The biggest challenge would be to get funding to run many slight variations of the same experiment without clear indications that this will produce many publishable results. Getting data from humans costs many orders of magnitude more than getting data from your code.