Jev-like models give calibrated decision probabilities, but at low accuracy.

So why didn't they show both??