High-Pass Filter in Human Connectome Project

I am currently setting up first-level analyses for the HCP tasks (using SPM) and wonder if there is a recommended value that one should set for the high-pass filter?

In the Barch et al. paper from 2013, they set a quite low high-pass filter (200s) but I cannot find any reasoning in this paper why they chose this specific value instead of the canonical 128s:

Subsequently, both the 4D timeseries and theGLMdesignwere tempo- rally filtered with a Gaussian-weighted linear highpass filter with a (soft) cutoff of 200 s. Finally, the time series was prewhitened within FILM to correct for autocorrelations in the fMRI data