Back to PLS HelpBehavior PLS with 2 covariates for fMRI analysis?
Posted on 10/29/09 18:51:44
Number of posts: 34
Hi.
I looked in the User's Guide but couldn't find a definitive answer to the question of whether or not it is possible to perform Behavior PLS using 2 separate but (potentially) related covariates, accuracy and reaction time (possibly to find brain regions whose activation correlates with accuracy-speed trade-off, for exmaple) on fMRI data. Would this be possible if the 2 covariates are loaded simultaneously (as shown in Figure 18 in the User's Guide), and how should we interpret the results? Any thoughts/advice will be greated appreciated!!
Miyoung
Replies:
Untitled Post
I'm Online
Posted on 10/30/09 11:58:55
Number of posts: 291
You may notice that in the User's Guide, I have mentioned:
" ... This User's Guide intend to show how to use PLSgui program to those users who have already understood the PLS method. If PLS is new to you, please read relevant articles first. ..."
Therefore, you cannot find the answer for your specific question.
However, as far as I know, "accuracy" and "reaction time" can be used as 2 separate behavior measurement, and they can be loaded simultaneously as shown in Figure 18 in User's Guide.
In terms of interpretation:
For brain data related results (e.g. BrainLV etc.), it displays the overall impact of the 2 behavior measurements.
For behavior data related results (e.g. Datamat Correlation Response etc.), it displays the plot for each different behavior measurement.
multiple behavior covariates question - part 2
Posted on 06/07/10 20:40:40
Number of posts: 34
quote:
You may notice that in the User's Guide, I have mentioned:
" ... This User's Guide intend to show how to use PLSgui program to those users who have already understood the PLS method. If PLS is new to you, please read relevant articles first. ..."
Therefore, you cannot find the answer for your specific question.
However, as far as I know, "accuracy" and "reaction time" can be used as 2 separate behavior measurement, and they can be loaded simultaneously as shown in Figure 18 in User's Guide.
In terms of interpretation:
For brain data related results (e.g. BrainLV etc.), it displays the overall impact of the 2 behavior measurements.
For behavior data related results (e.g. Datamat Correlation Response etc.), it displays the plot for each different behavior measurement.
Hi,
I've asked a question before related to setting up Behavior PLS for multiple behavioral covariates analysis and I have a new question now. I have accuracy (either in percentage or in decimals) and RT (in ms) that I'd like to load them simultaneously. Do they have to be on the same scale in order to do that, given accuracy is from 50 to 100 (or .5 to 1.0) and RT can run from roughly 200 ms out to 2000 ms? Please advise - thanks in advance!
Miyoung
You may notice that in the User's Guide, I have mentioned:
" ... This User's Guide intend to show how to use PLSgui program to those users who have already understood the PLS method. If PLS is new to you, please read relevant articles first. ..."
Therefore, you cannot find the answer for your specific question.
However, as far as I know, "accuracy" and "reaction time" can be used as 2 separate behavior measurement, and they can be loaded simultaneously as shown in Figure 18 in User's Guide.
In terms of interpretation:
For brain data related results (e.g. BrainLV etc.), it displays the overall impact of the 2 behavior measurements.
For behavior data related results (e.g. Datamat Correlation Response etc.), it displays the plot for each different behavior measurement.
Untitled Post
I'm Online
Posted on 06/07/10 21:01:54
Number of posts: 291
I don't think that you need to put ACCURACY & RT in the same scale range. However, for all ACCURACYs, they have to be in the same scale. i.e. either from 50 to 100, or from .5 to 1.0. Same for all RTs.
have trouble loading >1 behavior text file?
Posted on 06/08/10 14:54:17
Number of posts: 34
Make sense - thanks Jimmy! But now I can't seem to load 2 behavior files simultaneously (unlike the picture in the User Guide). I loaded the first behavior text file (eg. accuracy) and it appears in the Edit Behavior Data window. Then I load the second behavior text file (eg. RT) and it replaces the first file - have I missed something?..... Please advise - thanks!
Miyoung
Untitled Post
I'm Online
Posted on 06/08/10 15:11:13
Number of posts: 291
Method1:
- Open "Edit Behavior Data" window;
- Type the 1st Accuracy & RT in the first line of the "Behavior Data" edit area (the biggest white window);
- Then, type the 2nd, 3rd, ... line by line;
- Now you have N lines of 2-column data. The first column is Accuracy, and the second column is RT;
- Then, type string Accuracy RT in the "Behavior Name" edit field.
- Click "Save" to save your behavior data into a text file;
- Click "OK" to close "Edit Behavior Data" window.
- If you have Excel, you can copy and paste Accuracy & RT into 2 Excel column of the current worksheet. Please only put numerical value, without head line;
- Save the current worksheet as "Text (Tab delimited)(*.txt)" type of format;
- Open "Edit Behavior Data" window;
- Click "Load" to load the text file that you just saved in step 2;
- Now you have N lines of 2-column data. The first column is Accuracy, and the second column is RT;
- Then, type string Accuracy RT in the "Behavior Name" edit field.
- Click "OK" to close "Edit Behavior Data" window.
Untitled Post
Posted on 06/08/10 15:23:28
Number of posts: 34
quote:
Method1:
Thanks for the step-by-step directions Jimmy!! I had the 2 behavior files as 2 separate Text files and tried to load them simultaneously......I will definitely try Method2 (seems to be the easier of the two) :) Thanks once again for your help and time!!
-Miyoung
Method1:
- Open "Edit Behavior Data" window;
- Type the 1st Accuracy & RT in the first line of the "Behavior Data" edit area (the biggest white window);
- Then, type the 2nd, 3rd, ... line by line;
- Now you have N lines of 2-column data. The first column is Accuracy, and the second column is RT;
- Then, type string Accuracy RT in the "Behavior Name" edit field.
- Click "Save" to save your behavior data into a text file;
- Click "OK" to close "Edit Behavior Data" window.
- If you have Excel, you can copy and paste Accuracy & RT into 2 Excel column of the current worksheet. Please only put numerical value, without head line;
- Save the current worksheet as "Text (Tab delimited)(*.txt)" type of format;
- Open "Edit Behavior Data" window;
- Click "Load" to load the text file that you just saved in step 2;
- Now you have N lines of 2-column data. The first column is Accuracy, and the second column is RT;
- Then, type string Accuracy RT in the "Behavior Name" edit field.
- Click "OK" to close "Edit Behavior Data" window.
help interpreting results of multiple covariate behavior PLS analysis
Posted on 06/10/10 14:26:39
Number of posts: 34
Hi Jimmy,
I ran the multiple covariate behavior PLS that used accuracy and RT values (so 2 behavior columns like you suggested) for 2 groups and included 2 conditions to look at and I ended up with an LV profile that has 2 bars for each of the 2 conditions for each group. Now I'm trying to figure out which bar corresponds to which behavior covariate – if I loaded accuracy on the left column and RT on the right, do the bars show the following? (colon represents separation of bars) -
Group1/condition1/accuracy: Group1/condition1/RT: Group1/condition2/accuracy: Group1/condition2/RT:: Group2/condition1/accuracy: Group2/condition1/RT: Group2/condition2/accuracy: Group2/condition2/RT
Thanks for your help in advance!
-Miyoung
Untitled Post
I'm Online
Posted on 06/10/10 14:53:13
Number of posts: 291
I am assuming that you are talking about the "Correlation Overview Plot" window in "Brain scores plot for behavior analysis".
You are correct!
The "Groups" are labeled on X axis, and the "Conditions" are displayed with different colors that are indicated by the legend. The only thing that is not shown explicitly is the name of "behavior measurement". It is sequentially ordered from behavior 1 (the left-most column of your behavior data), to behavior END (the right-most column of your behavior data).
You are correct!
The "Groups" are labeled on X axis, and the "Conditions" are displayed with different colors that are indicated by the legend. The only thing that is not shown explicitly is the name of "behavior measurement". It is sequentially ordered from behavior 1 (the left-most column of your behavior data), to behavior END (the right-most column of your behavior data).
Untitled Post
Posted on 06/10/10 14:57:13
Number of posts: 34
Thank you so much once again for your time Jimmy! Cheers! :)
Miyoung
Login to reply to this topic.
