The first time a subject is run using the RJB protocol, during the first trial of the first run only, the cursor moves to the top of the screen and does not respond to use input. BCI2000 outputs increasilgy large negative numbers during this run.
This is not so much a bug as a side effect of the way the adaptation mechanism works. Quoting Gerv's email to me:
"The control signal is calculated (among other things) as the output value
from MUD minus the offset (UD_A) times the gain (UD_B). Now, depending on what UD_A and UD_B are set at when you start the system, the cursor will either go up or down. As an example, if the offset is set to zero, and the output of MUD is a large positive number, the cursor will jump right to the top. At the end of the first trial, the adaptation will have started to do its thing, and the cursor will start to go straight. What we do is to save parameter files at the end of each session, so that for the next session all of the adapting values have roughly good starting points."
This explains why the cursor moves to the top at first. Since the adaptation mechanism hasn't been trained, all it has are the output values of the MUD filters. In any case, the fact remains that the very first trial of the first run of the first session should be ignored for calculating user performance, as this trial is not affected by the user's brain activity. After the first trial, the adaptation mechanism has something to utilize.
Known Issues and Problems with BCI2000
1 post • Page 1 of 1
Who is online
Users browsing this forum: Bing [Bot] and 1 guest