This paper argues for a computational model of creativity as the unique and favorable combining of past experiences. It is hypothesized that the inner process that combines the experiences is universal and largely syntactic, i.e. it is only the emotional associations that are attached to encoded experience and not the semantic content of experience that is used by the inner creative process. A chess system, "Morph", has been developed that can be viewed as exploring this model. Morph is limited to 1-ply of search, little domain knowledge and no supervised training. It is only through playing another program, and encoding its experience (as "pattern-weight" pairs) that it can improve. Morph's learning system is a combination of machine learning methods that have been successful in other settings - and is covered in depth. Part of Morph's strength comes from a powerful representation for chess patterns that allows generalizations across positions (experiences).