Set bias 'z' as a prior?

[cmer...@berkeley.edu]

Hi, first thanks for making the toolbox available, it has some pretty neat functionality. 

I am curious if there is a way to either set the bias 'z' for each subject individually or set a prior at the group or subject level that will account for bias that subjects have before going into the experiment.

In the task participants are deciding to reach with their left or right hand, but all subjects are right handed and come in with a right hand bias. We manipulate the reward rate for reaching with their left and right hand - trying to get them to use their left hand more. Ideally we would fit a prior based on reaches that the participant makes before we begin to manipulate reward rate (which would give us a measure of their right hand bias).  But I am not sure how I would implement that into the model. 

I ran a model using stimulus coding in HDDM regression including z, but the bias was not what we were expecting, which made me feel uncomfortable to interpret other variables. I think setting the prior for z may be the next thing to try but please let me know if you have any other ideas. 

Thank you for your help!

Christina 

[Thomas Wiecki]

Hi,

On Fri, Feb 5, 2016 at 10:21 PM, <cmer...@berkeley.edu> wrote:

Hi, first thanks for making the toolbox available, it has some pretty neat functionality. 

I am curious if there is a way to either set the bias 'z' for each subject individually or set a prior at the group or subject level that will account for bias that subjects have before going into the experiment.

That's not so easy. 

In the task participants are deciding to reach with their left or right hand, but all subjects are right handed and come in with a right hand bias. We manipulate the reward rate for reaching with their left and right hand - trying to get them to use their left hand more. Ideally we would fit a prior based on reaches that the participant makes before we begin to manipulate reward rate (which would give us a measure of their right hand bias).  But I am not sure how I would implement that into the model. 

Priors aren't fit but set. It sounds more like you want to infer z for the different condition hand. That model should be easy to fit using the depends_on kwarg (see the docs for more info). 

I ran a model using stimulus coding in HDDM regression including z, but the bias was not what we were expecting, which made me feel uncomfortable to interpret other variables. I think setting the prior for z may be the next thing to try but please let me know if you have any other ideas. 

HDDMRegressor does not support stimulus coding I think? If you're using stim coding and want to infer z you can try split_by='v'. See the HDDMStimCoding doc for more info.

Best,

Thomas 

Thank you for your help!

Christina 

--
You received this message because you are subscribed to the Google Groups "hddm-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hddm-users+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.