Scientific knowledge does not necessarily replace intuitive misconceptions; successful physics reasoning may therefore require learners to control responses that remain cognitively available. However, inhibitory-control accounts have yet to explain whether the same conceptual conflict elicits a stable amount of control across representational contexts or whether control mobilization varies with contextual typicality and physics training. This study addressed this issue using a negative-priming (NP) paradigm in the domain of static friction. High-school students and undergraduate physics majors made static-friction judgments in a prototypical context and a nonprototypical context. Within the inhibitory account adopted here, NP magnitude was interpreted as a behavioral index of the control mobilized during the preceding incongruent prime. Probe-phase reaction times were analyzed on the log scale using a linear mixed-effects model with Group, Context, and Prime Type and their interactions. The analysis revealed a context-dependent pattern of NP, with the magnitude of this contextual modulation differing between the two groups. Specifically, NP was negligible in the prototypical context for both high-school students and physics majors. By contrast, reliable NP emerged in the nonprototypical context. This increase was substantially larger among high-school students than among physics majors. When translated back to the original reaction-time scale, NP in the nonprototypical context corresponded to an approximately increase for high-school students, compared with a increase for physics majors. Taken together, these findings show that the inhibitory control mobilized by the same conceptual conflict is not fixed across representations. The nonprototypical representation amplified inhibitory-control demand, whereas physics training attenuated this amplification. The training-related advantage is therefore better characterized as reduced contextual sensitivity of inhibitory mobilization than as a context-invariant absence of conceptual conflict. The results advance process-based accounts of conceptual change by linking problem representation, domain knowledge, and task-evoked inhibitory control.