To enhance effectiveness in real world applications, autonomous agents have to develop cognitive competencies and anticipatory capabilities. Here we point out their strong liaison with the functional roles of affective mental states as those of humanlike metaphor: not only the root elements for both surprise and anticipation are expectations, but also part of the effects of the former elicit efforts on the latter. By analyzing different kinds of expectations, we provide a general architecture enhancing practical reasoning with mental states, describing and empirically evaluating how mental and behavioral attitudes, emerging from mental states, can be applied for augmenting agent reactivity, opportunism and efficacy in terms of anticipation.