It's more than that, it's the danger of skewing your animal model in a direction you didn't predict and invalidating your results.
Joe Garner published an article in Scientific American over ten years ago in which he did research on the behavioral effects of standard laboratory environments on lab animals compared to control (living with lots of chips and paper tubes and food that needed to be found and a day/night schedule and socialization and all that stuff that's not too uncommon in a gerbil cage but completely absent in a lab environment). What he discovered was that pretty much any lab animal, from mice to monkeys, went what could be easily recognized as One-Flew-Over-The-Cuckoo's-Nest crazy as they transitioned from adolescence to adulthood. In other words, all our behavioral studies since the invention of the cage have been performed with animals that are depressed, neurotic, and irrational, often to the point of rendering their behaviors unrecognizable compared to their nature-raised peers.
To the best of my knowledge, nothing has changed.