Mean ratings for each body part, split by gender, modality, and target body, are visually displayed in Fig. 1, with descriptive data found in the Supplemental Information.
To analyze the arousal ratings, we first ran a PCA on the 41 rated body parts, statistical details of which are provided in the “Method” section. Three components were extracted. The pattern of loadings across all 41 body parts (see Table S3, Supplemental Information) enabled us to make some heuristic interpretations of the three components. The body parts that loaded most heavily onto the first component were all areas that could be defined as stereotypically arousing and generally directly involved in sexual behavior; these included the genitals, breasts, nipples, etc. Therefore, this component was labeled “sexual.” The second component contained body areas that might be considered sensual but not directly sexual, and commonly involved in sensual touch, foreplay, and massage; these included the head, nape of the neck, shoulders, lips, hands, and fingers. This component was therefore labeled “sensual.” Finally, a third component contained body parts that were generally not considered as typically arousing–these included the elbows, knees, chin, calves, etc., and therefore, this component was labeled “non-arousing.” Only the sensual and sexual components were retained for further analysis.
ANOVA on PCA Components, Assessing Effects of Gender, Modality and Target Body
The two remaining extracted components, for sexual and sensual areas, respectively, were analyzed in separate mixed 2 × 2 × 2 ANOVAs with Target Body (Partner-Body vs. Own-Body) and Modality (Touch vs. Look) as within-subject factors, and gender (man vs. woman) as a between-subjects factor. All statistics are reported with Greenhouse–Geisser correction. For both sexual and sensual areas, two key interactions were revealed; the first, a Gender × Target Body interaction, and the second, a Modality × Target Body interaction. These will be described in turn. All other interactions did not reach effect size criteria to be interpreted, and so will not be discussed further. The full ANOVA results for the sexual areas can be found in Table 2 and for the sensual areas in Table 3.
Table 2 ANOVA results for sexual areas Table 3 ANOVA results for sensual areas For sexual areas, there was a significant Gender × Target Body interaction, F(1, 611) = 621.3, p < .0001, η2p
= .504. Men gave higher arousal ratings to the sexual body parts of a partner than they did to those same parts on their own body, MOwn = − 0.27, 95% CI = [− 0.37, − 0.17], MPartner = 0.57, 95% CI = [0.48, 0.67], t(205) = − 18.0, p < .0001, d = 1.23. Women showed the opposite pattern; they rated sexual areas as more arousing on their own body than on a partner’s body, MOwn = 0.22, 95% CI = [0.14, 0.30], MPartner = − 0.38, 95% CI = [− 0.46, − 0.29], t(406) = 17.8, p < .0001, d = 0.89. For both target bodies, there were significant gender differences; for their own bodies, women gave significantly higher arousal ratings than men, t(611) = 7.47, p < .0001, d = 0.60; for a partner’s body, they gave significantly lower arousal ratings than men, t(611) = − 13.77, p < .0001, d = 1.11. This is illustrated in Fig. 2 (left panel).
There was also a Modality × Target Body interaction, F(1, 611) = 118.9, p < .0001, η2p
= .163; there was a general preference toward the tactile modality for both one’s own body, MTouch = 0.45, 95% CI = [0.40, 0.51], MLook = − 0.34, 95% CI = [− 0.43, − 0.25], t(612) = 20.7, p < .0001, d = .91, and a partner’s body, MTouch = 0.14, 95% CI = [0.07, 0.21], MLook = − 0.25, 95% CI = [− 0.33, − 0.17], t(612) = 16.5, p < .0001, d = .66, but the tactile preference for own body was significantly larger than for partner-body, t(612) = 10.4, p < .0001, d = 0.44. This is illustrated in Fig. 2 (right panel), using raincloud plots to show the distribution of raw data and summary statistics (produced in R using the procedure from Allen, Poggiali, Whitaker, Marshall, & Kievit, 2018).
For sensual areas, there was again a significant Target Body × Gender interaction, F(1, 611) = 111.5, p < .0001, η2p
= .154, but the pattern was strikingly different to that shown by the sexual component. Women gave much higher arousal ratings to sensual areas of their partner’s body than their own body, MOwn = − 0.22, 95% CI = [− 0.28, − 0.15], MPartner = 0.49 95% CI = [0.39, 0.59], t(406) = − 19.0, p < .0001, d = 1.07. In contrast, men showed no substantial differences in arousal for their own and a partner’s sensual areas, MOwn = − 0.31 95% CI = [− 0.41, − 0.21], MPartner = − 0.23, 95% CI = [− 0.35, − 0.11], t(205) = − 2.0, p = .044, d = 0.15. Women rated their partner’s sensual areas as significantly more arousing than did men, t (611) = 8.43, p < .0001, d = 0.68. This pattern is illustrated in Fig. 3 (left panel).
There was also a Target Body × Modality interaction revealed for the sensual component, F(1, 611) = 147.5, p < .0001, η2p
= .194. As with the sexual component, here, touching yielded significantly greater arousal ratings than did looking, for one’s own body MTouch = − 0.02, 95% CI = [− 0.08, 0.04], MLook = − 0.47, 95% CI = [− 0.54, − 0.41], d = 0.71. However, for a partner’s body, there were no modality differences, MTouch = 0.24, 95% CI = [0.16, 0.33], MLook = 0.25, 95% CI = [0.16, 0.34], d = 0.02, suggesting that both touching the partner’s sexual body-parts and looking at them were equally arousing. These results are illustrated in Fig. 3 (right panel).
The following two analyses aimed to test whether there were statistical correspondences between the topographic distribution of erogenous zones on one’s own versus a partner’s body and the topographic distribution of erogenous zones in the tactile versus visual modality.
The “Erogenous Mirror”: The Correlation Between Individuals’ Maps for Their Own versus a Partner’s Body
This analysis investigated whether the map of sexual pleasure across one’s own body matched the map of sexual pleasure across one’s partner’s body. Specifically, we investigated whether individual differences in ratings for our own body parts, i.e., individual idiosyncratic preferences over and above the sample average, could predict those same idiosyncratic preferences for one’s partner’s body parts. In other words, would an individual who has an above-average preference for being touched, e.g., on their toes, also have an above-average preference for touching their partner’s toes?
To investigate this, individual arousal ratings for each of the 41 body parts were de-meaned by subtracting the group average for that body part for their gender. This gave us a score reflecting how each participant’s ratings deviated from the gender-group “norm.” Then, these de-meaned scores were standardized across each participant, providing scores reflecting within-subject idiosyncratic preferences, uncontaminated by general group consensus or differences in overall whole-body arousal levels.
We then investigated whether there were correlations between these idiosyncratic preferences between one’s own body and a partner’s body. Interestingly, for both the touch and look modalities, the group mean correlation coefficients between individual preferences for giving and receiving were significantly greater than zero; for touch, mean r(39) = .33, 95% CI = [0.31, 0.35], t(612) = 36.7, p < .001, and for look, mean r(39) = .30, 95% CI = [0.29, 0.33], t(612) = 28.5, p < .001. The correlation coefficients for touch and look modalities did not substantially differ, as indicated by a pairwise t test which only revealed a small effect, t(612) = 2.10, d = 0.11. This means that individual preferences for certain body parts, over and above general group consensus, were common both to what we experience on our own body and what we experience on the body of our partner, independently of the sensory modality.
To investigate any gender differences in this “erogenous mirroring,” these scores were then entered into a mixed ANOVA, with modality as a repeated measure and gender as a between-subjects factor. This did not reveal any medium or large effects, suggesting self-other correspondence was equivalent for both men and women.
Multimodal Erogenous Zones: The Correlation Between Individuals’ Maps for Touching versus Looking
Using the same individual arousal ratings, as calculated for the previous analysis, we then calculated the correlation for each individual between idiosyncratic arousal preferences for touching and looking modalities. The group mean of these correlation coefficients between looking and touching modalities was significantly greater than zero both for one’s own body, mean r(39) = .35, 95% CI = [0.33, 0.37], t(612) = 35.4, p < .0001, d = 1.46, and a partner’s body, mean r(39) = .56, 95% CI = [0.54, 0.58], t(612) = 65.4, p < .0001, d = 2.67. At the group level, the correlation between visual and tactile modalities was significantly stronger for the partner’s body than for one’s own body, t(612) = 18.93, p < .0001, d = .77. This means that individual preferences for certain body parts, over and above general group consensus, were common both to the tactile and visual modalities, but that this commonality was more marked for one’s ratings of a partner’s body rather than one’s own. As before, gender differences were assessed in a mixed ANOVA with target body as a within-subjects factor, and gender as a between-subjects factor. No main effect or interaction involving gender was present.
The Mutual Pleasure Index: The Correlation Between Individuals’ Specific Partner-Body Maps and the Mean Maps of Partner’s Gender
Finally, we took data from the heterosexual participants only to investigate whether the individual-level “partner-body” preferences of one gender corresponded to the group-level “own-body” preferences of the opposite gender. In other words, did men’s preferences for touching women correlate with where women liked to be touched, and vice versa? To investigate this, we calculated the correlation between each man’s ratings for a female partner’s body parts with the group average of the females’ ratings of their own body parts. We also did the converse, calculating the correlation between each woman’s ratings for a male partners’ body parts with the group average of the males’ ratings of their own body parts. These calculations yielded “Mutual Pleasure” scores, indicating the extent to which each individual’s preferences for the opposite gender’s body corresponded with the opposite gender’s preferences for their own body. This was done for both the touch and look modalities separately. A heat map illustrating the distribution of mutual pleasure scores across the body surface can be found in Fig. 4.
We then entered these correlation coefficients (one score per individual per condition, calculated across body parts) into a mixed ANOVA with modality as a within-subjects factor and gender as a between-subjects factor. Interestingly, this revealed a medium-sized main effect of gender, F(1, 559) = 45.57, η2p
= .075. Men had significantly higher mutual pleasure scores than women, MMen = 0.69, 95% CI = [0.67, 0.72], MWomen = 0.59, 95% CI = [0.57, 0.60], t(559) = 6.75, d = .60 (medium-sized effect). Therefore, men’s preferences for touching or looking at women aligned more closely with women’s own preferences for being touched or looked at than did the converse. The main effect of modality and the Modality × Gender interaction were not sufficiently large effect sizes to be considered important (η2p
= .05 and .03, respectively).
To investigate the effects of any individual difference variables on the Mutual Pleasure Index, scores were averaged across modalities and entered as the dependent variable into a multiple linear regression, carried out separately for men and women. For men, relationship status, age, sexual satisfaction score, and sensuality score were entered as independent predictors. For women, an additional variable coding whether they were taking the contraceptive pill was also included. As this analysis was relatively exploratory, variables were entered using a stepwise entry method. For women, a significant model was identified containing relationship status, satisfaction score, and pill-status as significant, positive and independent predictors of mutual pleasure score. Women in a relationship had higher mutual pleasure scores than those who were single (β = 0.06, 95% CI = [0.02, 0.10], t = 3.05, p = .002). Women taking the contraceptive pill had higher mutual pleasure scores than those who were not (β = 0.05, 95% CI = [0.02, 0.09], t = 2.91, p = .004), and finally, those who rated themselves as more sexually satisfied had higher mutual pleasure scores (β = 0.04, 95% CI = [0.02, 0.07], t = 3.44, p = .001). Overall, the model was significant, r = .31, F(3, 326) = 11.24, p < .001, but explained a relatively small proportion of the variance (r2 = .09, 9% variance explained). For men, no suitable model was identified.
To investigate the existence of a mutual pleasure correspondence in non-heterosexual participants with their own gender, we repeated the first analysis just on those who responded with regard to a same-sex partner (including both homosexuals and bi/pansexuals; total N = 46; 18 women). First, the same ANOVA was run as in the previous analysis, with modality and gender as factors. This did not reveal any significant effects for modality, F(1, 44) = 1.11, p = .298, η2p
= .03, nor gender, F(1, 44) = 0.44, p = .510, η2p
= .01, nor the interaction, F(1, 44) = 0.70, p = .408, η2p
= 0.02. When comparing the mean mutual pleasure score, averaged across touch and vision modalities, between individuals who gave same-sex responses versus heterosexual participants who gave opposite-sex responses, no significant difference was observed, MSame = 0.65, 95% CI = [.62, .69], MOpposite = 0.62, 95% CI = [.61, .64], t (605) = 1.25, p = .213, d = 0.19.