Abstract

We begin by accepting that introspective evidence is important to cognitive science. However, as its history shows, introspection is risky, so methods should be used that minimize those risks. We argue that there are 13 ways that a beeper can reduce those risks, dividing those ways into three categories: time sampling per se, minimizing the reactive disturbance of evanescent phenomena, and aiding phenomenological fidelity. We turn aside six criticisms of beeper-based research, and describe five characteristics of a good beep.