No label defined (Q4501630)

Clamans in deserto the author renews in print his pedagogical efforts of 1966 on the methodology of teaching probability theory to beginners, taking expectation as the basic notion from which probability of an event can be derived. The essence of his argument is as follows: let \(E:\mathbb{R}^n\to\mathbb{R}\) be such that: 1) \(\forall{\mathbf x}=(x_1,\dots,x_n)\), \(\min_ix_i\leq E{\mathbf x}\leq\max_i x_i\); 2) \(\forall\lambda>0\), \(\forall{\mathbf x}\), \(E(\lambda{\mathbf x})=\lambda E({\mathbf x})\); 3) \(\forall{\mathbf x}\), \(\forall{\mathbf y}\), \(E({\mathbf x}+{\mathbf y})= E{\mathbf x}+E{\mathbf y}\). These lead to \(E{\mathbf x}=\sum^n_{i=1} p_ix_i\), for each \({\mathbf x}\) for some \(\{p_i\}\), \(p_i\geq 0\), \(\sum^n_{i=1} p_i=1\) determined by \(E\). Then if \({\mathbf e}_j=(0,0, \dots,0,1,0, \dots,0)\), \(E{\mathbf e}_j=p_j\). However, the purpose of the paper is historical, in its subsequent discussion of conditional expectation (leading to conditional probability) according to the ``Principle of Pascal'', since expectation is very largely a French invention. For the entire collection see [Zbl 0940.00005].