Research on adults' and children's perception of tone sequences or melodies reveals that they more readily detect the same magnitude of change in the context of a ``good'' or well-formed sequence than in the context of a ``poor'' or less well-formed sequence. Because well-formedness is typically confounded with familiarity, the origin of such performance asymmetries often remains unclear. In recent studies, however, 6-month-olds exhibited adult-like patterns of performance, raising the possibility of processing predispositions for good auditory sequences. Such sequences could be considered ``natural'' prototypes, and would be expected to occur frequently across cultures. Natural prototypes might serve a perceptual anchoring function for novice and experienced listeners, facilitating the acquisition of certain kinds of information. Kuhl argues, however, that phonetic prototypes operate differently, being less discriminable from other sounds within the same category than are nonprototypes. In other words, the good instances function as magnets rather than anchors. On the one hand, it is useful for natural prototypes to function as anchors, being highly distinctive and memorable. On the other hand, sufficient flexibility is necessary for the acquisition of culture-specific or ``magnetic'' prototypes.