We created experimental situations where observers (great tit, Parus major and marsh tit, P. palustris) acquired one of five types of experience near conspecific and non-conspecific demonstrators: (1) neither demonstrator was feeding; (2) only the conspecific was feeding; (3) only the non-conspecific was feeding; (4) both were feeding and the observer received a reinforcement (food) near the conspecific; and (5) as (4), but the reinforcement was given near the non-conspecific. After each treatment, we recorded whether the observer approached a caged conspecific or a caged non-conspecific. There was a baseline preference for approaching conspecifics but this could be overcome by learnt associations so that the birds would then approach non-conspecifics. When there was an opportunity to distinguish between successful and unsuccessful foragers (demonstrators), and the successful forager was not conspecific, observers of the dominant species approached the successful subordinate non-conspecifics. Observers of the subordinate species approached the dominant species only if they had received a food reinforcement near them. Observers followed non-conspecific individuals more often at temperatures below than above 0 degreesC and chose a conspecific individual more often above than below 0 degreesC. Copyright 1998 The Association for the Study of Animal Behaviour. Copyright 1998 The Association for the Study of Animal Behaviour.