Table 1.
Demographics of Experiment 1 participants.
Fig 1.
A visual illustration of the experimental design and posts of Experiment 1.
Fig 2.
Social intent scores following visual and tactile feedback.
Panel A shows participants’ social intent scores following feedback at high and low support levels. Panel B shows participants’ social intent scores following feedback (high support only) on positive and negative posts. The rainclouds represent the distribution of Social Intent when grouped by Feedback Mode and Support Level (in Panel A) and when grouped by Feedback Mode and Valence (in Panel B). The dots represent the average score per participant for each combination of Feedback Mode and Support Level (in Panel A) and Feedback Mode and Valence (in Panel B). Error bars indicate the 95% confidence interval (CI) around the mean (Mean of Social Intent +/- 95% CI). The dot on the error bar is the sample mean of Social Intent for the respective combination of Feedback Mode and Support Level (in Panel A) and Feedback Mode and Valence (in Panel B). The horizontal dashed line (at y = 0) indicates that the Social Intent scores could not be below 0.
Table 2.
Demographics of Experiment 2 participants.
Fig 3.
A visual illustration of the experimental design and posts of Experiment 2.
In Experiment 2 there were two main measures: Valence (Positive vs. Negative) and Feedback Mode (Visual vs. Visuotactile emoticons). Participants received visual and visuotactile feedback on posts of positive and negative valence. There were 4 possible combinations (e.g., “Valence: Positive, Feedback Mode: Visuotactile Emoticons) and 8 posts per combination (i.e., 48 posts in total; see below). and within each Valence and Feedback Mode combination the number of and type of visual emoticon reacts was varied (low, medium, and high; see below). However, for additional variability within each of the 4 combinations we varied the type of visual emoticon react (2 per intensity: low, medium, high; see below), thus resulting in 2 posts per Valence, Feedback Mode and Emoticon Reach combination. We also varied the number of emoticon reacts (low, medium, and high; see below). The number and type of emoticon reacts were not taken into consideration in the Experiment 2 analyses. The post examples are examples of what a participant would see when logged onto their ‘FaceJournal’ profile and receiving visual feedback. The screen display was similar to that of actual social media platforms on a browser (i.e., with the name of the website and the link on the top left, as well as typical browser functionalities on the top right (e.g., ‘bookmark’). Then, each post was individually shown like in this example. On each trial, participants’ name of choice (like the ‘Jane Doe’ example shown here) and the given icon they had set for their profile (always a round abstract image as the one shown here) were shown at the top left corner of the post. Below their name, participants could see how many minutes ago they posted on their profile (here, 2 minutes ago; all posts across all participants had the same timestamp), with the ‘connect’ logo of the platform next to the timestamp (as indexed by the grey earth-like icon). In the example at the top right corner of the figure, a post of positive valence is shown and thus the visual feedback is a smiley visual emoticon (i.e., a low intensity visual emoticon) with seemingly 24 such reactions (i.e., a number of likes from the low range). To create a more realistic experience of FaceJournal interactions, below the other users’ reactions participants could also see the ‘React’ and ‘Comment’ functionalities social media users also have below their own posts (should they wish to also react on their post and/or comment). The example at the bottom right corner of the figure is a post with the same design features as the top post, but with a sentence of negative valence and thus the emoticon react is a shocked face (here, medium intensity). The specific post was also paired with predetermined visuotactile feedback, thus the participant also received affective touch via the S-CAT at the same time as the visual emoticon appeared on their screen.
Fig 4.
The S-CAT and FaceJournal interface.
(A) Shows the inside materials of the S-CAT. This is the side that contacts with the skin, and the pneumatic actuators covered by the silicon layer on the outside (as shown in the image) were positioned on the participant’s forearm, as it is where touch was being delivered from (for more see [77]). (B) Shows a participant during the session, wearing the S-CAT on their left forearm, and looking at their post on the FaceJournal platform, while receiving visuotactile feedback via the S-CAT and the visual emoticon on the screen. The participant is sitting behind the makeshift wall and cannot see the experimenter sitting behind it, delivering the tactile feedback. (Note. The participant has consented).
Fig 5.
Trial timeline for Experiment 2.
*Feedback was either the total number of reacts and the visual emoticon displayed on the screen for 12 seconds, or a combination of this aforementioned feedback together with tactile feedback sent to the participant via the wearable sleeve for 12 seconds.
Fig 6.
Social intent scores following visual and visuotactile feedback.
Participants’ social intent scores following visual and visuotactile feedback on positive and negative posts. The rainclouds represent the distribution of Perceived Social Intent grouped by Feedback Mode and Valence. The dots represent the average score per participant for each combination of Feedback Mode and Valence. Error bars indicate the 95% confidence interval (CI) around the mean (Mean of Perceived Social Intent +/- 95% CI). The dot on the error bar is the sample mean of Perceived Social Intent for the respective combination of Feedback Mode and Valence. The horizontal dashed line (at y = 0) indicates that the Perceived Social Intent scores could not be below 0.
Fig 7.
Changes in physiological measures following visual and visuotactile feedback.
Heart rate (HR; Panel A) and skin conductance rate (SCR; Panel B) were baseline corrected–in other words, the mean HR and SCR scores from each condition block (e.g., visuotactile feedback on negative posts) were subtracted from the average HR and SCR scores we obtained before the task. Heart Rate Variability (HRV; Panel C) was calculated as the time elapsed between successive inter-beat intervals. The y’y axis on Panel C represents the mean of HRV values obtained during the main task, when participants were receiving feedback on their posts. The rainclouds–in all three panels–represent the distribution of the physiological measure grouped by Feedback Mode and Valence. The dots represent the average score per participant for each combination of Feedback Mode and Valence. Error bars indicate the 95% confidence interval (CI) around the mean (Mean of Baseline Corrected HR in Panel A, Baseline Corrected SCR in Panel B, and HRV in Panel C).