This touches on what for me is one of the big open questions on what it means to act rationally. I question the common position that the kinds of 'irrational' decisions you describe are actually all that irrational. Many such decisions seem to be rational decisions for an agent with a high time preference at the moment of decision. They may seem irrational from the perspective of a future self who looks back on the decisions when dealing with the consequences but I see the problem as more one of conflicting interests between present selves and past/future selves than one strictly of rationality. As the recent post discussed, rationality doesn't provide goals, it only offers a system for achieving goals. Many apparently irrational decisions are I suspect rational responses to short term goals that conflict with longer term goals.

If I decide to eat a chocolate bar now to satisfy a current craving, I am not really acting irrationally. I have a powerful short term drive to eat chocolate and there is nothing irrational in my actions to satisfy that short term goal. Later on I may look at the scales and regret eating the chocolate but that reflects either a conflict between short term and long term goals or a conflict between the goals of my present self and my past self (really just alternative ways of looking at the same problem). It is not a failure of rationality in terms of short term decision making, it is a problem of incentives not aligning across time frames or between present and future selves. In order to find solutions to such dilemmas it seems more useful to look to micro-economics and the design of incentive structures that align incentives across time scales than to ways to improve the rationality of decisions. The steps I take to acquire chocolate are perfectly rational, the problem is with the conflicts in my incentive structure.