Touch uses frictional cues to discriminate flat materials

In a forced-choice task, we asked human participants to discriminate by touch alone glass plates from transparent polymethyl methacrylate (PMMA) plastic plates. While the surfaces were flat and did not exhibit geometric features beyond a few tens of nanometres, the materials differed by their molecular structures. They produced similar coefficients of friction and thermal effects were controlled. Most participants performed well above chance and participants with dry fingers discriminated the materials especially well. Current models of tactile surface perception appeal to surface topography and cannot explain our results. A correlation analysis between detailed measurements of the interfacial forces and discrimination performance suggested that the perceptual task depended on the transitory contact phase leading to full slip. This result demonstrates that differences in interfacial mechanics between the finger and a material can be sensed by touch and that the evanescent mechanics that take place before the onset of steady slip have perceptual value.

the nanometre scale.
Supplementary Figure 1. Surface topography of the samples, related to materials and methods (subsection "Materials"). Height samples used during experiment 2 were randomly picked and assessed prior to the experiment: four samples of PMMA and four samples of glass.

Supplementary data 2: Homogeneity of performance across the experiments
The surfaces were shown to be quasi perfectly flat prior the experiment but a major concern during the two experiments was to avoid the influence of potential imperfections. In addition to the control procedure that was implemented during the experiments (See Materials and Methods: General procedure), the homogeneity of the performance across the different plates on which the samples were glued was also tested. Four distinct plates were used in each experiment to present all the possible orders since the plates with Glass -Glass -PMMA and PMMA -PMMA -Glass could be used for presenting different orders of the samples. Thus, an imperfection on a plate could have impacted the responses for either one or two of the possible orders of presentation. In experiment 1, the performance was found to be different depending on the target material (Fig 1a.). We therefore investigated if this bias was due to an undetected imperfection or to a natural difference like, for example, a higher probability of a transient increase in the stickiness (stick and slip) for one of the materials. If the bias was due to an imperfection on a plate, it should have impacted the participants' answers for the scratched plate but repeated-measures ANOVA In experiment 2, performance was similar for glass and PMMA but a non-parametric Friedman test showed a lower performance when the target sample was located in the middle (χ 2 = 11.87, p = 0.0026).
Thus, we tested if there was a discrepancy in the responses that would suggest the bias to be due to an imperfection rather than to the psychophysical criterion of the participants. Repeated-measures ANOVA with Geisser-Greenhouse correction was performed across all the plates where the target sample was at the distal or proximal location and showed no significant difference (F(2.565,28.22) = 2.845, p = 0.063).
Wilcoxon matched-pairs signed rank test was performed between the two cases when the target sample

Supplementary data 3: behavioural parameters in experiment 2
Explorations were constrained during experiment 2 but the participants had a large amount of behavioural freedom particularly in terms of exploration speed, normal force applied, and possible changes of strategy, which could influence the differences in contact mechanics between the two materials ( Supplementary   Fig. 2). Despite the different strategies implemented by the participants, we did not find any influence of normal force (R = 0.03, p = 0.95; Pearson's correlation for n = 10) and mean duration of exploration (R = 0.04, p = 0.92; Pearson's correlation for n = 10). These results confirm the hypothesis that the capacity to discriminate relied on specific properties of the fingertip-surface interaction rather than on an optimal choice of exploring strategy.

Illustration of partial slip
Supplementary Movie 1. Partial Slip, related to the discussion about Contact Mechanics. During an active exploration of a glass surface, a high-speed camera situated under the glass filmed the contact area. The first second of the movie illustrates the mechanics of partial slip described in the 'Contact Mechanics' section of the discussion. We see the peripheral slip spreading non-linearly to the central contact region.
One can also observe that the contact area slides uniformly during full slip, which occurred after approximately one second.