A common objection to aversive control, first proposed by B.F. Skinner in his 1953 book, Science and Human Behavior, is that punishment sucks because it may not stop the behavior forever. Meaning you may have to re-apply it if the behavior comes back. That Skinner had no idea how to stop a behavior for life or that it is entirely do-able is a different set of topics and I will not digress at this time. So I will present the rebuttal alone.
How is this different than a hot stove that always burns you or the crash that comes every time you lose your balance on a bike? Skinner and all the modern behaviorists and “scientific trainers” intentionally ignore the biological function of punishment. It keeps you safe by inhibiting behaviors that can get you killed or seriously injured. That is why YouTube videos of adolescent boys doing incredibly dangerous things is so bizarre to most adults. Only the young intentionally flaunt serious damage or death. (It should be noted that this behavior is the result of positive reinforcement – called “peer acceptance.” This is never chalked up as a downside to positive reinforcement for some strange reason) So, here’s a natural example. A coyote may be threatened by mountain lions more than once in its life. Each time, the coyote is scared spitless and runs like hell, unless it’s this coyote. http://www.liveleak.com/view?i=c77_1384292819. The fear is an integral part of the escape, freeze and/or fight complex. It is transient. As soon as the danger has passed the animal goes back to behaving normally – unless it has been babied it’s whole life.
It’s that babying that often encourages a dog to continue to engage in behaviors that get his butt kicked by other dogs in the house. I have seen many owners make such a fuss over a dog that has gotten into a fight of his own making that it actually becomes more rewarding to get into those fights again.