Nanoscale confinement alters the dynamics of glass-forming liquids in films of order 100 nm in thickness. A common hypothesis for the origin of this long range posits that interfacial dynamics propagate into the film via cooperatively rearranging regions (CRRs). However, the precise nature of the dynamic interface remains uncertain, and its identification with CRRs has yet to be firmly established. Based on results from coarse-grained molecular dynamics simulations of a freestanding polymer film, here we show that an apparent qualitative discrepancy between computational and several experimental measures of the interfacial dynamic length scale results from a difference in definition and not from a difference in underlying behavior of experimental and simulated systems. We then show that the computational definition exhibits a direct correspondence with the expected behavior of CRRs as predicted by the Adam-Gibbs theory, and we suggest that it should be possible to extract this length scale from experimental measurements.