Am I using this correctly?

[tj]

I'm a newbie to HDDM and I'm trying to make sure that I am using this modeling program correctly. I have 6 different conditions in my experiment. On each trial, subjects are shown one of three cues, followed by one of two kinds of stimuli (hence the 6 different conditions). I want to be able to compare the components of the DD model, like z or bias, in these different conditions. To do that, we are using the "depends_on" variable in the hddm.model.HDDMBase function, which we have used to divide our trials by either cue condition, stimuli condition or both. 

My question is: Will I get meaningful results from dividing the data up this way? Am I using this variable as it is intended to be used?  

Thanks so much!

[Thomas Wiecki]

Hi,

That all sounds correct. Please read this:

http://ski.clps.brown.edu/hddm_docs/howto.html#have-separate-parameters-for-different-conditions-using-depends-on

And let me know if there are further questions.

Finally, make sure to set include='z' (or bias=True) if you want to include bias.

Thomas

[tj]

Thank you so much, Thomas!! I really appreciate it. 

[tj]

Thanks again for your earlier response - very helpful!

One more question about this analysis: I was looking over our average drift rates, and they seem to be the opposite of what I would expect. My understanding is that drift rate is related inversely to reaction time, with fast reaction times associated with high drift rates, and slow reaction times associated with low drift rates. Currently, in my analysis, it is the opposite. Any thoughts about why this may be?

Thanks so much!

[Thomas Wiecki]

While true it turns out that drift-rate is more strongly affected by accuracy. Do the estimates make more sense when looking at that?

Thomas

[tj]

Hi Thomas,

Thanks so much for your quick reply. Yes, when I look at accuracy, it makes a lot more sense. I will read up on drift rate more to try to undertand the connection between it, accuracy and reaction time. 

[Thomas Wiecki]

You might want to try running hddm_demo.py to get a feel for how the parameters affect the RT distributions. I found that very helpful. I don't think it works on OSX though.

[tj]

Thanks so much, Thomas, I will try that!

[tj]

Hi Thomas - 

I've looked a few different places, and have not been able to locate hddm_demo.py

Could you tell me where to find it?

Thanks so much!

[Thomas Wiecki]

Which OS are you running?

Thomas

[tj]

I am using windows 7 Ultimate.

Thank you!

[Thomas Wiecki]

What happens if you just start cmd.exe and then type hddm_demo.py?

[tj]

Got it working. Thank you again!

[tj]

Hi Thomas,

Got the hddm_demo program running. Being able to manipulate the DDM components and see how they are graphed is great, but is there anyway to manipulate reaction time and accuracy, and see how those effect V, Z, A etc?

Thanks so much!!

[Thomas Wiecki]

There is not although I must say that that would be very useful!