Polarization of starlight and thermal dust emission caused by aligned dust grains is a valuable tool to characterize magnetic fields (B-fields) and constrain dust properties. However, the grain alignment physics is not fully understood. To test the radiative torque (RAT) paradigm, including RAT alignment (RAT-A) and disruption (RAT-D), we use dust polarization observed by Planck and Stratosphere Observatory for Infrared Astronomy/High-resolution Airborne Wideband Camera Plus toward two filaments, Musca and OMC-1, with contrasting physical conditions. Musca, a quiescent filament, is ideal for testing RAT-A, while OMC-1, an active star-forming region OMC-1, is most suitable for testing RAT-D. We found that polarization fraction, P, decreases with increasing polarization angle dispersion function, S , and increasing column density, N(H2), consistent with RAT-A. However, P increases with increasing dust temperature, T d, but decreases when T d reaches a certain high value. We compute the polarization fraction for the ideal models with B-fields in the plane of sky based on the RAT paradigm, accounting for the depolarization effect by B-field tangling. We then compare the realistic polarization model with observations. For Musca with well-ordered B-fields, our numerical model successfully reproduces the decline of P toward the filament spine (aka polarization hole), having higher N(H2) and lower T d, indicating the loss of grain alignment efficiency due to RAT-A. For OMC-1, with stronger B-field variations and higher temperatures, our model can reproduce the observed P–T d and P−N(H2) relations only if the B-field tangling and RAT-D effect are incorporated. Our results provide more robust observational evidence for the RAT paradigm, particularly the recently discovered RAT-D.