"It's just what I do in the evening"

The three jobs cover most of what the tin does, but if you kept the notes I suggested in chapter 1, you may have found some entries that fit none of them. You weren't rewarding yourself, you weren't tired and you weren't upset. It was nine o'clock, the television was on, and you had something sweet because that's what happens at nine o'clock.

This kind of eating deserves its own short chapter, because the arguments of the last three chapters don't touch it. There's no belief in it to examine. If I asked why you were eating the biscuit, you'd say "I don't know, I just always do."

There's a well-known experiment on this kind of eating. David Neal, Wendy Wood and colleagues gave people going into a cinema a free box of popcorn. Some boxes were fresh. Others contained popcorn that was a week old and had gone stale. Afterwards the researchers weighed what was left. People who rarely ate popcorn at the cinema ate much less of the stale popcorn than the fresh, as you'd expect, because it tasted bad. People with a strong cinema popcorn habit ate just as much of the stale as the fresh. They said afterwards that they hadn't liked it. How hungry they were made little difference either.

The researchers then repeated the experiment in a meeting room, with people watching music videos. There, the habitual popcorn eaters behaved like everyone else and left the stale popcorn alone. The habit ran only in the setting where it had been built. In a second study, habitual eaters in the cinema were asked to eat with the hand they didn't normally use, and that small disruption was also enough to make the taste matter again.

That's how the nine o'clock biscuit works. The sofa, the programme and the time of evening together call up the eating, and whether you want it or enjoy it has little say. It's how you came to eat the third and fourth biscuits on Thursday without tasting them.

This kind of eating gives you the least, since there isn't even a reason attached, and it's also the easiest kind to change, because nothing is holding it up except the setting. The popcorn experiment shows how. If you change the setting a little, the automatic link stops firing. Here are a few ways of doing that, which are my suggestions.

Keep the tin somewhere that isn't the sitting room or the route to it, as chapter 13 describes. If eating means getting up, finding a chair and reaching into a high cupboard, it stops being automatic.

Have your pudding at the table, with dinner, before the sofa. Then the sweet part of the evening has already happened by the time the programme starts.

Give your hands something else to do for the first couple of weeks. A cup of tea works, and so does knitting, a crossword or the dog's ears. It sounds trivial. The habit is partly a hand habit, as the researchers found when they made people switch hands.

And expect the reach. For a while, at nine o'clock, some part of you will feel that something is missing. It's the same feeling as patting your pocket for keys you've already put down. It doesn't mean you need a biscuit, and it fades quickly once the evening stops including one.