Chapter 6 put forward the idea that scientists are concerned to accept theories that are informative and close to the truth; and it was assumed there that these values can be represented by a cognitive utility function. Chapter 8 has provided support for the idea that cognitive values can be represented by a cognitive utility function. However, a person's preferences can satisfy all the axioms of Chapter 8 without that person having the sorts of values we would call scientific; in this case, the person's preferences are representable by a cognitive utility function, but that function does not reflect scientific values.

This chapter investigates the question of what sort of cognitive values count as scientific. I articulate and defend the view put forward in Chapter 6: that science values informativeness and closeness to the truth and nothing else.

The notion of truth figures centrally in my account of scientific values. This notion is sometimes regarded as a dubious one, in need of clarification before it can be used with a good conscience. So I should perhaps begin by putting such worries to rest. Fortunately, that is easily done.

All we need to know about truth is that it has what Blackburn (1984) calls the transparency property. This property is that statements of the form ‘It is true that p’ mean the same as ‘p’ (where p is to be replaced by any sentence).