Has Anyone Used A Drift Diffusion Model?

[Therapist4Chnge]

---

Members don't see this ad.

In the most recent Neuropsychology there is a really interesting article looking at ADHD effects in response speed an variability (Karalunas et al., 2012). They used DDM and explained the basics of the approach, but I'm wonder people's thoughts of it. Ollie, JS, and other stat/cog psych fans? A quick check on Dr. Google reveals an "intro" paper by an ASU prof that made my head hurt. 😉

 

[Ollie123]

Not familiar with it (addictions is usually pretty far behind the curve), but just tracked down the article. At a glance it seems relevant to some points I've been trying to get through to people in my area for years now (i.e. that the traditional "Average RT across trials" approach is an overly simplistic view of a rich behavioral dataset) but have yet to actually make the effort to write up. I'll go through the article in more depth soon and let you know.

 

[AcronymAllergy]

Haven't yet had a chance to check out this article, but Castellanos et al. have done some different analyses of RT in ADHD beyond the typical RT measures (i.e., average RT, RT block change, and RT standard deviation).

 

[Ollie123]

This is interesting stuff. Thanks for giving me another thing I have to find time to learn I have oodles of data this could be applied to and it may be useful for my dissertation too so this could actually be quite useful.

The details elude me, but this definitely seems to fall under the "Computational modeling" umbrella, so there are plenty of books on the topic. I've had this on my shelf for awhile and am trying to find time to go through it but have not done so yet. What looks like the original paper (Ratcliff & Rouder 1998) gives a bit more detail without going into that highly technical ASU overview that I didn't understand a lick of either. It also looks like there is software to automate extraction of the parameters here. My experiences with these things is that it works much like other stats, with a lot of people trusting in the algorithms and not really understanding the nuances. As much as I dislike it and am fighting it, I'm growing to realize that approach is kind of necessary to be successful in this field.

 

[AcronymAllergy]

This is interesting stuff. Thanks for giving me another thing I have to find time to learn I have oodles of data this could be applied to and it may be useful for my dissertation too so this could actually be quite useful.

The details elude me, but this definitely seems to fall under the "Computational modeling" umbrella, so there are plenty of books on the topic. I've had this on my shelf for awhile and am trying to find time to go through it but have not done so yet. What looks like the original paper (Ratcliff & Rouder 1998) gives a bit more detail without going into that highly technical ASU overview that I didn't understand a lick of either. It also looks like there is software to automate extraction of the parameters here. My experiences with these things is that it works much like other stats, with a lot of people trusting in the algorithms and not really understanding the nuances. As much as I dislike it and am fighting it, I'm growing to realize that approach is kind of necessary to be successful in this field.

Sadly, we can't hope to know everything about everything (a realization I constantly struggle to accept as well).

And now I'm probably going to go search for this paper as a means of distracting myself from what I really should be doing right now.

Edit: Just noticed that in one of the author's other recent works on ADHD, they discuss graph theory, so there seems to be a trend of "computational modeling" analyses in these labs.

 

[Therapist4Chnge]

The paper I referenced talked about the applicability of DDM to ADHD research because it ties in well with current theories of the underpinnings of ADHD. I have been kicking around some ideas for developing a new attentional assessment measure, but only recently have I started to consider how I'd improve instead of just replicate certain measures and constructs. DDM may address some of the shortcomings of existing assessments, though it is still a tall hill to climb to actually evolve how we measure what we proport to measure w. attentional tasks.

*snip, too off topic*

The 1998 article is my next read, as that seems to be Square One. Any cognitive psych folks use DDM, as the authors mentioned the natural fit?

 

[paramour]

In the most recent Neuropsychology there is a really interesting article looking at ADHD effects in response speed an variability (Karalunas et al., 2012). They used DDM and explained the basics of the approach, but I'm wonder people's thoughts of it. Ollie, JS, and other stat/cog psych fans? A quick check on Dr. Google reveals an "intro" paper by an ASU prof that made my head hurt. 😉

Seems to be going around. Too bad I didn't check this out before I went in for my latest dissertation meeting earlier today. Perhaps I could have thrown this in and given my poor advisor even more of a headache. 😛

 

[Therapist4Chnge]

This is interesting stuff. Thanks for giving me another thing I have to find time to learn I have oodles of data this could be applied to and it may be useful for my dissertation too so this could actually be useful....

If nothing else it gives you a chance to look at your data from another view. I am intrigued enough to do more reading despite the piles of other work I need to do. Or maybe it is bc of said pile of work...

 

[Therapist4Chnge]

Edit: Just noticed that in one of the author's other recent works on ADHD, they discuss graph theory, so there seems to be a trend of "computational modeling" analyses in these labs.

Any idea of a good primer (journal or book) for someone looking to learn a bit more about computational modeling, or is that akin to asking to be voluntarily beaten about the head with a blunt object for no good reason? I can take courses for free through my uni, but I am hesitant to jump into any grad class without really brushing up on my stats and programming knowledge. I figure reading a few chapters will let me know quickly if this was a dumb idea.

Amazon recommends this book: http://www.amazon.com/Computational-Modeling-Cognition-Principles-Practice/dp/1412970768

 

[AcronymAllergy]

Any idea of a good primer (journal or book) for someone looking to learn a bit more about computational modeling, or is that akin to asking to be voluntarily beaten about the head with a blunt object for no good reason? I can take courses for free through my uni, but I am hesitant to jump into any grad class without really brushing up on my stats and programming knowledge. I figure reading a few chapters will let me know quickly if this was a dumb idea.

Amazon recommends this book: http://www.amazon.com/Computational-Modeling-Cognition-Principles-Practice/dp/1412970768

I'll check in with a friend of mine who got his doctorate in higher-level math (ala graph theory) and see if he has any recommendations.

 

[Therapist4Chnge]

I'll check in with a friend of mine who got his doctorate in higher-level math (ala graph theory) and see if he has any recommendations.

Much appreciated!

 

[Ollie123]

Sure...credit amazon for the find and not the link I gave you to the same book a few posts up🙂 Haven't gone through it in full yet, but the first chapter is good! It will give you broad strokes though - perhaps enough to read the literature and "kinda get it", certainly not enough to do one on your own without herculean effort.

You'll want to get into Bayesian stats too, as that provides a foundation for many of the modeling approaches. I've heard good things about this as well: click here but can make no guarantees. Computational neuroscience is "slightly" different, but these are new and developing fields so I imagine much of the methodology is comparable.

Being the giant computer geek that I am, I'm actually really interested in the area and regret being totally unaware of it when I was applying to grad school as I might have applied to places that offered it. If I can figure out a way to make it happen, I would love love love to find a post-doc where I could get training in it.

 

[Therapist4Chnge]

Sure...credit amazon for the find and not the link I gave you to the same book a few posts up🙂

HAHA! I missed that the first go around. 😀 TY for the recommendations, I'll def. check them out. I enjoy nerding out with this kind of stuff, as it is a nice YING to the YANG of straight clinical work.