“I never want to do X again. I’ve decided it before, but this time is different. This time I really mean it, I feel it so strongly, I’m definitely never doing X again.”
Two potential models of regret:
Hydraulic model of regret: regret builds up over time. The first time you regret doing X, you won’t stop doing it. But as the weight builds up over time, as you keep doing X, you start feeling more and more strongly that you really should quit. Eventually you feel so strongly that you just stop, out of willpower.1
Memoryless model with black swans: almost every time you regret doing X it feels the same as all the previous times. You feel equally strongly every time, but you keep failing–keep relapsing. Then, one day, something changes. You feel regret and you change your situation or drastically change your outlook and all of a sudden, it’s not a struggle any more–you just Don’t Want To Do X or Set Up Your Life So That X Doesn’t Fit. Now, if only you knew what was different those times, you’d be set, but they seem to appear randomly.
I think that my experience is more consistent with model 2. Thoughts?
UPDATE (7/22/13, 5:48 PM): Two alternative models of regret have emerged from the pundit sphere.
The first, from @EMGurevitch: “you regret x, repress x, unconsciously reenact x. cf Freud’s “remembering, repeating, and working through” “. Interesting, though I don’t quite what this predicts in terms of doing (or refraining from) X in the future.
The second, from @letthemeatfood; “You do X and regret it but the short-term benefit outweighs the regret to the extent that you do X again and again.” I like this point a lot. Regret doesn’t mean that you made a bad decision–the regret could be an acceptable cost. In fact, suppose that regret for doing X gets weaker over time as you become habituated more and more to doing X. Then weaning yourself off these unhelpful emotions could be an active planned consequence of what you do. Point well taken!
Note: This is consistent with a Bayesian model–every time you feel regret you update your priors a little bit until the evidence is strong enough. ↩