On the possible scientific laws (Q809027)

Suppose that one has a psychophysical or other scientific law in which the relationship between the dependent and independent variables is \(x_{n+1}=u(x_ 1,...,x_ n)\). For each variable \(x_ i\) there is a set \({\mathbb{T}}_ i\) of admissible transformations that represents the scale invariance of the variable. \textit{R. D. Luce} [see J. math. Psychology 1, No.1, 278-284 (1964; Zbl 0136.415)] pointed out that if there are no `dimensional parameters' which cancel out the effects of the transformations, there is a transformation \(D(T_ 1,...,T_ n)\) such that \[ u(T_ 1(x_ 1),...,T_ n(x_ n))=D(T_ 1,...,T_ n)[u(x_ 1,...,x_ n)]. \] Luce calls this transformation the `principle of the theory'. In fact one would expect that knowledge of the scale invariance of the variables would determine the general form of the function u. This paper generalizes earlier results in this direction for the case where u is continuous. The results for different assumptions on the scale type of the independent and dependent variables are presented in the form of a table.