Fig 1.
(a) Muscle actuation and exoskeleton actuation. Various implementations for exoskeletons have been proposed, such as McKibben artificial muscle [2, 5]. (b) Illustration of perceptual simultaneity between intentional movements and induced movements. Four conditions. Note that in this illustration the range is not precise, and will be psychophysically investigated in this paper.
Fig 2.
The neurocognitive comparator mechanism and our working hypothesis.
(a) A simple illustration of the neurocognitive comparator mechanism with sensing and actuation (based on [7]). A forward model makes predicted sensory feedback based on an efference copy, which is an internal copy of a movement-producing signal. The comparison between predictions and actual sensory feedback results in either a sensory match or a discrepancy. A sensory discrepancy between the predicted and actual sensory feedback results in a lack of SoA and accordingly creates disturbances in motor control. The blue arrows represent external sensing processes and the red arrow represents an actuation process. Where sensing occurs on this pathway depends on the sensing method, i.e., (i) electroencephalogram, (ii) electromyogram, (iii) accelerometer. Brain activity-based sensing is earliest, followed by muscle activity sensing and then motion sensing. (b) Illustration of our working hypothesis.
Fig 3.
(a) System diagram. The system detects the arm movement onset via EMG and position with ACC. (b) Detection-stimulus workflow. S is the detection threshold and t is the controllable interval. (c) Detection and stimulation sequence. The system delay between the detection and stimulation is approximately 50 ms. This total delay is controllable by inserting an additional interval. If the arm returns to a resting position and remains still for 1500 ms, subsequent EMG detection is enabled.
Fig 4.
Example of the EMG and ACC signals.
Blue and red lines indicate movement detection and the stimulation onset, respectively. The interval is defined as the DSI. Note that the EMG signal goes off scale after the stimulation and then returns to a stable state.
Fig 5.
(a) Participants are required to make small movements. (b) With EMS, the movements are amplified to larger movements. (c) and (d) show an example of where the electrodes and ACC sensor were placed.
Fig 6.
The timing of the random stimuli varies for each trial. To get participants to focus only on their movement intention and the timing of the EMS, the “0” digit was not presented. EMS was induced with a delay randomly chosen from a set of 11 values between −200 ms and +200 ms, before and after the instructed timing.
Fig 7.
Results in Experiment 1 (n = 10).
(a) Grand average of response rates in Experiment 1 (n = 10). Each point indicates the middle of a bin of ±20 ms. The sync rate reached a peak where the DSI was approximately 80–160 ms. The early and late rates intersected near 130 ms (between the “120 ms” and “160 ms” bins). (b) The distribution of detection timing towards instructed timing. The onset detection timings did not deviate from the instructed timing significantly. (c) The resulting DSI histogram. All data consists of trials with stimulation after detection.
Fig 8.
Response rates in Experiment 1.
Data were fit with a probit function to capture the rate of “late” responses to DSI in the task. The crossover point of each line and the horizontal line at 50% rate of “late” responses were taken as the point of subjective equality (PSE). The just noticeable difference (JND) indicates the distance between any two X-axis values where the function crosses 0.50 and 0.75 (or 0.25 and 0.50). The average PSE and JND between “early/sync” and “late” were 228 ms and 69 ms, respectively.
Fig 9.
(a) Three types of stimuli. Note that the detection and stimuli timings vary and different in each trial. (b) The trial structure is the same as in Experiment 1, except that the countdown method was changed to beeping sounds and the simultaneity report was simplified to three scales. (c) Session structure. (d) The data flow of the analysis.
Fig 10.
(a) Responses in Experiment 2 (n = 12). (b) PSE and JND means in Experiment 2. The box is drawn from Q1 to Q3 with a horizontal line used in the middle to denote the median. The PSE obtained from the 50 ms adaptation sessions was significantly smaller than that of the 150 ms sessions. (c) The rate of early responses when the stimulation was followed by detection. (d) The distribution of detection-instruction interval. (e) The resulting DSI histogram. All data consists of trials with stimulation after detection.