Teleosemantics holds that the contents of psychological states depend crucially on the functions of such states. Etiological accounts of function hold that the functions of things depend on their histories, especially their evolutionary or learning histories. Etiological teleosemantics combines these two features. Consider the case of beliefs. Since selection rests on the stable effects of things, since beliefs have no obvious effects independent of unstable desires, and since desires themselves have mental content, beliefs may seem a hard case for etiological teleosemantics. But David Papineau deploys the effects of beliefs mediated by conation in an artful way to evade these difficulties. I argue that accounts with such an architecture are false.