I think your hypothesis might be improved if you tried to better define (formalise) the concept of lossless information transfer. It is not obvious to me that losslessness is the best descpription of the difference between gorilla and human communication.

Also, you apparently generalise from one example and even that isn't particularly clear. As you describe it, the gorillas learn that trying to reach the banana causes hosing and they prefer not having the banana to being hosed. When the causal link between banana and hosing is broken, the gorillas still act like nothing has changed. But:

1. Your explanation by imperfect information transfer doesn't work. The gorillas clearly communicate the cause, since new gorillas know that it is eating the banana which is forbidden. It is not clear whether the effect is communicated since the new gorillas clearly know that they shouldn't eat the banana but needn't to know what happens if they break the rule unless they try to, which they presumably never do.
2. The observed evidence (and the assumed difference between gorillas and humans, about whom you seem to think that they would discover that banana doesn't correlate with hose anymore) can be explained in several ways that have nothing to do with information loss: perhaps the gorillas simply don't discover that banana does no more cause hosing. Or they are more hosing risk averse and once they discover that the banana might be associated with the unpleasant experience they avoid it even if they have reasons to think that the risk is relatively low.
3. Beware thoughts along the lines "there is one single thing which causes this important difference and we should only discover which it is". Why do you suppose that there is a single strict and simple principle that distinguishes humans from other animals?