• Recognition of new instances based on comparison to detailed memories was examined in meter perception.
• Meter—the “beat” of a piece of music—is thought to be calculated by reference to idealized representations.
• Different listening experiences led to different metrical percepts in metrically-ambiguous melodies—metrical “restoration”.
• Musical properties affected metrical restoration when they were consistently associated with meters during exposure.
• Metrical restoration is not a simple function of music recognition.
How do perceivers apply knowledge to instances they have never experienced before? On one hand, listeners might use idealized representations that do not contain specific details. On the other, they might recognize and process information based on more detailed memory representations. The current study examined the latter possibility with respect to musical meter perception, previously thought to be computed based on highly-idealized (isochronous) internal representations. In six experiments, listeners heard sets of metrically-ambiguous melodies. Each melody was played in a simultaneous musical context with unambiguous metrical cues (3/4 or 6/8). Cross-melody similarity was manipulated by pairing certain cues—timbre (musical instrument) and motif content (2–6-note patterns)—with each meter, or distributing cues across meters. After multiple exposures, listeners heard each melody without context, and judged metrical continuations (all Experiments) or familiarity (Experiments 5–6). Responses were assessed for “metrical restoration”—the tendency to make metrical judgments that fit the melody’s previously-heard metrical context. Cross-melody similarity affected the presence and degree of metrical restoration, and timbre affected familiarity. Results suggest that metrical processing may be calculated based on fairly detailed representations rather than idealized isochronous pulses, and is dissociated somewhat from familiarity judgments. Implications for theories of meter perception are discussed.