Participants
A total of 20 healthy right-handed women (age [mean ± standard deviation]: 33.4±3.5 years) participated in the present study. All recruited participants were regular users of one of the three face serums of famous luxury brands (A, B, and C), consuming them more than three times per week. These cosmetics were all within the same price-range (A ~ ¥13,500/50 mL; B ~ ¥13,500/60 mL; C ~ ¥13,000/50 mL). The number of users of each serum was similar (7 participants for A, 7 for B, and 6 for C), and more than five bottles of the serum (5.2 ± 2.0 bottles) were used at the time of the fMRI experiment. They had no history of neurological or psychiatric disorders and provided written and oral informed consent to participate in the present study. The Research Ethics Committee of the Shiseido Global Innovation Center approved the present study, and all the experiments were conducted in accordance with the approved guidelines.
Stimuli, trial protocol, and procedure
We used three types of face serums (A, B, and C) that had different textures and were in bottles with different visual appearances; each participant could clearly discriminate her regular serum from the others. The fMRI experiment consisted of two sessions for every participant with two conditions per session: (i) a visual session consisting of the regular-use serum (the participant's preferred cosmetic) and the control (the other non-preferred cosmetics) conditions, repeated twice per condition, and (ii) a visual with tactile session consisting of the preferred cosmetic conditions and the non-preferred cosmetic conditions, twice per condition. Each stimulus was presented for 30 s (task block) with a 30 s interval (rest block). The stimulus presentation order was counterbalanced across participants. In the visual session, participants viewed the photo of a face serum bottle as stimulus in the MRI scanner using goggles that allowed the photos to be projected. In the visual with tactile session, participants were applied with the face serum on the back of their left hands, while they viewed the photo of the face serum bottle. The amount of serum applied was 0.2 mL per task block. A beauty specialist applied the serum on the back of the participant's hand using the fingers and palm of her right hand. She applied it to the skin and moved her hand in a circle, slowly and softly. The speed of movement was approximately 2.5 s per cycle. The application procedure was the same for all task blocks. During the rest block, she removed the serum from the participants' hand using a warm wet towel.
Subjective evaluations
After the fMRI experiment, outside the MRI scanner, participants evaluated all the face serums on a visual analog scale (from 0 to 100 points) of six items: "positive feeling of texture," "expectation for skincare effect," "want to buy," "security," "satisfaction," and "attachment" after being applied with the serum in the same manner as they had in the scanner. The average scores of each subjective evaluation were compared between conditions using a paired t-test with a significance level of p = 0.01. In addition, we conducted a multiple regression analysis with "attachment" as the dependent variable and the other items as the independent variables, with a significance level of p = 0.05 based on the stepwise method. Furthermore, we checked the residuals for the regression analysis by performing the Shapiro-Wilk test of normality with a significance level of p = 0.05 and calculated the Durbin-Watson statistic for the null hypothesis of no autocorrelation. Statistical analysis was carried out using SPSS software, version 21.0 (SPSS, Inc., Chicago, IL).
fMRI data analysis
Scanning was conducted with a 3.0T MRI system (Achieva Quasar Dual; Philips Medical Systems, Best, the Netherlands). Blood oxygenation level dependent T2*-weighted MR signals were measured with a gradient echo-planar imaging (EPI) sequence (repetition time [TR] = 3,000 ms, echo time [TE] = 35 ms, flip angle [FA] = 90°, field of view [FOV] = 230 × 230 mm2, scan matrix = 128 × 128, total scan time = 492 s, dynamic scans = 164 volumes, slice thickness = 5 mm, and 23 slices per volume). Image processing was conducted using the statistical parametric mapping software (SPM8, Wellcome Department of Imaging Neuroscience, London, United Kingdom; http://www.fil.ion.ucl.ac.uk/spm/software/spm8). T1-weighted anatomical images were acquired (150 slices, thickness 1.0 mm, TE = 2.0 ms, TR = 23 ms, FOV = 240 × 240 mm2, FA = 30°, and matrix size = 240 × 240). EPIs were spatially realigned, co-registered, and normalized to the Montreal Neurological Institute template. Normalized images were smoothed using an 8 mm full-width half-maximum Gaussian kernel. The data were temporally convolved with a hemodynamic response function and high-pass filtered with a cutoff period of 128 s. Each condition was modeled using a separate regressor, and the second-level analysis was performed. For the ROI analysis, we set the brain regions that had been identified in the previous fMRI studies of human love attachment in relationships [13-15, 26] (Table 1), as a set of spherical ROIs (radius 5 mm) for the contrast of the preferred cosmetic condition with the non-preferred cosmetic condition in each session, and tested their significance in the differential brain activity between the visual with tactile session and the visual session, by using the SVC test (significance level; p = 0.05, FWE).
Moreover, we conducted a multiple regression analysis with the score of a subjectively evaluated item that was significantly correlated with a feeling of "attachment" as the dependent variable, and the eigenvariate values of the ROIs that were significant by the SVC test as the independent variables, in each of the visual with tactile and visual sessions, with a significance level of p=0.05 based on the step-wise method. In addition, we added the somatosensory-related ROIs {the right SI (40 -31 59), secondary somatosensory area (SII) (52 -24 20), and posterior insula (PI) (42 -31 21), and the left SII (-58 -20 14)} identified in our previous studies [27, 28] to the above ROIs in order to confirm the tactile effect on the somatosensory-related regions. Next, we conducted 2×2 (conditions × sessions) repeated measures analysis of variance for the ROI activity versus baseline in each condition (p < 0.05), and compared their activities between the visual with tactile and the visual sessions using a paired t-test based on a Bonferroni correction (p < 0.025 = 0.05/2) when there was a significant interaction between conditions and sessions. Moreover, we conducted multiple regression analyses with the eigenvariate value of each brainstem region (DRN, PAG and SN/VTA) as the dependent variable, and activities in each of the memory-related ROIs and the somatosensory-related ROIs, based on the step-wise method corrected by Holm method (p < 0.05). Furthermore, we checked the residuals for all regression analyses by performing a Shapiro-Wilk test of normality (significance level; p = 0.05) and calculated the D-W statistic for the null hypothesis of no autocorrelation.
Data availability
Due to the confidentiality agreements with the participants, the data in this study are available only at the Shiseido Global Innovation Center and Tokyo Metropolitan University.