*2.2. Experiment*

The participants were seated in a chair which is adjustable for comfortable height and were instructed to hold the arm position with the elbow in 90◦ of flexion and the shoulder in 45◦ of abduction during the data collection (Figure 1A). A Velcro strap was used to restrain the shoulder and trunk from moving during the experiment. The wrist and forearm were immobilized in a handmade fiberglass cast and placed on a fixed platform. The wrist joint was restricted inside a ring interface, which was mounted to the platform. The ring interface was connected to a load cell (ATI, Apex, NC, USA). Force signals were recorded with a sampling frequency of 2 kHz and digitized by a BNC-2090A data acquisition board (National Instruments, Austin, TX, USA). The fiberglass cast helped to fix the upper limb well and minimize the movement and variation between subjects.

**Figure 1.** Experimental setup. (**A**) Force display and EMG recording; (**B Figure 1.** Experimental setup. (**A**) Force display and EMG recording; (**B**) The linear electrode array used for surface EMG recording; (**C**) The placement of the linear electrode array on BB muscle belly; (**D**) The fixation of the linear electrode array.

‐

‐

‐ ‐

‐

‐

‐ ‐

‐ ‐

The maximum voluntary contraction (MVC) of the weaker side of the SCI subjects (determined by self-report and clinical assessment) and the non-dominant side of the control subjects was determined. Each subject conducted three MVC trials of the BB muscle. The largest one was defined as the MVC value. The target force of 10–80% MVC (in 10% MVC increment) was marked as a circle with a line connecting the circle and the center of the computer screen (representing the rest state) (Figure 1A). For each desired force level, there was a computer-generated cursor tracking the force in real time. The subject was asked to move the cursor to follow the force line. When the target circle was reached (indicated by color change in the cursor), the subject was instructed to keep the cursor as stable as possible inside the target circle. In the case of having difficulty in reaching the target (especially for the high force level tasks), the subject was verbally encouraged to control the cursor as close as possible to the target. The whole process (moving the cursor to the target and holding the cursor) lasted for at least 10 s. Each subject was allowed to perform practice trials to become familiar with the contraction task before data recording. The sequence of different muscle contraction levels was randomized. Each muscle contraction level was repeated twice. The subjects were explicitly instructed not to change or move trunk position during task performance. To avoid mental or muscle fatigue, subjects were allowed to have at least 2 min break between trials.

Surface EMG was captured from the BB muscle by a linear electrode array designed and manufactured in our lab. The array has 20 silver bars, with each bar being 10 mm in length and 1 mm in width. The inter-bar distance is 5 mm (Figure 1B). Skin preparation was performed with sandpaper, alcohol pads, and conductive gel. The array was positioned over the midline of the BB muscle longitudinally from the bicipital groove to the biceps tendon insertion (Figure 1C). Such placement ensured that the electrode array covered the major portion of the muscle. In addition, self-adhesive cuff was used to wrap the linear electrode array and secure a good attachment on the skin surface of the BB muscle during the experiment (Figure 1D). The reference electrode was attached on the lateral condyle of the subject's tested arm. Surface EMG signals were recorded via the Porti EMG acquisition system (TMS International, Oldenzaal, The Netherlands). The sampling frequency was 2 kHz per channel. There is a 1st order low pass filter before the ADC with a −3 dB point at 4.8 kHz. The ADC of the Porti has a digital sinc3 filter with a cutoff frequency of 0.27× sample frequency.
