Friendship‐and‐love affords bonding that satisfies what can be called “transcendental preferences”—in contradistinction of “substantive preferences” afforded by, for example, food, clothes, and shelter. Substantive preferences involve ordinary “substantive cost,” whereas transcendental preferences involve “bonding cost” that includes heartaches, obsession, and emotional turmoil. What about the cost of gifts such as flowers, time, and other carriers of friendship‐and‐love? The greater is the expenditure on gifts, the greater the bonding cost. This paper investigates the following question: How should we model bonding cost, which includes the cost of gift, in relation to substantive cost? Given bonding cost and substantive cost share the same budget, neoclassical economists treat them as commensurable and, hence, transcendental and substantive preferences make up a unidimensional objective function. This treatment, however, originates the “gift anomaly”: If people easily substitute between the two genera of preferences, why do they consider the demand of payments for visiting their grandmothers—or payments for voting and sexual intercourse—as repugnant (taboo)? To solve the gift anomaly, this paper is critical of the standard economist's entry point. This paper proposes bonding and substantive costs as incommensurable and, corollary, transcendental and substantive preferences as incommensurable as well. This paper further shows how, without undermining the incommensurability thesis, the incommensurability is up to a limit: the two genera of costs and, corollary, the two genera of preferences are still linked via the income effect—as opposed to the substitution effect.