Fig. 3.

The mean distribution of smoking behaviour after ten model runs of 200 generations. Here smoking behaviour is clustered around the sampling bias peak of ten cigarettes smoked per day (adapted from Claidière and Sperber 2007)