How is median RT computed in HCP behavioral data?

[Steve Norkus]

Hello everyone,

In the HCP behavioral datasets (e.g. working memory, relational processing tasks), the reported “median RT” is provided as a summary measure.

Does anyone know the exact procedure used to compute this median?

In particular:

Is it computed over all valid trials as-is, or are any trials excluded beforehand (e.g. anticipations, timeouts, extreme RTs)? If preprocessing is applied, is there a documented thresholding rule or trimming procedure (e.g. fixed RT bounds, percentile trimming, or SD/MAD-based exclusion)?

I couldn’t find a precise description in the HCP task documentation and would appreciate any pointers to the implementation details or relevant technical notes.

Best,
Steve Norkus