One of the modern classics of economics is an article from 2006 with the self-explanatory title Paying Not To Go To The Gym, in which researchers Stefano DellaVigna and Ulrike Malmendier studied the behaviour of nearly 8,000 gym members and found it “difficult to reconcile with standard preferences and beliefs”.

By that, they meant that gym members seemed to be delusional, weak-willed or both. People on a monthly contract paid more per visit than those who simply showed up and paid at the door, suggesting they either had a very basic problem with arithmetic or, more likely, optimistic expectations about how often they would exercise. People on the rolling monthly contract also tended to let more than two months elapse between the last visit and the moment they got round to cancelling their membership.