The Creative Mind: Myths and Mechanisms

Rating:
Summary: Superb
Review: The nature and origins of creativity is of great interest in the field of human psychology, and research on creativity has occupied the time of many individuals. This book, written by an individual well recognized in the field, gives a fascinating account of her ideas on creativity, and has been a major influence for those attempting to realize creativity in non-human machines. Throughout its pages, the author attempts not only to define creativity, but also to give criteria for distinguishing new ideas that are creative from simply new ideas. The most important goal of the book though is to frame the nature of creativity in a computational paradigm.

Philosophers, writers, artists, musicians, and others in the humanities will perhaps frown on the author's project to use computational concepts to assist in the understanding of human creativity. Many of those who work in these professions will point to "intuition" as being the source of creative ideas and works, regardless of their vague understanding of this term. The author demands, rightfully, a clarification of the notion of intuition, but she does not seek to do this herself in the book, but instead looks to the computational paradigm to settle the issue of human and machine creativity. She argues at length throughout the book that not only can this paradigm assist in the understanding of human creativity, but that non-human machines can appear to be creative and can also recognize creativity. The question as to whether non-human machines really are creative is also addressed towards the end of book.

Having a computational model of creativity would allow the design of machines that could produce novel scientific theories, musical compositions, architectural designs, and works of art. It would also give those working in the cognitive sciences greater insight into the understanding of the mechanisms involved in human creative thinking, such as the ability to make analogies, formulate complicated mathematical ideas, or classify patterns. In addition, having a generic computational model of creativity would give insight on how to create novel ideas in a manner that is independent of the domain or context in which one is working. Machines could thus be built that would be able to generate creative ideas in many different domains. For example, a machine that generates new mathematical ideas would also be able to create interesting works of art. This would be a major advance in machine intelligence if such machines could be built.

In the book, the author introduces two notions of creativity that she feels is necessary for her study. One of these is called 'P-creativity', and is a personal and psychological notion. The other is called 'H-creativity' and is a historical notion. Given an idea that arises in a particular person, this idea would be called P-creative if the person could not have had it before. It is irrelevant how many other people already have the same idea. An idea is called H-creative if it is P-creative and no one in human history has ever had it before. As can be seen from the definition, it would be difficult to check whether an idea is really H-creative, since it would require quite an extensive statistical survey. But such a survey would be unnecessary, for the author contends that there is no theory that can explain solely H-creative ideas, but that since H-creative ideas are also P-creative, an explanation of P-creative ideas would also apply to H-creative ideas.

According to the author, the ability to distinguish first-time novelty from genuine originality involves an understanding of what generative processes were involved in the production of the idea. A "novel" idea is one that can be described or produced by the same set of generative rules as other familiar ideas, whereas a genuinely original idea is one that cannot. Therefore, the designation of ideas as "genuinely creative" is always done relative to a specific generative system.

Conceptual spaces, in the author's view, are styles of thinking, and creativity must be thought of in terms of the mapping, exploration, and transformation of conceptual spaces. Conceptual spaces are used to organize and give structure to a particular domain of interest, such as checkers game or a Bach concerto. The structure of a conceptual space can be mapped by forming mental representations of it, and these maps can be use to explore them and to change them if necessary. As one example of this, the author quotes the investigations of tonal harmony in post-Renaissance Western music. The dropping of the home-key constraint in tonal music transformed the conceptual space of tonal music to create the conceptual space of atonal music. The deliberate dropping of a constraint is one method by which conceptual spaces can be generated. Another method is to negate a constraint. The author illustrates this method with a discussion of the discovery of the benzene ring due to Friederich Kekule.

A highly interesting and totally unexpected discussion in the book involves the connection of unpredictability with creativity. Since H-creative ideas are ones that have not appeared before in human history, they are unpredicted. The author emphasizes though that this does not mean that H-creative ideas are unpredictable. The role of serendipity, chance, and coincidences in creative thinking are discussed in terms of the computational paradigm. The author concludes from this discussion that creativity cannot be due to chance alone, but that chance with judgment can. The author also discusses three different notions of randomness, with the intent of shedding light on the belief that determinism is incompatible with creativity. These three notions are labeled 'absolute', 'explanatory', and 'relative'. The first is total absence of any order or structure within the domain; the second the lack of any explanation or cause, and the third lack of any order relevant to some specific consideration. Human creativity is relatively unpredictable the author asserts, and it is easy to believe this considering our folklore understanding of it. The computational paradigm of the author has assisted our understanding of it, but more research is needed.