How does honest costly signaling work?

The honesty of animal communication has long been disputed and continues to be largely associated with the "handicap principle," in which strategic costs are paid by signalers in order to maintain signal honesty. Recently, several manuscripts have forcefully made the point that costly signaling models do not require handicaps (realized signal costs) to be paid by signalers. The aim of the present manuscript is to reiterate, clarify, and develop some of those ideas. Although the message that handicaps are not necessary for honest signaling seems to have been readily accepted, the point that handicaps are also not sufficient for honest signaling has been less well understood by the communication community. I argue that the central separation should not be between "handicaps" and "other" mechanisms such as "indices" and "punishment of cheaters." Indices, for example, are in fact a trivial exemplar of costly signaling, in which the cost functions associated with exhibiting a greater level of signal of signal expression are simply infinite. I advocate that the essential distinction is between circumstances where there is alignment of interest between signaler and receiver and those where there is not. I reenforce the view that the current focus on measuring the realized costs of signals ("handicaps") is misplaced, as without further constraints the costly signaling paradigm makes no predictions about whether honest signalers pay strategic costs for giving their own signals. Rather, it only predicts potential costs for individuals giving dishonest signals—costs that are often not paid in an honest system.