Two approaches to the evolution of phenotypic plasticity in heterogeneous environments have recently been put forward. The first focuses on selection on the character expression within each environment; plasticity is seen as a by-product of local selection in various habitats. The second approach focuses on selection on the parameters of the response function of genotypes, and selection is thought to change the frequencies of 'plasticity' genes that affect the function. This paper discusses the relationship between the two approaches, with emphasis on applications. A method is described that allows switching from one approach to the other. It is argued that character state and reaction norm approaches, while to a large extent interchangeable, usually differ in the response function chosen. This choice, however, may strongly affect the biological interpretation. The methods outlined in this paper permit one to look at the data from different perspectives in order to avoid this danger.