1. Brain-computer communication: Self-regulation of slow cortical potentials for verbal communication 
Andrea Kübler, PhD, Nicola Neumann, DiplPsych, Jochen Kaiser, PhD, Boris Kotchoubey, PhD, Thilo Hinterberger, PhD, Niels P. Birbaumer, PhD 
Archives of Physical Medicine and Rehabilitation 
Volume 82, Issue 11, Pages (November 2001)
DOI: /apmr
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

2. Fig. 1
The Thought Translation Device set-up. The patient sat in front of the computer notebook that provided him with feedback of the slow cortical potentials. The EEG was recorded at the vertex. An 8-channel EEG amplifier transferred the signal to a computer equipped with an analog-to-digital card. The notebook's output signal produced written communication, letter by letter, on screen.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

3. Fig. 2
Notebook screen display. Horizontal cursor movement started on the left margin of the screen. During the active phase, the cursor moved to the right margin. The patients' task was to keep the cursor above or below the midline of the screen.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

4. Fig. 3
Trial structure. A trial's duration varied from 4 to 6 seconds. (1) A 2-second passive phase was introduced by a high-pitched tone, signalling to the patient that the cursor could not be moved and that he had to prepare for the active phase. (2) A low-pitched tone introduced the active feedback phase during which the cursor could be moved and feedback was provided. It varied in duration from 2 to 4 seconds, depending on the patients' brain response. (3) Cursor movement could occur during the entire active phase or for only a few hundred milliseconds as indicated. (4) Baseline was recorded 500ms before the active phase began.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

5. Fig. 4
The 5-level matrix for letters and words presented on screen. At level 1, the entire symbol set was subdivided into 2 letter banks, each containing 16 letters, punctuation marks, and a space symbol. The figure shows the division of the first letter bank (E to X) up to the fifth level. If the patient selected that bank, it was again split in 2 and, in the following trial, the patient was presented with a letter bank containing 8 letters (E to P or N to X). If the patient selected a letter bank erroneously, he could reject the 2 letter banks presented at this level and select the “return” function to return to the previous level. Letter banks were split in 2 after each selection until, at the fifth level, the letter bank contained only 1 letter for selection. The second letter bank at the first level (I to ·) was split in the same way as the E to X letter bank. Level 1 contained a “delete” option to erase the last selected letter. Letters were arranged according to their frequency in the German language. Abbreviation: r, return.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

6. Fig. 5
Slow cortical potentials (top panels) and vEOG curves (bottom panels) averaged across about 200 trials conducted at (A) start of training and (B) end of training. Bold lines show cortical negativity condition; thin lines, cortical positivity condition.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions

7. Fig. 6
Learning progress of (A) patient 1 and (B) patient 2. The percentage of correct responses is shown as a function of sessions. (A) Depicts 245 sessions (basic training, 82; error-ignoring training, 25; copy spelling, 127). (B) depicts 250 sessions (basic training, 121; error-ignoring training, 64; copy spelling, 65). Both curves show a significant linear trend.
Archives of Physical Medicine and Rehabilitation  , DOI: ( /apmr )
Copyright © 2001 American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation Terms and Conditions