thanks @twiecki. Although I am not sure this would solve it: What I would like is a series of binomial samples for which the p parameter of each draw depends on the previous draw. Because of this, I thought scan() would be right way of going about it.
How would you go about implementing this?