search

AGhaderi / hDDM_attention

The current repository is able to assess the relationship between EEG components and HDDM parameters of top-down attention in perceptual decision-making using a multiple regression model
13 stars 4 forks source link

Question about stimulus coding vs accuracy coding in HDDM? #1

Open hcp4715 opened 3 years ago

hcp4715 commented 3 years ago

Dear authors,

Thank very much for your open practices. Scripts here are very helpful for me to understand your study. After reading your manuscript on biorXiv, I have a question about one particular detail of the hddm modelling:

In the last sentence of 1.0.3 Cognitive modeling and model comparison in the manuscript, you wrote that

In this work, the stimulus-based approach is employed.

However, functions used in models_utils.py are hddm.HDDM(), instead of hddm.HDDMStimCoding(). I am a bit confused, because the tutorial of HDDM suggests that stimulus-coding approach should use hddm.HDDMStimCoding().

AGhaderi commented 3 years ago

Dear authors,

Thank very much for your open practices. Scripts here are very helpful for me to understand your study. After reading your manuscript on biorXiv, I have a question about one particular detail of the hddm modelling:

In the last sentence of 1.0.3 Cognitive modeling and model comparison in the manuscript, you wrote that

In this work, the stimulus-based approach is employed.

However, functions used in models_utils.py are hddm.HDDM(), instead of hddm.HDDMStimCoding(). I am a bit confused, because the tutorial of HDDM suggests that stimulus-coding approach should use hddm.HDDMStimCoding().

Dear colleague,

Thanks for your meticulous point and your interest in our research.

As we explained in the preprint version, there are two general implementations for drift-diffusion model, accuracy-based and stimulus-based. These two approaches were used even before HDDM, for example fast-dm and others. Thus, the question is how the stimulus-based approach has been implemented in previous packages. However, we utilize different jargons of accuracy-based and stimulus-based instead of "accuracy coded" and "stimulus-encoding" which have been used in HDDM. In this approach, we assumed that stimuli are similar to conditions, therefore we could make it free in our implementations. This kind of implementation is more convenient and practical for the authors.

Kind regards

hcp4715 commented 3 years ago

Hi, @AGhaderi

Thx for your quick feedback, now I see your point.

Best, Chuan-Peng