after winter is over. And you probably want to move (or start a restaurant with) the rabbits not in your program that might be hanging around.
The progress would be quite difficult to predict. That is because it will depend on mutation, which is to some extent random. Plus, you will be building in the instincts required one neuron at a time so-to-speak. You randomly get a brain config that tells the rabbit to go south 50 km, and then you have to find the sucker before a hawk or a coyote does. Migration is a complicated thing to do by instinct, requiring timing, direction, and overall distance. And the requirement to get some extra body fat before leaving. And the previously mentioned metabolism changes. This is likely to require a bunch of mutations that have to be connected in a useful fashion. It is extremely difficult to predict how long this process might take.
There are some species of rodents that are migratory. Lemmings for example. Possibly if you can do some gene splicing, you can steal some migratory tendencies from lemmings and do a transfer. That might give you a big advance.