there's a lovely review by my supervisor that lists (non-exhaustively) 36+55 genes known to be involved in learning and memory in Drosophila. Here's part of that list taken from Keene and Waddell, 2007, NatRevNeuro:
The 55 more refer to this relatively recent mutant screen from 2003 that looked specifically at long-term memory: http://www.sciencedirect.com/science/article/pii/S0960982203000642
There are surely more genes, many of which have been discovered since that review appeared. There have been many learning mutant screens since the 70s pioneered by Tully and Quinn from Seymour Benzer's lab (dunce was the first one in 1976), and the field has widely moved from the question of identifying genes to identifying the neurons in which these genes are expressed, what they do there, and how these neurons interact with one another in the brain. So as you see, your question has been largely answered already, and it might make sense to rather work with the genes that have already been uncovered rather than re-inventing the wheel. On the other hand, mutants in the past have mostly been generated by random mutations, and they may thus be not as 'clean' as the CRISPR technology would allow, so there's a chance for some false positives in the previous datasets that you could uncover.
Sorry, but i am asking question over here. Do you have CRISPER/Cas9 technology at the institute? Do you work on Drosophila or any other organism as well?