Comment on Reproducing prospect theory with 'differentiable decision theories'