On the possible merging functions (Q1122455)

This paper explores which `laws of science' (as determined by \textit{R. D. Luce} [Psychol. Rev. 66, 81-95 (1959); J. Math. Psychol. 1, No.1, 278-284 (1964; Zbl 0136.415)], \textit{D. K. Osborne} [ibid. 7, 236-242 (1970; Zbl 0205.491)], the first author, \textit{F. S. Roberts} and \textit{Z. Rosenbaum} [J. Math. Anal. Appl. 119, 389-416 (1986; Zbl 0605.39004)], \textit{S. Kim} [On the possible scientific laws, Tech. Rep. RRR-22-88, Rutgers Center Oper. Res., Rutgers Univ. (New Brunswick/N. J. 1988)] and others; some of these results are modified or extended in this paper) can serve as `merging functions' of individual numerical judgements. Scale types (ratio, interval, log-interval, ordinal), meaningfulness, `agreement' (u(x,\(\cdot \cdot \cdot,x)=x)\), symmetry, linear homogeneity and a large number of functional equations play prominent roles. In many cases the arithmetic mean and in even more the geometric mean turn out to be the only merging functions satisfying the conditions.