The comparable likeness of being

Was George Bush justified in likening Saddam Hussein to Adolf Hitler in the run-up to the Persian Gulf War? What role do analogies play in scientific discovery, courtroom procedures, psychotherapy and computer modelling? These are some of the questions raised in this wide-ranging book written jointly by a psychologist and a philosopher-psychologist with an interest in computer science.

The focus of the book is to present what cognitive science has learned about analogical thinking. The authors' own approach, which they call "multiconstraint theory", understands analogy as involving three distinct but interrelated constraints, namely similarity, structure and purpose. If "thinking by analogy is an implicit procedure applied to explicit representations", one of the goals of cognitive science is to make this operation explicit. Concepts are formed according to the detection of similarities between situations despite their differences, then more complex structures are created through combining these concepts; and this is done in view of a particular purpose, such as explanation, problem solving or planning.

Three levels of complexity in analogical thinking are distinguished: attribute mapping based on the attributes of objects, relational mapping based on relations between objects, and system mapping based on higher-order relations (such as cause) between relations. This analysis is applied "in the thinking of organisms (sic) as diverse as apes, politicians, scientists and poets". The chapters describing experiments with apes and with young children demonstrate the evolution of analogical thinking and the stages of human cognitive development. Monkeys seem to be capable of attribute mapping, while specially-trained chimpanzees get as far as relational mapping before "hitting a wall". Young children acquire the skill of relational mapping around the age of three, and five-year-olds can reason about higher-order relations.

A basic mechanism in the construction of similarity is the perceived relationship between what the authors call the source analogue to the target analogue: to what extent can the target be seen as "the same kind of thing"? Here we come to the Saddam/Hitler analogy that became a tool for argument and indeed an instrument of propaganda during the Gulf War. A recent book by Raymond Gibbs - The Poetics of Mind - cites United States senators characterising Saddam as a "geo-political glutton", a tactic also designed to elicit specific emotional responses.

The book looks at this particular analogy at two levels: the nature of the analogical mapping and the consequent inference towards deciding on a certain course of action. This is analysed in some detail, showing how two readings of this putative parallel are possible. The first is to map Bush on to Roosevelt. The second arises from the consideration that the US did not go to war until bombed by Japan, and therefore suggests mapping Bush on to Churchill. Correspondingly, Kuwait becomes Poland or Austria, and Saudi Arabia France. One of the authors then carried out a study using two differently emphasised accounts of the second world war and established that the subsequent choice of mapping was significantly correlated with the version which the subjects had just read. This is extended in an intriguing way to another study showing how seemingly disparate problem situations can more easily be solved when we apply the lesson of one to another - in this case the necessity of a multi-pronged attack on a fortress to the optimum approach of laser therapy on a tumour.

After an interesting consideration of the role of precedents and hence analogies in the legal process, the authors return to the role of analogies in American foreign policy with a detailed look at the arguments surrounding US entry into the Vietnam conflict. They rephrase George Santayana's maxim that "those who do not remember the past are condemned to relive it" by warning that those who do not forget the past can be led to misapply it (readers who are golfers will recognise that both versions apply to club selection under different conditions on the same hole). President Johnson and his advisers used the Munich appeasement analogy in public, but in practice relied more heavily on the positive Korea analogy, which was by no means isomorphic. It is here that skill in evaluating analogies is absolutely vital, both in the context of decision-making and in the use of analogies in teaching. Since then, the lesson of Vietnam itself has loomed large in foreign policy decisions about potential US interventions.

The use of analogy is then considered in philosophy - Plato's cave, then quotations from Francis Bacon and W. V. O. Quine, as well as the argument for other minds and William Paley's argument from design for the existence of God. In this last context, a discussion of the origins and pervasiveness of the mechanistic world-view would have been helpful, all the more so in view of the chapter on analogical computers where it becomes clear that the authors hold a fairly strong artificial intelligence position. Nor is there any mention of the anthropic principle as a sophisticated reformulation of Paley's argument, which is presumed to have been demolished by Charles Darwin. It is also surprising, especially in an American book, that while there is a section on religious and philosophical analogies from Asia, there is no mention of the use of parables in the New Testament.

The treatment of scientific analogies - for discovery, development, evaluation and exposition - clearly shows the role of analogical thinking in creative thought, which the authors extend in the final chapter to specially designed computer programmes. The authors quote Samuel Butler's remark that "though analogy is often misleading, it is the least misleading thing we have".

The book is at once a contribution to the continuing scholarly debate on the nature of analogy and an accessible interdisciplinary overview for the general reader of this central process of thinking and communication.

David Lorimer is director, Scientific and Medical Network.