hddm in the context of a 3-AFC task

[Varg Königsmark]

Hi Everyone,

I'm fitting the model to face recognition data from the Cambridge Face Memory Test. I know that the classic ddm assumes a binary choice, e.g., choosing either target or distractor stimuli. However in the CFMT participants always choose from 3 alternatives, 2 distractors alongside the target. This is somewhat crucial as the chance level differs compared to a 2AFC design, and hence I am concerned that the model could potentially underestimate the difficulty of the task. Conceptually though, the distractors do not represent distinct categories, they are just two different face stimuli and so I am not sure whether it would make sense to e.g., compute drift rates for all alternatives (as in a multinominal ddm) and whether this would even be possible in the context of the hddm? 

Is there, by chance, anything you could recommend or otherwise point me to? Thanks a lot in advance!

Cheers,

Varg