The Relational Responding Task (RRT: De Houwer et al., 2015) was designed to capture implicit beliefs. It was modeled on the Implicit Relational Assessment Procedure (IRAP: Barnes-Holmes et al., 2010), with the dual goals of lowering attrition rates and allowing for the presentation of natural-language statements.
The original version was programmed in Affect 4.0 and can be found on the LIPLAB website. In order to better understand the RRT task I wrote an implementation in PsychoPy. I’ve make the code free and open source under an GPLv3+ so that others might use or improve it.