The data reported in this paper illustrate the importance of incorporating elements of naturalistic contexts into experimental design and how the failure to do so can lead to erroneous conclusions regarding behavioural plasticity in a communication system. The slow repetition rate of the notes of the mallard, Anas platyrhynchos, maternal alarm call is the primary acoustic feature affecting freezing (vocal and locomotor inhibition) in ducklings. Previous experiments indicate that alarm calls with notes pulsed between 0.8 and 1.8 notes/s are particularly effective in promoting behavioural inhibition in domestic mallard (Peking) ducklings reared and tested individually. In light of field observations, the ineffectiveness of repetition rates slower than 0.8 notes/s was surprising and might possibly have been caused by species-atypical rearing and testing contexts. Thus, in the present study, Peking ducklings were reared and tested socially to alarm calls that had their notes temporally altered such that the repetition rates fell outside the range normally effective for non-social ducklings (0.2 and 2.6 notes/s). Compared with ducklings reared and tested individually, the incidence of freezing by social ducklings was significantly greater at 0.2 notes/s and was comparable to the level of freezing to such slow-rate notes on nests in the field. There were no differences in freezing between social and individual ducklings at 2.6 notes/s. These data reconcile a discrepancy between held observations of alarm-call responsivity and certain laboratory tests and show that repetition-rate specificity can be altered as a function of rearing and testing context. Furthermore, the observed change in repetition-rate specificity is specific to a form of stimulation that ducklings normally encounter on natural nests (i.e. slow-rate notes). These and other data illustrate that context can be an overwhelmingly important, although often neglected, variable in behavioural and neural research.