This paper defines foresight as being a mental model about the future and considers the role of foresight in shaping actions and events reflected in imperious, heroic, tragic and chaotic futures (defined within the paper). The paper contends that success in foresight is not about acquiring knowledge or using it to build pictures about the future. Rather, it is the expectations that come with such processes that cause organisational closure, and thus chaotic and tragic futures. The argument is made that firms need to doubt much more than they do. Two processes of doubting are described: the first (single loop doubting) shows how differences between expectations and perception cause doubt that (whenever the underlying mental model is sufficiently plastic) is accommodated by social processes without change. The second process, called double loop doubting, is based on genuine attempts to refute, rather than confirm, mental models about the future. The contention is that such processes would lower expectations and certainty, thereby opening the organisation and enabling mental models to be more accurate. (C) 2003 Published by Elsevier Ltd.