Read Aloud the Text Content
This audio was created by Woord's Text to Speech service by content creators from all around the world.
Text Content or SSML code:
The study was divided into two blocks, with a one minute break between blocks. In each block the participant received 10 (5 velocities x 2 intensities) stimuli which were presented in random order. After rating the last stimulus, the vibration motors were removed from the participant’s arm. Finally, the participant completed a demographics questionnaire and was debriefed about the goal of the study. 2.4 Results Separate two-way repeated measures ANOVAs with the independent variables velocity (5 levels), and intensity (2 levels) were conducted with the dependent variables perceived- velocity, continuity, straightness, intensity, and pleasantness. Ratings in both blocks were aggregated. For all post-hoc and pairwise comparisons Bonferroni correction was used. Following [17], ⌘2 values were calculated manually for all F-tests. In accordance with studies into actual touch [18], regression analyses were performed to investigate a potential inverted U-curve shape for the distribution of pleasantness ratings. The independent variable velocity was log10 transformed.The curve fit of a linear regression model (reduced regression model) was tested against the fit of a quadratic regression model (full regression model), with an F-test for significant reduction of the error sum of squares in the full compared to the reduced model [4]. Perceived velocity For perceived velocity (Figure 2A) a significant main e↵ect of velocity was found (Greenhouse-Geisser correction applied, F(1.63, 29.32) = 72.70, p <.001, ⌘2 =.67), and no other main or interaction e↵ects were found. Post-hoc analysis showed that vibrotactile stroking velocities of 0.5 and 1 cm/s were significantly di↵erent from each other at p <.05, and all other velocities di↵ered from each other at p <.001, with higher velocities being rated significantly higher on perceived velocity. These findings indicate that the velocity manipulations of the vibrotactile stroking stimuli were successful. Perceived continuity For perceived continuity (Figure 2B) no significant main- or interaction e↵ects were found. Overall, ratings for perceived continuity had a mean rating of 64.01 (SD = 10.04). This indicates that, overall, perceived continuity was comparable for all stimuli. The average rating for continuity was found to be significantly higher by 14.51 (t(18) = 6.30, p <.001, 95% CI, 9.67 to 19.34) than the median of the scale. The overall rating for continuity for all velocities can be considered acceptable for the purposes of the current study. Perceived straightness For perceived straightness (Figure 2C) a significant main e↵ect of velocity was found (F(4, 72) = 3.22, p <.05, ⌘2 = .06). Posthoc analysis revealed that this e↵ect was explained by a significant di↵erence in perceived straightness between stimuli with a velocity of 0.5 cm/s (M = 57.66, SD = 11.72) and stimuli with a velocity of 10 cm/s (M = 70.11, SD = 12.05). These velocities had, respectively, the minimum and maximum scores for perceived straightness overall. These findings suggest that extremely slow stimuli are perceived as less straight than faster stimuli, but this e↵ect was not structural in the current study. Perceived intensity For perceived intensity (Figure 2D) a significant main effect of velocity was found (Greenhouse-Geisser correction applied, F(2.42, 43.52) = 24.23, p <.001, ⌘2 = .34). Overall, stimuli with a lower velocity were rated higher on perceived intensity. Post-hoc analysis revealed a significant di↵erence (p <.05) between stimuli with a velocity of 0.5 cm/s (M = 71.87, SD = 11.90) and 1 cm/s (M = 66.28, SD = 12.15), and all other velocities. Stimuli with a velocity of 3cm/s (M = 58.37, SD = 11.67) di↵ered significantly (p <.05) from all other stimuli except those of 10 cm/s (M = 52.08, SD = 11.01). Finally, the di↵erence between stimuli with a velocity of 10 cm/s and 30 cm/s (M = 43.32, SD = 19.52) was not significant. Furthermore, a significant main e↵ect of intensity (F(1, 18) = 28.50, p <.001, ⌘2 = .21) was found. High intensity stimuli (M = 63.84, SD = 8.66) were rated as significantly more intense than low intensity stimuli (M = 52.92, SD = 13.25). These findings indicate that the intensity manipulation was successful. However, velocity also influenced perceptions of intensity. Stimuli with a lower velocity were perceived as more intense than stimuli with a higher velocity, pointing to a temporal summation e↵ect [31], where vibrotactile actuators that are activated for a longer duration are perceived as more intense. Perceived pleasantness For perceived pleasantness (Figure 3) a significant main e↵ect for velocity (Greenhouse-Geisser correction applied F(1.78, 32.02) = 3.47, p <.05, ⌘2 = .11) was found. Post-hoc analysis revealed a significant di↵erence (p <.05) between stimuli with a velocity of 0.5 cm/s (M = 50.76, SD = 21.69) and 3 cm/s (M = 63.03, SD = 13.34). Furthermore, a significant main e↵ect of intensity (F(1, 18) = 4.62, p <.05, ⌘2 = .02) was found. Post-hoc analysis (p <.05) revealed that low intensity stimuli (M = 60.52, SD = 11.19) were rated as significantly more pleasant than high intensity stimuli (M = 56.07, SD = 11.76). Finally, a significant interaction e↵ect between velocity and intensity (F(4, 72) = 2.62, p <.05, ⌘2 = .03) was found. Pairwise comparisons revealed no statistically significant di↵erences between the di↵erent velocities for either low or high intensity stimuli. Statistically significant di↵erences were found for pairwise comparisons between low and high intensity stimuli for 1cm/s (F(1, 18) = 5.17, p <.05, ⌘2 = .22) and 10 cm/s (F(1, 18) = 6.54, p <.05, ⌘2 = .27). For both these velocities low intensity stimuli were rated significantly more pleasant than high intensity stimuli. Polynomial contrasts revealed a significant e↵ect of a quadratic trend (F(1,18) = 9.19, p <.01, ⌘2 = .18) for velocity, indicating a curved relation between vibrotactile stroking velocity and pleasantness ratings. From Figure 3 it can be observed that the pleasantness ratings for low intensity vibrotactile stroking, but not for high intensity vibrotactile stroking seem to form an inverted U-curve. Indeed, an F-tests comparing the fit of a linear regression model to the fit of a quadratic regression model for low intensity stimuli (F(2, 92) = 5.44, p <.01, R2 = .11) showed a better fit for a negative quadratic model. This was not the case for high intensity stimuli (F(2, 92) = 1.70, p >.05, R2 = .04). The peak of the fitted negative quadratic curve for low intensity vibrotactile stroking stimuli was at 6.41 cm/s. 3 Conclusions and discussion The results of the current study show that the Tactile Brush algorithm [14] is suitable for generating vibrotactile stroking stimuli of di↵erent velocities, and intensities. Overall, stimuli of di↵erent velocities were indeed found to di↵er significantly on perceived velocity. Perceived continuity of the stimuli was comparable for all velocities, and non-structural di↵erences were found for perceived straightness of the stimuli. Although it would be expected that slow moving stimuli are perceived as less straight [16], such an e↵ect did not appear consistently for all stimuli. Furthermore, the actual intensity of the stimuli influenced perceptions of perceived intensity. In addition, a temporal summation e↵ect was found [31], where stimuli that involved prolonged activation of individual actuators (i.e. low velocity stimuli) resulted in higher ratings for perceived intensity. One way to further investigate this e↵ect, would be to conduct a similar study in which all stimuli are equal in duration, by repeating (i.e. backwards-forwards stroking) faster stimuli. Another possibility would be to add actuators to the array, decreasing the spatial distance between them, and thus decreasing the duration that each actuator is active. Finally, the e↵ects of intensity could also be investigated more structurally using linear actuators for which the vibration frequency and amplitude can be independently controlled. With regard to the perceived pleasantness of the stimuli, the results show that, overall, the relation between the velocity of the vibrotactile stroking stimuli and pleasantness ratings follow an inverted U-curve. After further examination of this e↵ect, following methods used in studies on actual a↵ective touch [18], a significant fit for a negative quadratic curve was found for low intensity vibrotactile stroking stimuli, but not for high intensity stimuli. It therefore seems that the suggestion that a↵ective aspects of touch are most strongly related to gentle stroking touches [2, 21], also applies to vibrotactile stroking. Moreover, the peak of the fitted negative quadratic curve for low intensity vibrotactile stroking was at 6.41 cm/s, which is within the optimum velocity range of 1-10 cm/s, which is rated most pleasant for actual stroking touches [18]. Considering the fact that vibrotactile actuators do not stimulate CT a↵erents [2], it is plausible that the velocity information of vibrotactile stroking, through a more cognitively involved process, possibly anchored in one’s own perceptual experience with actual stroking touches [19, 21], serves as an a↵ective cue for the interpretation of the tactile sensation.