Sensory substitution, the use of a sensory modality to assist or replace another one, is a promising method to restore or compensate sensory loss in a context of amputation. The missing sense can be substituted using stretch, haptic, electric, tactile, visual or auditory feedback [
1‐
5]. Research on sensory substitution has particular interest in the prosthetic domain, and especially for individuals with an upper limb amputation [
1,
2,
6‐
12]. However, one of the biggest issues associated with myoelectric prosthesis control is the absence of efficient sensory feedback. This feedback could enable effective closed-loop control for comparison with actual correction based upon visual feedback loop [
8]. The absence of sensory feedback for prosthetic control is highlighted by Peerdeman et al. as one of three main reasons for patients to stop using their prosthesis [
13], together with non-intuitive control and insufficient functionality. Because the recovery of the sensory feedback could have a high impact in daily life usage of the prosthesis, this subject is drawing increasing research attention. Using a non-visual feedback signal to control the prosthesis could be advantageous to liberate one’s visual attention which could be directed toward the interaction with the environment, or other tasks.To address this question, sensory substitution has been studied in different contexts looking at substituting grasp force [
14,
15], joint position [
16,
17], finger force [
18], passive touch [
19] and hand configurations [
3] (see the review of Antfolk et al. for more details [
7]). Using the surface of the skin as the interface for sensory substitution has several advantages due to its sensitivity to various stimuli such as temperature, pressure, distortion and vibration [
19‐
22]. In addition, the skin has the ability to transmit both spatial and temporal information. To stimulate the skin, vibrotactile stimulation is commonly used [
23‐
25]. The advantages of such stimuli are the multiple parameters that may be tuned. A vibrotactile stimulation is often characterized by the amplitude and the frequency of vibrations. Other characteristics such as stimulation duration, body localization and intensity of the stimulation may produce signals that could be perceived as distinct [
26]. This process is emphasized by the topographic innervation of the skin which provides the element to make the skin an excellent interface for different kind of stimulations. The skin of the arm is innervated by 5 different dermatomes emerging from the spinal roots from C5 to T1. These dermatomes are organized in longitudinal bands around the arm. The roots give birth to cutaneous nerves, which innervate different areas of the arm. These neurological landmarks have been evoked by Cody et al. [
27,
28] in their exploration of tactile acuity on different sites in the human upper limb. In this context, accuracy in tactile discrimination is of primary importance. Two studies compared tactile perception of stimulations arranged in longitudinal and transverse orientations in a discrimination task [
4,
27]. In the study of Cody et al.[
27], the tactile discrimination was explored using a single von Frey hair (rounded tip diameter 0.6mm, rating 150mN at the onset of bending). Better localization acuity was found for the transverse axis. In the study of Witteveen [
4], the performance of longitudinal and transversal configurations of vibrors for signaling grasp forces and/or hand aperture by means of vibrotactile stimulation was compared. No significant difference was found between the configurations. However, this study mainly focused on how well people performed the task, but did not provide information about how accurately stimulations were localized. The results of this study completed and confirmed the previous work of Weber and Hamburger on the exploration of tactile stimuli [
28,
29]. To our knowledge these studies [
4,
28,
29] are the only ones comparing such orientation. The exploration of upperlimb sensory characteristics shows that most of the research has been done at the forearm level [
4,
12,
23,
26,
27,
30,
31] and very few at the upper arm level [
32,
33]. Based upon principles of sensory physiology and findings of previous studies [
4,
28,
29], we presume that vibrotactile stimulations at the upper arm level will be better discriminated when provided circumferentially (in a transversal axis) than linearly (in a longitudinal axis). This hypotheis is based of the fact that the stimulations sent with a circumferential orientation of the vibrors will be more likely to activate nerves endings from various dermatomes compared to stimulations provided linearly which may potentially implicate only one dermatome. We address this question in our experiment where vibrotactile discrimination is tested according to four different arrangements of vibror stimulators, involving two different orientations: a linear orientation aligned with the upper arm longitudinal axis and a circular orientation on the upper arm circumference. Aside from the orientation, the number of textcolorbluevibrors and the space occupied by them are important parameters to consider with the aim to build a set-up that could be integrated into a prosthesis. Previous work reports that the discrimination distance for the upper arm is approximately 3cm [
34]. Based on this data, and to test the possible advantage of exploiting the full upper arm surface of subjects, we set two categories of spacings between the center of two vibrors. The first spacing was equal to 3cm and was applied to both orientations. The second spacing was set to be proportional to either the upper arm length or its circumference. This produced inter-vibrors distances longer than 3cm. Combined with the two orientations, these two conditions of spacing created the 4 arrangements tested. In addition to the orientation and spacing between vibrors, the physical characteristics of vibrations may also serve to modulate tactile perception of the signal. The vibrors we chose have been used in numerous studies for their ease of use, small size and low cost [
3,
4,
10,
12,
15,
26,
30,
32].
Interestingly, none of these studies have explored the mechanical characteristics of the produced vibrations, such as frequency, intensity, and waveform shape, nor their influence of the perceived signal itself. Instead, these studies directly investigated the capability provided by the vibrotactile signal to identify level of grasping forces [
3,
10,
15,
30,
32], or amplitude discrimination [
3,
26]. For instance, grasping was feedback either using different locations of vibration on the forearm, or different frequency levels of stimulation [
4,
10,
15,
32]. Duration of stimulation could also vary and were often long, from 1.3 sec for Cipriani et al. [
26] to as long as the object was held [
4]. A specific purpose of our study was to first investigate the influence of some important vibrotactile stimulation parameters on the mechanical characteristics of the vibration produced and on the resulting perception, before using them in a specific task. Among such important stimulation parameters are both the location and the duration of the stimulation [
35]. The smallest duration tested so far in studies using the same type of vibror was about 200ms [
36]. However, it has been reported that durations longer than 200ms are perceived as bothersome, and that stimulus between 50ms and 200ms are preferred [
37]. To produce fast and discrete stimulations and avoid the disadvantages of longer stimulations, we therefore focused on stimulus durations of
60, 100 and 140ms.To explore and understand the effect of different settings of the stimulation on vibrotactile perception of the skin, we investigated combinations of duration and intensity in two discrimination tasks with the idea that each vibror could convey multiple types of information. The first experiment aimed to evaluate which of 4 arrangements of vibrors elicit the best score in a spatial discrimination task on the upper arm of non-amputee subjects. The second experiment explored how the same combinations of duration and intensity of the stimulation influence the level of perceived intensity of the stimulation, which could be rated as absent (0) weak (1), medium (2) or strong (3). In a second phase, the arrangement that elicited the best scores on healthy subjects was specifically tested for spatial discrimination on 7 participants suffering from an amputation to verify the validity of this arrangement.