D2Box freezes during the second run

[arunmib]

My problem is the D2Box freezes during the second consecutive run. I start the D2box, ARsignal processing and random number generator (or the application for getting signals from an EEG cap) applications. Run a trial. Towards the end of the trial, if I hit the "Suspend" and then the "Resume" button, the application promptly starts a new trial, but the cursor just freezes and doesn't moves at all... This 'freezing' doesn't always happens in the second trial, it sometimes happens in the third trial also...But this freezing doesn't happens if the application successfully hits the target and restarts itself....
I am using version 1.4

[gschalk]

Hi,

The "freezing" could have number of causes. If the whole BCI2000 system locks up, then you have discovered a bug. If the system keeps running (e.g., the EEG display is being updated), then the control signals that are output to the cursor task are simply very small, so that they don't make the cursor move. Because the system itself is quite stable, reason 2 seems much more likely. The reason for the control signals becoming so could be a misconfiguration of the adaptation, so that the UD_A and UD_B values become very small. There is some documentation on this configuration in the BCI2000 Workshop manual (mu rhythm section) that comes with 1.4.

However, I would strongly suggest updating to V2.0. It features tons of improvements throughout (including more sensible parameter names ), and comprehensive documentation on doc.bci2000.org.

I hope this helps.

Gerv

[arunmib]

Thanks for the reply and information . I have another doubt, can I use D2Box application for single axis movement training only. That is lets say, I ignore the x-axis movement, whatever its value may be and set it to be a constant value and just update the y-axis movement value based on the signal processing. Sorry, if this question is so dumb , but I need to clarify this with some one....

[gschalk]

Hi,

Yes, you can use the D2Box task for one-dimensional training. What you need to do is to configure the D2Box task with 2 targets (either up/down, or left/right). If you use targets up and down, you need to configure the MUD matrix to classify control signals for up/down control, set the weights in the MLR matrix (3rd column) to 0 so that the output of the classifier is 0, turn off the adaptation for the X direction (I forget the parameter name that controlled this in V1, but it was in Statistics), and potentially give the cursor movement in x a bias in case you want it to travel across the screen during the trial. In this case, you would set XMean to some negative value (e.g., -5) and XGain to some positive value (e.g, 5).

Again, the best documentation of these issues that we have is in the BCI2000 Workshop manual, and again I highly recommend upgrading to v2 since all of this is comprehensively documented for v2.

The Gerv

[arunmib]

Thanks for the replies.....
I have one more question regarding the TUser::ScaleX and TUser:ScaleY. I noticed that, if I reduce their values (from original 0x7fff) the cursor moves at a faster rate. I need to vary the cursor speed based on few
other external parameters. Is varying these values is the only way to do
it? Also, If you can share with me as how you came up with these
numbers, may be I can use the same logic to vary the cursor speed in my
application. It will be of great help to me.

[gschalk]

You do not have to modify the code. As mentioned before, the X/YGain variables specify how fast the cursor is moving. If adaptation is turned on, then, the system will change X/YMean and X/YGain such that the resulting speed of the cursor movement is equal to X/YPixelsPerSecond. Thus, if adaptation is turned on, you can simply change that parameter to change the output speed of the system.

Again, I strongly suggest upgrading to version v2.0 for which all relevant functionality is documented.

Gerv