The most widely used experimental paradigm to study the sensorimotor synchronization phenomenon is paced finger tapping, where the subject synchronizes with a metronome with period perturbations to analyze the underlying error correction mechanism. In previous work, we verified the existence of a perturbation context, formed from the experience accumulated during the experiment, either by continuous exposure to a single type of perturbation or by the random alternation of two types, which affects the response. In this work, we experimentally show that within the "pure" perturbation context, responses are incompatible with each other when they come from different types of perturbation. We theoretically support this result by fitting a nonlinear model to these responses and show that the fitted parameter values are mutually exclusive. We also show that, in the case of the "combined" context, where responses are compatible with each other, the model is able to correctly fit both types of time series. These results suggest that the underlying mechanism can be represented by a single model that correctly describes a variety of situations (e.g., different sizes and signs of perturbation), but with the freedom to tune its parameter values to the context and type of perturbation.