Neural substrates of sexual arousal revisited: Dependent on sex

Challenging previous work (1), a recent metaanalysis suggests “that the neuronal circuitries activated by visual sexual stimuli are independent of biological sex” (ref. 2, p. 15671). Neuroimaging metaanalyses are indispensable for robust conclusions on neuronal correlates of mental functions (3). To attain this goal, best-practice guidelines have been jointly proposed by the developers of all major analytic approaches (3). A careful look at the aforementioned analysis reveals several issues that may have affected the outcome in nontrivial ways.

The authors performed contrast analyses “to combinatorically compare ALE [activation likelihood estimation] datasets in a pairwise manner, in particular with respect to … sex” (ref. 2, p. 15675). According to their appendix and laudably shared dataset, experiments comprising mixed samples (i.e., men and women) might have been assigned to both sex-specific metaanalyses that were eventually compared, obscuring potential sex differences. Another 11% of experiments did not report within-sex brain activations but reported group comparisons, sample size information seemed off for about 25%, and 14% merged multiple contrasts, while, for 26%, additional contrasts were available but not included, all of which may have potentially added further noise.

Given strong recommendations against including region-of-interest (ROI) analyses in metaanalyses due to spatial bias, the observations that almost 40% of the included experiments were ROI-based and that the proportion of ROI analyses differed between sexes (P = 0.014) (2) further complicates the performed analysis. Furthermore, inference relied on false discovery rate correction for multiple comparisons, which is inappropriate for statistical parametric maps (4) and yields both poor sensitivity and specificity for ALE (5). Additionally, the authors applied a (arbitrary and unjustified) cluster threshold of 200 mm³, potentially biasing the results against smaller clusters of convergence, for example, in the hypothalamus.

Conceptually, inclusion of experiments assessing nonheterosexual subjects may add noise in sexual preference-encoding regions (6), further impairing sensitivity for true differences. Finally, more than one-third of the experiments did not primarily address sexual arousal but addressed, for example, resting-state connectivity, sexual disgust (by nonpreferred sexual stimuli), deactivations, or medication effects. Conversely, relevant applicable publications were omitted (e.g., refs. 7–10).

Reanalyzing the provided dataset employing the original statistical parameters reproduced the null result. Using best-practice statistical approaches on the original data, however, we found stronger convergence of neural activity for women in the middle occipital cortex, inferior parietal lobule, and pallidum, while stronger convergence for men was observed in the lateral visual areas, anterior cingulate cortex, caudate, thalamus, and hypothalamus. When additionally addressing the conceptual concerns about study selection and preparation by only including matched pairs of experiments in heterosexual men/women from the updated literature, we observed that, pooled across sexes, sexual arousal entailed a typical pattern of brain activations (Fig. 1). Metaanalytic comparison of within-group experiments revealed significantly stronger convergence of activity in visual areas for women during sexual arousal but in dorsal anterior cingulate and hypothalamus in men (Fig. 2).

Fig. 1.

Functional neuroanatomy of sexual arousal. Significantly converging activity (P < 0.05, cluster-level family-wise error [cFWE] corrected) during visually induced sexual arousal across 32 experiments (559 foci) involving heterosexual men and women. Brain slices shown at Montreal Neurological Institute (MNI) coordinates (x, y, z). Experiments involving one sex (e.g., men) only included if analogous within-group experiment also performed in the other (i.e., women); dACC, dorsal anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; IPL, inferior parietal lobule; L, left; lOFC, lateral orbitofrontal cortex; MCC, middle cingulate cortex; NAcc, nucleus accumbens; PCC, posterior cingulate cortex; pgACC, pregenual anterior cingulate cortex; R, right.

Fig. 2.

Sex differences in convergence of brain activity during visually induced sexual arousal. Comparison of brain activity between heterosexual men and women as revealed by subtraction analyses. Significance threshold set to P > 0.95 posterior probability. Brain slices shown at coordinates (x, y, z) in MNI space.

Taken together, replicating the metaanalysis by Mitricheva et al. (2) using current best-practice recommendations shows a different picture, indicating significant sex dependence of neuronal activity due to sexual arousal. Further investigations focusing on sex differences in the neural substrates of sexual arousal are thus needed to complete the still limited data pool of pertinent studies.