Neuroscience and Biobehavioral Reviews

Transcription

1 Neuroscience and Biobehavioral Reviews 36 (2012) Contents lists available at SciVerse ScienceDirect Neuroscience and Biobehavioral Reviews jou rnal h omepa ge: Review Functional neuroimaging studies of sexual arousal and orgasm in healthy men and women: A review and meta-analysis Serge Stoléru a,b,, Véronique Fonteille a,b, Christel Cornélis c, Christian Joyal c, Virginie Moulier a,b a Inserm, U669, 123 rue de Reuilly, Paris, F-75012, France b Univ Paris-Sud and Univ Paris Descartes, UMR-S0669, Paris cedex 14, France c Département de Psychologie, Université du Québec à Trois-Rivières, 3351, boul. des Forges, C.P. 500, Trois-Rivières, (QC) G9A 5H7, Canada a r t i c l e i n f o Article history: Received 11 June 2011 Received in revised form 3 March 2012 Accepted 14 March 2012 Keywords: Functional neuroimaging Positron emission tomography Functional magnetic resonance imaging Brain Sexual behavior Libido Sexual arousal Sexual desire Erection Orgasm Heterosexuality Homosexuality Visual sexual stimuli Meta-analysis a b s t r a c t In the last fifteen years, functional neuroimaging techniques have been used to investigate the neuroanatomical correlates of sexual arousal in healthy human subjects. In most studies, subjects have been requested to watch visual sexual stimuli and control stimuli. Our review and meta-analysis found that in heterosexual men, sites of cortical activation consistently reported across studies are the lateral occipitotemporal, inferotemporal, parietal, orbitofrontal, medial prefrontal, insular, anterior cingulate, and frontal premotor cortices as well as, for subcortical regions, the amygdalas, claustrum, hypothalamus, caudate nucleus, thalami, cerebellum, and substantia nigra. Heterosexual and gay men show a similar pattern of activation. Visual sexual stimuli activate the amygdalas and thalami more in men than in women. Ejaculation is associated with decreased activation throughout the prefrontal cortex. We present a neurophenomenological model to understand how these multiple regional brain responses could account for the varied facets of the subjective experience of sexual arousal. Further research should shift from passive to active paradigms, focus on functional connectivity and use subliminal presentation of stimuli Elsevier Ltd. All rights reserved. Contents 1. Introduction Methods Search strategy and selection criteria Organization and presentation of results Meta-analysis Brain areas involved in sexual arousal Healthy male volunteers Review Meta-analysis Healthy female volunteers Comparison between healthy men and women Brain areas involved in orgasm Male ejaculation and orgasm Female orgasm Corresponding author at: INSERM Unité 669, 123 rue de Reuilly, Paris, F-75012, France. Tel.: ; fax: addresses: serge.stoleru@inserm.fr, serge.stoleru@free.fr (S. Stoléru), veronique.fonteille@inserm.fr (V. Fonteille), Christel.Cornelis@uqtr.ca (C. Cornélis), Christian.Joyal@uqtr.ca (C. Joyal), virginie.moulier@inserm.fr (V. Moulier) /$ see front matter 2012 Elsevier Ltd. All rights reserved. doi: /j.neubiorev

2 1482 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Time course of brain responses to sexual stimuli A neurophenomenological model of sexual arousal Future directions for research References Introduction About a hundred years ago, S. Freud started to develop a psychological theory where sexual drives played a key role both in healthy individuals and in psychopathological conditions. Although controverted, psychoanalytical theory has drawn the attention of generations of clinical psychologists, psychiatrists and of some prominent neuroscientists (e.g., Kandel, 1999) to the concept of sexual drives. By an instinct is provisionally to be understood the psychical representative of an endosomatic, continuously flowing source of stimulation, as contrasted with a stimulus, which is set up by single excitations coming from without. The concept of instinct is thus one of those lying on the frontier between the mental and the physical. (Freud, 1905). Later, he wrote: The deficiencies in our description would probably vanish if we were already in a position to replace the psychological terms by physiological or chemical ones. [...] Biology is truly a land of unlimited possibilities. We may expect it to give us the most surprising information and we cannot guess what answers it will return in a few dozen years to the questions we have put to it. (Freud, 1920). It may well be that new insights drawn from functional neuroimaging investigations of sexual arousal (SA) a method also lying on the frontier between the mental and the physical support Freud s prediction. SA may be broadly defined as the physical and psychological readiness to perform sexual behavior. Episodes of this psychophysiological state may be triggered by external stimuli or may occur without any apparent external cause. Manifestations of SA may be psychological, e.g., sexual desire, and/or physiological, e.g., genital responses. The highest level of SA reached during any particular episode may vary from a transient desire to maximum level of SA and orgasm. Congruent with the above manifestations, the measurement of the level of SA is based on self-report of psychological manifestations and on objective measurements of physiological responses, e.g., phallometry and vaginal photoplethysmography (Rosen and Beck, 1988). The new frontier of the scientific investigation of human SA is the identification and understanding of its neural correlates. The human brain is involved in all the successive steps of human sexual behavior, from the assessment of the sexual relevance of external stimuli to the control of sexual behavior (Meisel and Sachs, 1994). A thorough study of the neural underpinnings of human SA is important for both theoretical and practical reasons. From a theoretical standpoint, the understanding of the neural correlates of SA should provide insights into human sexual motivation, human reproductive behavior, and about the processing of sexual incentives, which belong to primary reinforcers. From a practical point of view, a better understanding of the neural mechanisms of SA should contribute to solve public health problems such as sexual disorders and sexual offending. Before the development of brain functional imaging techniques, studies of the cerebral basis of human SA relied for a great part on animal models (Bancroft, 1989; Herbert, 1996; Meisel and Sachs, 1994). However, human sexual behavior has unique characteristics, e.g., sexual imagery, that distinguish it from the homologous behavior in other species. Therefore, studies on human beings are needed to characterize the regions of the brain involved in the species-specific aspects of human SA. A second source of knowledge on the brain basis of human SA has been the study of neurological patients, for instance those presenting epileptic seizures with sexual manifestations, or those presenting sexual symptoms associated with focalized or disseminated lesions (Rees et al., 2007). However, brain lesions are rarely restricted to regions of interest. Instead, many patients have diffuse damage resulting for instance from head trauma or stroke, and their lesions often encompass multiple brain regions. In postmortem studies, psychological autopsies are particularly difficult and uncertain in the domain of sexual behavior. Thus, those neurological studies, although useful, have been insufficient to describe the cerebral correlates of SA in healthy individuals. Thirdly, neurosurgery, whether by removing brain lesions or by inadvertently causing them, has provided findings relevant to the understanding of cerebral correlates of human SA (Burns and Swerdlow, 2003; Devinsky et al., 2010; Dieckmann et al., 1988; Freeman, 1973). In the last decade, modern functional neuroimaging techniques have brought major advances in this domain of research for several reasons. Firstly, being minimally invasive, they may be used both in healthy volunteers and in patients with sexual disorders. Secondly, instead of being limited to the study of some specific areas, as are brain lesion studies, they allow for the study of the whole brain. Thirdly, functional neuroimaging techniques can be used to investigate cognitive aspects of sexual behavior. Finally, technological advances have improved both their spatial and temporal resolutions. The axial spatial resolution of the latest positron imaging (PET) devices is about 2.2 mm. Depending on technical characteristics, the spatial resolution of functional magnetic resonance imaging (fmri) scanners is about 1 3 mm. PET and fmri present different temporal resolutions, i.e., about 2 3 s for fmri and about 1 min for [ 15 O] H 2 O PET. Functional neuroimaging techniques have become, for the above reasons, one of the key approaches to understand the brain basis of SA both in healthy subjects and in sexual disorders. At present, 73 published original studies have used these techniques to specify areas that respond to sexual stimuli in healthy human subjects. Thus, this research field is now sufficiently mature for a detailed review to be performed. Hence, the purpose of this study is to review the functional neuroimaging studies of brain regions mediating SA in healthy men and women. Because of space limitations, functional neuroimaging studies of patients presenting sexual disorders, which are an important source of knowledge about the cerebral mechanisms of SA, will be the focus of a separate review. Because this review showed that the consistency of findings of different studies varied across brain regions, findings were subjected to a meta-analysis, a technique used to identify areas of consistent activation (Turkeltaub et al., 2012). 2. Methods 2.1. Search strategy and selection criteria We systematically searched peer-reviewed journals indexed in large databases (PubMed, PsychInfo, Ovid, Embase) for Englishlanguage published manuscripts of single photon emission tomography (SPECT), PET, fmri, functional near-infrared spectroscopy and magnetoencephalography (MEG) studies of SA published between January 1994 and May The databases were searched for the following keywords or expressions: functional magnetic resonance imaging; or positron emission tomography; or magnetoencephalography; or single-photon computed tomography; or near-infrared spectroscopy; or neuroimaging; combined with

3 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Fig. 1. Brain areas showing activation in response to sexual stimuli. Note: Activations are superimposed onto a normalized individual T1-weighted structural scan provided in MRIcro software (Rorden and Brett, 2000). Above each coronal slice, the y-coordinate refers to the MNI space. Foci of activation are automatically placed on the slice closest to their y-coordinate. For each study, the type of sexual stimuli is coded through shape (empty circle: video clips; filled square: photographs; cross: other stimuli) and the neuroimaging technique is color-coded (yellow: fmri; red: PET). Brain regions. (a) lateral occipital and/or lateral temporal cortex; (b) fusiform gyrus; (c) amygdala; (d) inferior parietal cortex; (e) superior parietal cortex; (f) claustrum; (g) orbitofrontal cortex; (h) medial prefrontal cortex; (i) insula; (j) hypothalamus; (k) caudate nucleus; (l) putamen; (m) nucleus accumbens/ventral striatum; (n) thalamus; (o) anterior cingulate cortex; (p) premotor areas; (q) cerebellum; (r) midbrain.

4 1484 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Fig. 2. Activations in anterior cingulate cortex in response to sexual stimuli Notes. Activations are superimposed onto a normalized individual T1-weighted structural scan provided in MRIcro software (Rorden and Brett, 2000). sexual arousal; or sexual desire; or sexual behavior; or erotic; or orgasm; or ejaculation; or sexual disorder; or libido; or erection. Retrieved studies and their references were examined to determine whether they investigated SA or orgasm through one of the above-mentioned neuroimaging methods. Studies cited in the papers retrieved through the databases could be considered in this review if they met all the following criteria: (1) to focus on healthy human beings; (2) to investigate SA and/or orgasm, not solely love feelings; (3) to base identification of regions on explicit probability thresholds. Fifty eight studies (40 fmri, 14 PET, 2 MEG, 1 SPECT and 1 near-infrared spectroscopy studies) were included for review. Nine studies were not included because they did not meet abovementioned criterion No. 1 (n = 6) or No. 3 (n = 3). All regional brain responses are reported in this review if they were considered as statistically significant according to the criteria used in the individual studies. The first study of SA using a functional neuroimaging technique was based on SPECT, a technique with relatively low resolution (Tiihonen et al., 1994). Since then, the most commonly utilized techniques have been PET and fmri. All fmri studies have been based on the Blood Oxygen Level Dependent (BOLD) contrast, except one that relied on perfusion imaging (Georgiadis et al., 2010) Organization and presentation of results We begin by reviewing brain regions that have repeatedly been found associated with SA in heterosexual men. For each region, we discuss which particular function(s) it may serve within the more general process of SA. Not surprisingly, a particular function, e.g., evaluative processes, served by a brain region may in some cases not be specific of sexual behavior, but be an instance of a more general process encountered in nonsexual contexts. Then, we compare results in heterosexual and homosexual men. Then, findings in heterosexual women are reviewed and compared with those in heterosexual men. A specific section focuses on studies of male and female orgasm. As the SA response develops over several minutes, emerging knowledge on the time course of brain responses to sexual stimuli is presented. Finally, we propose a model aiming to account for the accumulated findings in a meaningful manner, both from a neural and a phenomenological standpoint. Because the great majority of studies used visual sexual stimuli (VSS), we do not present studies grouped by induction methods. To illustrate the findings of this review, the standard coordinates of activation peaks reported by individual studies were plotted onto coronal sections of a T1-weighted structural scan of a healthy male subject s brain normalized into the standardized anatomical space provided by the Montreal Neurological Institute (MNI) template (provided in MRIcro software; Rorden and Brett, 2000). Figs. 1 and 2 show activation foci according to contrasts of experimental conditions and to correlational analyses. In order to make direct comparisons across studies, we converted reported Talairach coordinates (Talairach and Tournoux, 1988) into MNI coordinates (transformation developed by M. Brett, Because differences in image smoothing techniques have an effect on the number and extent of activation foci, we did not show the spatial extent of activation foci in the figures. The methodological features of the studies of SA (excluding orgasm) in healthy men and women are reported in Table 1. By far, the most often used experimental paradigm has been the passive viewing of VSS. Both still pictures and film clips have been used, within event-related or block designs. Although in the great majority of studies, scanning started with the presentation of stimuli, in four papers (Bocher et al., 2001; Miyagawa et al., 2007; Redouté et al., 2000; Stoléru et al., 1999), stimulation began before scanning. While most fmri studies used 1.5 Tesla (T) scanners, an increasing number of studies used a 3 T- scanner, and two papers were based on a 7 T-scanner (Metzger et al., 2010; Walter et al., 2008a). The brain regions reported to respond to sexual stimuli in healthy men and women are presented in Table 2. It is essential to note that SA comprises two aspects: one is the pleasure reported by participants, which means that VSS can be considered received rewards ; the other aspect is that SA generates the desire to receive more sexual stimulation and more sexual rewards, which means that SA comprises an element comparable to craving. Thus, SA induces both liking and wanting (Berridge, 1996), which should be kept in mind when trying to understand the functional meaning of observed responses to VSS. In keeping with the putative successive stages through which emotionally competent visual stimuli are processed (Damasio, 1999), we begin by reviewing the findings on the dorsal and ventral visual streams, i.e., in lateral occipito-parietal and inferotemporal cortices, respectively. We then present findings regarding brain regions considered as involved in affective and motivational processes Meta-analysis Articles meeting all the following criteria were included in the meta-analysis: (i) to focus on heterosexual men, because of the small number of studies on women and on homosexual subjects; (ii) to use VSS, to avoid heterogeneity potentially associated with the few studies using other stimuli; (iii) to report whole-brain results; (iv) to report coordinates in the Talairach or MNI stereotactic spaces. An activation likelihood estimation (ALE) meta-analysis (Laird et al., 2011) was conducted using GingerALE software ( ALE models the uncertainty in localization of activation foci using Gaussian probability density distributions. The voxelwise union of these distributions yields the ALE value, an estimate of the likelihood that at least one of the foci in a dataset is truly located at a given voxel (Turkeltaub et al., 2012). The algorithm of version of the software prevented subject groups with multiple experiments in a dataset from influencing ALE values more than other groups and made ALE maps more conservative (Turkeltaub et al., 2012). The false discovery rate (FDR) method was employed to correct for multiple comparisons at a significance threshold of p < 0.05 and a cluster threshold of 200 adjacent voxels.

5 Table 1 Methodological features of studies in healthy men and women. Reference Participants Technique Blocks vs. events Sexual stimuli Control condition Measures Probability thresholds Gender Sexual orientation N Age: range; mean; SD Type Length (s) Type Length (s) Objective Subjective Rauch et al. (1999) M ; 25; 4.4 Stoléru et al. (1999) a M Hetero ; 23; Redouté et al. M Hetero ; (2000) a 30.7; Beauregard et al. M ; (2001) a 23.5; Bocher et al. M Hetero ; 27; (2001) a Savic et al. (2001) PET NA Audio presentation of sexual script Audio presentation of neutral script PET NA V: M F intercourse 90 V: neutral; humorous PET NA V: M F 120 V: neutral; intercourse; humorous; Ph: Ph: nude Fs neutral 1.5 T fmri Blocks V: M F and F F 39 V: M F and F F sexual nonsexual s s PET NA audiov: M F 600 audiov: cross intercourse and white noise; nature; talk show M Hetero ; ; PET NA Smelling EST 15; 4 times Smelling air or AND F Hetero ; ; Smelling AND 15; 4 times Smelling air or EST Park et al. (2001a) F ; 33; 1.5 T fmri Blocks V 240 V: documentary Park et al. (2001b) M ; 23; 1.5 T fmri Blocks V 120 V: documentary Arnow et al. M Hetero ; ; 3 T fmri Blocks V: M F sexual 120 V: relaxation; (2002); 1st/2nd sports run a Karama et al. (2002) a Costa et al. (2003) Mouras et al. (2003) Hamann et al. (2004) M Hetero 20 ; 25; T fmri Blocks V: M F sexual F 20 ; 24; 3 M 12 ; 24.7; 3.8 MEG Events Ph: nude Fs; nude Ms 543 V: sports; relaxation 179 V: neutral social 1 or 2 Ph: household objects HR; GSR; EMG 90 PPG; T; HR; RR 120 PPG; T; HR; BP PSA, Valence,.001, uncorr. General arousal, Happiness PSA, Humor 0.001, uncorr. PSA, Humor 0.001, uncorr. 39 No PSA 1) ROI:.05, corr.; 2) WB:.005, corr. No PSA.001, uncorr. 15 Respiratory movements 15 Pleasantness, irritability, intensity and familiarity mean rcbf across ROI:.05; voxel-wise analysis across ROI:.01, corr. 60 No PSA.05, number of activated pixels 60 No PSA, perceived.05, number of erection activated pixels 30 and 60; PPG; HR; PSA, Perceived WB:.05, corr.; 120 RR erection ROIs:.001, uncorr. and.05, SVC 120; No PSA WB:.05, corr.; ROIs:.001, uncorr. 1 or 2 No Valence and arousal F 12 ; 22.3; 2.9 M Hetero ; 26; 1.5 T fmri Blocks Ph: nude Fs 21 Ph: Fs, neutral 21 No Perceived erection, beauty, desire M Hetero 14 ; 25.9; 1.5 T fmri Blocks Ph: M F sexual Ph: nonsexual ; M F nude Fs F 14 ; 25.0; Ph: M F sexual ; nude Ms ; fixation cross No Sexual attractiveness; physical arousal Magnetic contingent variation and visual evoked magnetic fields:.05 WB:.05, corr.; ROIs:.001, uncorr..001, uncorr. S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

6 Table 1 (Continued) Reference Participants Technique Blocks vs. events Sabatinelli et al. (2004) Gender Sexual orientation N Age: range; mean; SD M 14 ; 18.6; 3 T fmri Blocks Ph: M F sexual F 14 ; 18.3; 1.4 Ferretti et al. (2005) a M Hetero ; ; 1.5 T fmri Blocks V: M F sexual Georgiadis and Holstege (2005) Berglund et al. (2006) Georgiadis et al. (2006) Gizewski et al. (2006) Sexual stimuli Control condition Measures Probability thresholds Type Length (s) Type Length (s) Objective Subjective 12 Ph: nonsexual 12 No No WB:.001, uncorr.; average signal across ROI: V: sports; emotionally neutral scenes 120 PPG PSA WB:.05, corr.; ROI averaged signal:.05 3 Ph: sports 3 Events Ph: M F sexual M Hetero ; 33; PET NA masturbation by partner M Hetero 12 ; 28; 2 PET NA Smelling EST 15; 4 times Smelling air 15; 4 times Respiratory movements F Hetero 12 ; 26; 2 Smelling AND F Homo 12 ; 33; 6 Smelling EST F Hetero ; 34; PET NA Clitoral masturbation by partner M ; 28; 1.5 T fmri Blocks V: M F sexual F ; 27; Heinzel et al. (2006) a M 13 ; 38.7; T fmri Events Ph: not specified Kim et al. (2006) a M Hetero ; 52; Moulier et al. M Hetero ; (2006) a 21.7; Ponseti et al. (2006) 1.5 T fmri Blocks V: M F sexual M Hetero 12 ; 26.8; T fmri Events Ph: nude M and F aroused genitals F Hetero 12 ; 22.7; 3.0 M Homo 15 ; 27.2; 3.8 F Homo 14 ; 25.1; 5.9 PET Blocks audiov: M F intercourse Tsujimura et al. M Hetero ; (2006) a 35.5; rest 120 No No.05, corr. 120 Rest 120 Rectal pressure 31 V: nonsexual M F 5 Emotional nonsexual; neutral; fixation cross 120 V: sports; relaxation; Pleasantness, irritability, intensity and familiarity WB:.05, corr. PSA.001, uncorr. 31 None PSA WB:.05, corr.; ROIs:.001, uncorr. 5; 5; 8 10 Reaction time 120; 30 or 60 Self relatedness, valence, emotional intensity.001, uncorr. No No WB:.05, corr.; ROIs:.001, uncorr..001, uncorr. erection, beauty, desire, pleasure, displeasure 1.5 fmri Blocks Ph: nude Fs 35 Ph: dressed Fs 35 PPG Perceived 0.3 Ph: nature, sport events 90 audiov: nonsexual human ; mosaic images 0.3 No Valence and arousal.05, corr. 90 PPG None.05, corr S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

7 Ethofer et al. (2007) M Hetero 12 ; 25.1; fmri Events Words, erotic prosody F Hetero 12 M ; fnirs Blocks V: Roman orgy 25.8; 7.6 showing Leon-Carrion various kinds et al. (2007) of sexual F 15 Meseguer et al. M Hetero 14 ; 28.8; 1.5 T fmri Blocks Ph: M F sexual (2007) a Miyagawa et al. M Hetero ; (2007) a 34.7; 3.3 Sabatinelli et al. (2007) Safron et al. (2007) M 11 ; introductory psychology students PET NA audiov: M F intercourse 3 T fmri Events Ph: M F sexual F 11 M Hetero 11 ; 21; 3 T fmri Events Ph: M M or F F sexual s Homo 11 ; 21; Brunetti et al. (2008) a M Hetero ; 24.9; Bühler et al. (2008) a 1.5 T fmri Blocks V: M F sexual Words spoken, neutral prosody V: man walking in a crowd 30 Ph: neutral, inanimate objects 90 audiov: nonsexual human ; mosaic images 3 Ph: neutral people, romantic M F couples, mutilations, dental scenes, snakes, threatening people 3.5 Ph: M and F sports activity 180 V: neutral; M and F sport activity 19.8 Ph: neutral people, plants, objects No No.05, corr. No Valence and arousal 30 No Valence and arousal.001 (oxygenated hemoglobin level).005, uncorr. 90 PPG none WB:.05, corr.; ROI averaged signal:.05 3 No No.001, uncorr. 3.5 No Liking; disliking WB:.001, uncorr.; ROI averaged signal: PPG PSA.001, uncorr. M Hetero 10 ; 32; T fmri Blocks Ph: nude Fs, M F sexual s 19.8 No Valence and arousal.05, corr. Events T fmri Blocks V: M F sexual 50 V: neutral; 50 PPG Perceived.05, corr. humorous erection, beauty, desire, pleasure, displeasure M Hetero ; 1.5 T fmri Blocks V: M F sexual 38.5 V: couples, 38.5 No PSA WB:.001, 34.8; 7.6 ; nonsexual uncorr.; ROI: M M sexual activity.05, uncorr. Homo ; 32.0; 6.8 M Homo 12 ; 32.0; T fmri Blocks Ph: nude M 38.5 Ph: dressed M 38.5 No PSA.05, corr. Mouras et al. M Hetero ; (2008) a 21.9; 1.6 Paul et al. (2008) a Schiffer et al. (2008a) Schiffer et al. M Hetero 12 ; 36.1; T fmri Blocks Ph: nude F 38.5 Ph: dressed F 38.5 No PSA.05, corr. (2008b) a S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

8 Table 1 (Continued) Reference Participants Technique Blocks vs. events Zhou and Chen (2008) Gender Sexual orientation N Age: range; mean; SD Sexual stimuli Control condition Measures Probability thresholds Type Length (s) Type Length (s) Objective Subjective Walter et al. (2008a) M Hetero 6 ; 25.6; T fmri Events Ph: erotic 4 Ph: non-erotic emotional; fixation cross 4; fixation cross: 7.5 to 10.5 No PSA, emotional intensity, pleasure.05, corr. M Hetero ; 25.7; 1.5 T fmri Events Ph: partially or completely nude Fs or Ms 5 Ph: emotionally relevant sports 5 Reaction time PSA.001, corr. or heterosexual scenes or social Walter et al. couples s; (2008b) emotionally relevant non-human motives F ; 23.9; M Hetero 10 ; 27.9; T fmri Blocks V: M F, M M 180 Fixation cross 180 No PSA.001, uncorr. and F F Hu et al. (2008) a couples, sexual Homo 10 ; 26.5; 5.1 F Hetero 19 ; 23.4; T fmri Events Airborne male 8 Neutral sweat; 8 RR, Intensity and 1) ROI:.005, sexual sweat androstadienone; respiratory pleasantness of corr.; 2) WB: amplitude smell.0005, uncorr. phenylethyl alcohol Arnow et al. (2009) F Hetero ; ; 3 T fmri Blocks V: M F sexual PSA, through.001, uncorr. potentiometer Klucken et al. (2009) Rudrauf et al. (2009) M Hetero 20 ; 23.0; T fmri Events Ph: M F couples, sexual F 20 ; 23.0; 3.3 M ; 26; MEG Events V: erotic content 180 V: F sports 120 Vaginal photoplethysmography 4 Ph: Ms and Fs, neutral scenes 10 V: fearful or disgusting; neutral: landscapes, faces, objects GSR PSA, valence, arousal, disgust 10 EKG Valence and arousal ROIs;.05, corr. Rupp et al. (2009) F Hetero ; ; 3 T fmri Events Ph: M faces 4 Ph: houses No Rating: How 0.001, uncorr. likely would participant be to have sex with depicted M Seo et al. (2009) a M Hetero 12 ; 30.5; T fmri Blocks Ph: M F sexual 24 Ph: nature, 24 No PSA, food.05, uncorr. s food, smiling M F couples craving Georgiadis et al. (2010) M Hetero ; 29.3; 3 T fmri Blocks Masturbation by partner until maximum arousal or ejaculation rest 300 PPG, EKG, RR PSA.001, uncorr..05, corr S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

9 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) , uncorr. 4 No Perceived SA, emotional intensity, salience, and pleasure 4 Ph: non-erotic emotional M Hetero 10 ; 25.6; T fmri Events Ph: not specified Metzger et al. (2010).05, corr. 180 Rest 60 No Attractiveness, PSA 30 None None.05, corr. 30 V: M F nonsexual Sundaram et al. M Hetero ; 25; 3 T fmri Blocks V: M F sexual (2010) a Zhu et al. (2010) F Hetero 15 ; 26.3; T fmri Blocks V: M F sexual Abbreviations: : not mentioned in article; AND: androgen-like compound; BP: blood pressure; corr.: corrected for multiple comparisons; EKG: electrocardiogram; EMG: electromyogram; EST: estrogen-like compound; F: females; fmri: functional magnetic resonance imaging; fnirs: functional near-infrared spectroscopy; GSR: galvanic skin response; hetero: heterosexual participants; homo: homosexual participants; HR: heart rate; M: males; MEG: Magnetoencephalography; NA: not applicable; PET: positron emission tomography; Ph: photographs; PPG: penile plethysmography; PSA: perception of sexual arousal; ROI: region of interest; RR: respiratory rate; s: seconds; SA: sexual arousal; SVC: small-volume correction; T: testosterone; uncorr.: uncorrected for multiple comparisons; V: videos; WB: whole brain. a Articles included in meta-analysis. Because findings on deactivation were rarely reported, we analyzed only foci where activation was related to VSS or was positively correlated with markers of SA. The coordinates of foci referring to Talairach space were converted to coordinates in MNI space using the program supplied by GingerALE Brain areas involved in sexual arousal As shown in Table 1, the majority (n = 36, i.e., 73.5%) of the 49 articles on SA (excluding orgasm) were based on fmri, 10 (20.4%) on PET, 2 (4.1%) on MEG and 1 (2.0%) on functional near-infrared spectroscopy. Most studies relied on a subtractive approach, i.e., on the contrast between a sexual and a control condition, assumed to differ only in one property represented by induced SA. In some studies, these analyses were supplemented by the combination of different contrasts (exclusive masking and conjunction analyses), and by correlational analyses between level of regional brain activation and level of a psychological or physiological marker of SA Healthy male volunteers Review Lateral occipital and/or lateral temporal cortex. The majority (n = 31; 83.8%) of the 37 articles reporting separately on men have found an activation of visual areas of the lateral occipital and/or lateral temporal cortices in response to VSS, i.e., in the inferior occipital gyri, middle occipital gyri, inferior or middle temporal gyri (Table 2; Fig. 1). These findings suggest that these areas might process information related to the sexually arousing character of visual stimuli over and above the information related to low-level visual features of stimuli, such as color and luminance, and over and above the general (non specifically sexual) emotional salience of these stimuli. If these areas are involved in processing the sexual character of stimuli, is the categorization of stimuli as sexually relevant related to the activation of these areas or is this categorization related to the activation of higher-order areas, such as the orbitofrontal cortex (OFC)? In the latter case, the activation of occipital/temporal areas might result from a top-down mechanism. Current studies do not answer this question. Methods for the study of functional connectivity (Lin et al., 2009) have not yet been used to shed light on this issue. Moreover, are these lateral occipitotemporal activations a true effect of the sexually arousing character of VSS or do they merely result from using inadequate control stimuli? In other words, did possible differences between sexual and control stimuli regarding their low-level visual features account for observed differences in lateral occipitotemporal activation? A strategy to answer this question is to compare the response of these areas to the same visual stimuli across different groups, as the same stimuli may represent sexual stimuli for one group while being sexually irrelevant for another. If, indeed, these regions are activated in response to sexual stimuli, they should be more activated in response to pictures of nude women in heterosexual men than in men with pedophilia exclusively attracted to prepubescent girls. However, Schiffer et al. (2008b) did not observe such higher activation. Similarly, they did not find a higher activation of these areas in response to pictures of nude men in homosexual men than in men with pedophilia exclusively attracted to boys (Schiffer et al., 2008a). Similarly, these areas showed bilateral activation in heterosexual and in homosexual men, whether they were presented with video clips corresponding to, or opposite to, their sexual orientation (Ponseti et al., 2006). Again, the comparison of preferred stimuli to nonpreferred stimuli in homosexual and heterosexual subjects failed to show a differential activation of these areas. A second way to address the problem of potentially inadequate control stimuli is to determine whether the magnitude of

10 Table 2 Brain regions responding to sexual stimuli in healthy men and women. Reference Gender/ Activation in response to sexual stimulus/positive correlation with sexual arousal Deactivation/Negative correlation with Orientation sexual arousal LOCC and/or LTC IT AMY SPL IPL Cl OFC mpfc INS Hy CN Pu VS Th ACC PM Cb MB PCC mofc Left IFG Rauch et al. (1999) M No No No No No L No No No No No No No No R No No No No No No No L Stoléru et al. (1999) M B No No No No R R No R No R No No No L No No No B N L No No Redouté et al. M B B L R R B R R No Yes R B L B B B B No B R L B B (2000) Beauregard et al. M B R R B No No No No No Yes No No No No No No L No No No L B No (2001) Bocher et al. (2001) M B B No B L No No No No No No No No No No No No R B R No B B Savic et al. M No R No No No No No No No R No No No No No No No No (2001) F No R No No No No No No No L No No No No No No No No Park et al. (2001a) F Yes B No No No No No No B No B No No B L No No No Park et al. (2001b) M Yes B No No No No No No B No B No No B Yes No No No Arnow et al. (2002) M R No No No No B No No B R L B No No B R No No No No No R No Karama et al. M B No B No No No B B B B No No B B B No No No (2002) F B No B No No No B B B No No No B No B No No No Costa et al. M, hetero B (2003) F No Mouras et al. M B B No B L No No No B No No No No No No B No No R No No B B (2003) Hamann et al. M B No B B No No No No No L No No B R Yes No R No (2004) F B No No B No No No No No No No No B Yes No No Sabatinelli M B B et al. (2004) F No No Ferretti et al. M, Blocks B B B No B No No No B B No No No B B No No No (2005) M, Events B B B No B No No No No No No No No No No B No No Georgiadis et al. (2006) M No No No No R R No No R No No R No No No No No No L M, hetero No L L No No No No No No R No No No No No No No No Berglund et al. F, hetero No R No No No No No No No B No No No No No No No No (2006) F, homo No L L No No No No No L No No No No No No No L No Georgiadis et al. F No No No L L No L No No No R No No No No R L R No No No L L (2006) Heinzel et al. M No No R No No No No L No No No No No No R L No R (2006) Kim et al. (2006) M R No No R No No B No R No B No No No L R No No Moulier et al. (2006) M B B No B B No R R B No L No No B R B R No B Yes L B R Ponseti et al. (2006) Tsujimura et al. (2006) Ethofer et al. (2007) Meseguer et al. (2007) Miyagawa et al. (2007) Sabatinelli et al. (2007) M, hetero R No No B No No No No No No No No R R No B No No F, hetero No M, homo No F, homo No M B No No No No No R No L No No No No No R No R No No No L L L M R F M B B L R No No L No L No No B L B L L L No M B No No No No No No No No L No R R No No No No No No No No L No M B B B No B No No L L No No No B No R R No No F LOCC and/or LTC IT 1490 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

11 M, hetero and homo; preferred B No No L R No No No L No B No No L B No No No No No No No No vs. sports Safron et al. stimuli (2007) M, hetero B No L L No L Yes No L Yes B L Yes B B No B Yes No No No No No and homo; preferred vs. nonpreferred stimuli Brunetti et al. M L B B B B No No No B B No No No B B B No No (2008) Bühler et al. M; blocks B R No L R No No No No No No No No No No B L No No No No R No (2008) M; events B No No No No No R No B No No No No L R L L No No No No No No Mouras et al. M R B No B B B No Y + L B No B R L B B B R No L No L B No (2008) Paul et al. M, hetero B B No No B No No L R L L No No No B No B R (2008) M, homo B B No R L No L R No L No No No No No No B No Schiffer et al. M, homo B B No R No No No No No No No No No No No No No No (2008a) Schiffer et al. M, hetero B B B B L No B No No No B No No B L B No R No No No No No (2008b) Walter et al. M B No B No B (2008a) Walter et al. M B R R B No No No L No B No No B No Yes No No L (2008b) F Hu et al. (2008) M, hetero B R No B L No No No B No No No No L L L B No M, homo B No No L No No No B L No L No No B R R L No Zhou and Chen F No R No No No No R No No R No No No No No No No No (2008) Arnow et al. (2009) F B B L B L R L L No No B No No B B B B L L Klucken et al. M B No B No No No B No B No No No B B B No No L (2009) F B No B No No No B No B No No No B No L No No No Rudrauf et al. M B B No No No No B B L No No No No No B B No No (2009) Seo et al. (2009) M L No No B L No R B R No B R R L No B L R Georgiadis et al. M B No No R No B No No R L No L L R B R No No No B L B No (2010) Metzger et al. M B B B L No B B R B R L (2010) Sundaram et al. M B B B No No No No No L L B R No B B Yes L L (2010) Zhu et al. (2010) F B B B B B No No No No B No R No B No R B B Yes No No B Notes: (i) Participants whose sexual orientation is not specified are heterosexual; (ii) for papers reporting no significant Gender by Stimulus Category, cells for men and women are merged; (iii) as the article by Gizewski et al. (2006) focuses on between-gender differences and does not specify within-gender activations, it is omitted from Table 2 and presented in Table 3; (iv) similarly, as the paper by Leon-Carrion et al. (2007) focuses exclusively on the dorsolateral prefrontal cortex, it is omitted; as the study by Rupp et al. (2009) exclusively target the effects of menstrual cycle phase on neural activation, it is omitted. Abbreviations: : not covered by scan or not mentioned in article; ACC: anterior cingulate cortex; AMY: amygdala; B: bilateral response; Cb: Cerebellum; Cl: claustrum; CN: caudate nucleus; F: females; hetero: heterosexual participants; homo: homosexual participants; Hy: hypothalamus; IFG: inferior frontal gyrus; INS: insula; IT: Inferotemporal cortex; L: response in left hemisphere; LOCC: lateral occipital cortex; LTC: lateral temporal cortex; M: males; MB: midbrain; mofc: medial orbitofrontal cortex; mpfc: medial prefrontal cortex; PCC: posterior cingulate cortex; PM: premotor areas; Pu: putamen; R: response in right hemisphere; Th: thalamus; VS: ventral striatum; Yes: response, but hemisphere not specified. S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

12 1492 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) the response of these visual areas is correlated with a behavioral index of SA, such as perceived SA or penile erection. Indeed, the degree of activation of these areas was correlated with the level of the erectile response (right middle occipital gyrus, Arnow et al., 2002; bilateral cuneus, Ferretti et al., 2005; bilateral middle occipital gyrus, Moulier et al., 2006). In order to clarify the role of these areas, Moulier et al. (2006) analyzed the relation between the time courses of the BOLD signal and of the plethysmographic response and found that the former drives the latter with a 20 s delay. A third way to address the problem of potentially inadequate control stimuli was provided by a classical conditioning study of brain responses to conditioned stimuli (CS+: geometrical shapes) that were paired with sexually stimulating photographs (unconditioned stimuli; UCS), on the one hand, and to CS (other geometrical shapes) that were paired with neutral photographs (unconditioned stimuli), on the other hand (Klucken et al., 2009). In the left occipital cortex, responses were stronger to the CS+ than to the CS. About half of the sample became aware of the type of geometric figure that preceded the presentation of sexually stimulating pictures, spontaneously realizing the contingency between CS+ and UCS. Post hoc t-tests showed significant differences in arousal, valence, and SA ratings in response to CS+ compared with CS in the aware group, but not in the unaware group. Only in the group of aware subjects, the left occipital cortex was more activated in response to the CS+ than to the CS. This conditioning study rules out the possibility that activation of lateral occipitotemporal areas was related to different low-level visual features of VSS and control stimuli and suggests it was a true correlate of the sexually arousing character taken on by these stimuli. Thus, the subtractive analyses on the one hand, and the correlational analyses and the conditioning study, on the other, do not support the same conclusions as to the functional significance of lateral occipitotemporal activation in response to VSS and this issue remains open. Furthermore, is the activation of lateral visual areas related to the sexually arousing character of visual stimuli, or would any emotional visual stimuli elicit such activation? The former interpretation is supported by the finding that men showed a significantly greater signal increase in these areas in response to erotic, compared to emotionally aversive pictures (Sabatinelli et al., 2004). However, contrasting the response of visual areas to VSS and to humorous stimuli did not show any significant difference (Redouté et al., 2000). In their review of emotion activation studies based on PET or fmri, Phan et al. (2002) found that among 35 visual induction studies, 60% reported activation in the occipital cortex. They considered two possible explanations for these activations: (i) the occipital cortex might appraise the emotional valence of complex visual stimuli; (ii) the occipital cortex might be recruited because the emotional visual stimuli have higher image complexity than control stimuli. In two studies that controlled for emotional valence and for emotional arousal of sexual and nonsexual stimuli (Ponseti et al., 2006; Walter et al., 2008b), right and/or left lateral visual areas remained more activated in response to VSS than to nonsexual emotional stimuli. This finding suggests that activation in lateral visual areas is related to the sexually arousing character assigned to stimuli and is not a mere effect of valence or of general emotional arousal. Nevertheless, higher activation in lateral occipitotemporal areas in response to VSS could still be related to their higher visual salience, irrespective of their emotional valence or arousing character. Finally, does the activation of the extrastriate body areas (Downing et al., 2001) account for activation of lateral visual areas, irrespective of the sexually arousing character of the visual stimuli? The extrastriate body area is a distinct region that was found bilaterally in the lateral occipitotemporal cortex in all subjects tested and that apparently reflects a neural system specialized for the visual perception of the human body. The mean (SD) MNI coordinates were x, y, z = 51 (2), 71 (4), 1 (3), and x, y, z = 51(4), 72 (5), 8 (4), for the right and left extrastriate body areas, respectively (Downing et al., 2001). In several papers, (e.g., Miyagawa et al., 2007; Moulier et al., 2006; Schiffer et al., 2008b; and Stoléru et al., 1999), some of the extrastriate visual activations fell in the extrastriate body area coordinates. The activation of the extrastriate body area could also account for the finding that, compared with nonsexual stimuli, pictures representing male and female sexually aroused genitals activated a large region encompassing the extrastriate body area, regardless of the sexual orientation of the participants (Ponseti et al., 2006) Inferior temporal cortex. The majority of studies (Table 2; n = 20, 54.1%; Fig. 1) have reported an activation in the inferior temporal cortex essentially in the fusiform gyri in response to VSS or to putative pheromonal stimuli (Savic et al., 2001). By contrast with findings on lateral visual areas, in response to clips of heterosexual couples engaged in sexual activity, inferotemporal areas showed activation in heterosexual men, but not in homosexual men (Paul et al., 2008); similarly, in response to clips of homosexual couples engaged in sexual activity, an inferotemporal activation was recorded only in homosexual men. Again, in contrast with findings related to lateral areas, in response to photographs of nude women, Schiffer et al. (2008b) did find a greater activation in inferotemporal areas in heterosexual men than in men with pedophilia exclusively attracted to female children. The level of activation of inferotemporal areas was linearly correlated with the level of penile tumescence (Ferretti et al., 2005; Redouté et al., 2000), which suggests that the activation of the inferotemporal cortex is related to the sexually arousing character of visual stimuli and not just to their low-level visual features. In a study controlling for emotional valence and for emotional arousal associated with sexual and nonsexual stimuli (Walter et al., 2008b), the right inferotemporal area remained more activated in response to VSS than to nonsexual emotional stimuli. Thus, the evidence for a relation between the activation in inferotemporal areas and the sexually arousing character of VSS is more clear-cut than in the case of lateral visual areas. According to Rolls (2000), the identification of visual stimuli within the ventral stream of visual processing proceeds schematically from the primary visual cortex, V1, through V2 and V4, to the posterior inferior temporal visual cortex (TEO) and the anterior inferior temporal cortex (TE). In rhesus monkeys, it was found that the responses of neurons of the inferior temporal cortex were independent from whether visual stimuli were associated with reward, punishment, or were neutral (Rolls et al., 1977). Rolls (2000) suggested that the extrastriate visual cortex and the inferior temporal cortex are involved in the visual processing of objects, i.e., with what is being looked at, independently of reward vs. punishment associations. However, above mentioned evidence from neuroimaging studies of SA indicate that the response of inferotemporal areas may be related to differential perceptions of stimuli, i.e., as sexual or nonsexual stimuli, and that their response is correlated with the magnitude of erection. No study has yet investigated whether these responses of inferotemporal areas to VSS may result from top-down influences, i.e., from higher-order areas involved in assessing the motivational relevance of stimuli. Various kinds of motivationally relevant stimuli, not just VSS, have been found to induce activation in the fusiform gyri. Thus, Kilts et al. (2001) reported positive correlations between self-rated cocaine craving in response to script-guided mental imagery associated with cocaine use and regional cerebral blood flow (rcbf) in the right fusiform gyrus. Similarly, food-related visual stimuli elicited greater responses in the right anterior fusiform gyrus when participants were in a hungry state relative to a satiated state (LaBar et al., 2001). In addition, the fusiform gyri may be activated even when only internal signals induce the motivational state and no

13 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) sensory cues are used in the experiment. Thus, Parsons et al. (2000) found an activation in the fusiform gyri after thirst was induced by hypertonic saline infusion, in the absence of any sensory cue, such as the visual presentation of a beverage. While the above studies support the involvement of inferotemporal regions in motivational processes in general, they also question the sexual specificity of activations found in response to VSS. Finally, it is likely that faces are more represented among VSS than among stimuli used in control conditions such as sports highlights. Thus, an alternative interpretation of the activation of the fusiform gyri in response to VSS is suggested by several studies showing the crucial role played by the fusiform face areas in the perception of faces (e.g., Kanwisher et al., 1997; Katanoda et al., 2000). Indeed, in some studies (Bocher et al., 2001; Ferretti et al., 2005; Moulier et al., 2006; Redouté et al., 2000), peaks of activation fell in the fusiform face area. In addition, in a study that used male and female sexually aroused genitals as VSS, no activation of the inferotemporal cortex was reported (Ponseti et al., 2006). This alternative interpretation, however, is unlikely to fully account for the reported activation of the fusiform gyrus, as this region was reported activated by odorous sex hormone-like compounds (Savic et al., 2001) and in studies that did not use faces as visual stimuli (e.g., Zhou and Chen, 2008) Amygdala. Neuroimaging studies have yielded highly variable findings regarding the response of the amygdalas to sexual stimuli, with 13 studies reporting activation (35.1%) (Table 2; Fig. 1). This variability contrasts with findings in male rodents where chemosensory inputs from the main olfactory and from the vomeronasal systems are probably the most important stimuli for sexual behavior and are processed in the medial amygdala (Hull and Dominguez, 2007). More specifically, whereas lesions of the medial amygdala impair the ability of male rats to respond to olfactory and other sexual cues from a receptive female, which implicates this structure in the appetitive stage of sexual behavior, amygdalar lesions do not interfere with the consummatory aspects of sexual behavior (Everitt, 1990). The prevailing interpretation of amygdalar activation is that this structure receives multimodal sensory inputs and relays processed information to the ventral striatum, hypothalamus, autonomic brainstem areas, and the prefrontal cortex. In this process, the amygdala participates in the evaluation of the emotional content of the complex perceptual information associated with erotic stimuli (Ferretti et al., 2005). The activation of the fusiform gyri discussed above might depend on the influence of the amygdalae, as has been demonstrated in response to fearful faces, another kind of emotional visual stimulus (Vuilleumier et al., 2004). In addition to this role in emotional appraisal, Hamann et al. (2004) related amygdalar activation to motivational processes and specifically to the role of the amygdala in encoding the current value of reward representations accessible to predictive cues. This role is not specific for sexual incentives and has been demonstrated for food and drug incentives (reviewed in Murray, 2007). The study of Walter et al. (2008b) is relevant to analyzing this distinction between two possible roles of the amygdala, one related to emotion and the other to motivation. They concluded that the amygdala was associated only with a general emotional component associated with SA. According to Murray s review (Murray, 2007), the roles of the amygdala in emotional reactions and in reward processing should be distinguished. Whereas the role of the amygdala is crucial in emotional responses, recent research based on selective lesions of the amygdala made with fiber-sparing excitotoxins has shown that the performance on experimental tasks, which historically had linked the amygdala with stimulus reward association in monkeys, does not actually depend on the amygdala (Murray, 2007) but on s between inferotemporal/perirhinal cortices and OFC. In a proposed model (Murray, 2007), both the amygdala and inferior temporal/perirhinal cortices interact with the OFC to guide decisions. Following this new view, the amygdala would process the emotional aspects rather than the motivational relevance of VSS. Contrasting with the activation discussed above, the amygdala has also been reported to be deactivated when SA was induced by manual stimulation of the penis by the subjects partners (Georgiadis and Holstege, 2005) and during orgasm (see below) (Holstege et al., 2003). The amygdalar deactivation associated with SA is consistent with the state of hypersexuality and indiscriminate sexual behavior observed in the Kluver and Bucy syndrome after bilateral ablation of temporal lobes, including the amygdalae (Klüver and Bucy, 1939; Terzian and Ore, 1955). However, the specific regions whose lesions cause this syndrome have not been well delineated and bilateral lesions limited to the amygdalas in humans do not consistently produce hypersexuality or other features of the syndrome (Baird et al., 2007). Finally, the lack of any amygdalar response in many studies is puzzling. It could be related to partial volume effects, i.e., the combination within single voxels of signal contributions from two or more amygdalar functional regions presenting opposite responses to sexual stimuli Parietal cortex. The majority of papers (n = 21; 56.8%) have reported an activation in the parietal cortex in response to VSS [superior parietal lobule (n = 19, 51.4%); inferior parietal lobule (n = 15, 40.5%)] (Table 2; Fig. 1). While the parietal cortex is part of the dorsal visual stream, this activation does not seem to be entirely accounted for by the visual nature of stimuli as it was also found when SA was induced by tactile stimuli (Georgiadis and Holstege, 2005; Kell et al., 2005). Within the parietal cortex various areas have been found activated. We shall review these various parietal areas as each of them seems to play distinct functions in SA. Firstly, tactile stimulation of the penis induced an activation in the contralateral postcentral gyrus (Georgiadis et al., 2009; Kell et al., 2005), i.e., in the primary sensory cortex. Although early studies indicated that the penile representation in the primary somatosensory cortex lay on the mesial part of the parietal lobe (e.g., Penfield and Rasmussen, 1950), neuroimaging studies show that the activation lies in the uppermost part of the postcentral gyrus (Kell et al., 2005). This finding is consistent with the positive correlation between the level of penile tumescence induced by VSS and the level of rcbf in the uppermost part of the left postcentral gyrus (Redouté et al., 2000). Interestingly, these converging results were obtained with different means of inducing penile sensations, i.e., tactile stimulation (Georgiadis et al., 2009; Kell et al., 2005) and VSS (Redouté et al., 2000). Surprisingly, bilaterally in the lower part of the primary somatosensory cortex, a region corresponding to the mouth and hands, the BOLD level was positively correlated with the degree of penile tumescence (Mouras et al., 2008). In addition, the activation in this region seemed to drive the penile response, as the BOLD signal was correlated with the plethysmographic response measured 20 s later. One possible explanation may be related to the actions depicted in the clips presented in this study, namely, viewing a man caressing a woman with his hand and being kissed by her on the mouth might induce activations in the part of the somatosensory cortex corresponding to the hand and the mouth. The postcentral activation found in another study based on videoclips was also in the lower part of the somatosensory cortex (Hu et al., 2008). These findings draw attention to the neglected concepts of somatosensory imagery and of somatosensory representation, i.e., while participants witness the sensations, actions and somatic pain of others, they consistently show vicarious activation in their own somatosensory cortices (Keysers et al., 2010). Secondly, tactile stimulation of the penis induced an activation in the parietal operculum, on the right side (Georgiadis and Holstege, 2005) or bilaterally (Georgiadis et al., 2010; Kell et al.,

14 1494 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) ). The levels of penile tumescence and of perceived SA were positively correlated with the level of parietal opercular activation, whether SA was induced by tactile stimulation (Georgiadis et al., 2010) or by VSS (Bocher et al., 2001; Ferretti et al., 2005; Redouté et al., 2000). These parietal opercular foci lie possibly in the secondary sensory cortex. Thirdly, a role of the inferior parietal lobule in motivational processes has been demonstrated. In monkeys, an early electrophysiological study showed that inferior parietal lobule neurons increased their discharge frequencies during steady fixations of motivationally relevant visual targets ( visual objects the animal desires ; Lynch et al., 1977), while they remained relatively inactive during episodic fixations occurring during environmental exploration. Accordingly, Lynch et al. (1977) suggested that the parietal lobe matches visual inputs to the internal drive state of the organism. The inferior parietal lobule could therefore contribute to maintain attention to motivationally relevant targets. Fourthly, another parietal region, the posterior parietal cortex, has also been implicated in motivational processes, as neurons in this region encode the value of expected outcomes of actions (Musallam et al., 2004; Sugrue et al., 2004). Studies in monkeys show that the lateral intraparietal area a part of the posterior parietal cortex provides topographic representations of the environment, which encode the salience of different objects. This salience map may be instrumental in the selection of attentionworthy objects, which often also constitute potential, or desirable, targets for eye movements (Gottlieb, 2007). Studies suggest that the human equivalent for monkey lateral intraparietal area is located in the medial intraparietal sulcus (Grefkes and Fink, 2005). Indeed, the medial intraparietal sulcus shows activation in response to VSS (Mouras et al., 2003; Ponseti et al., 2006; Sabatinelli et al., 2007). The activation of a brain area involved in attention is consistent with the important role played in SA by increased attention to targets once they have been appraised as sexual cues (Barlow, 1986; Janssen and Everaerd, 1993). In patients undergoing awake brain surgery, stimulating the posterior parietal cortex (Brodmann areas 39 and 40) triggered a strong intention and desire to move (Desmurget et al., 2009) (e.g., stimulation on the right parietal regions, desire to move contralateral hand, arm, or foot; on the left, intention to move the lips and to talk). Thus, the posterior parietal cortex is thought to be involved not only in attentional, but also in intentional processes. While in response to VSS, Brodmann area 39 was found deactivated or negatively correlated with penile tumescence (Moulier et al., 2006; Redouté et al., 2000), Brodmann area 40 was found activated or positively correlated with penile tumescence in many studies (e.g., Bühler et al., 2008; Ferretti et al., 2005; Kim et al., 2006; Mouras et al., 2008; Redouté et al., 2000). This activation could be the correlate of the participants expressed desire to perform sexual actions in response to VSS, e.g., the same actions as those depicted in the clips (Redouté et al., 2000). Fifthly, during debriefing of sessions, subjects often report sexual motor imagery in response to VSS. Motor imagery has been defined as a dynamic state during which the representation of a given motor act is internally rehearsed within working memory without overt motor output (Decety and Grèzes, 1999). Motor imagery tasks unrelated to sexuality induce activations in the superior and inferior parietal lobules (e.g., Grèzes and Decety, 2001; Stephan et al., 1995) in foci located close to those activated in response to VSS. It has been suggested that these areas play a role in sexual motor imagery, which is an important cognitive component of SA (Moulier et al., 2006). Finally, Mouras et al. (2008) reported that the levels of activation in the inferior parietal lobules and in the left frontal operculum, areas which both belong to the mirror-neuron system, were positively correlated with, and predicted by a few seconds, the magnitude of the erectile response. These findings suggest that the response of the mirror-neuron system may not only code for the motor correlates of observed actions, but also for the autonomic correlates of these actions Claustrum. Claustral activation has been reported in 10 studies (27.0%) (Table 2; Fig. 1). Among those, a PET study (Redouté et al., 2000) and three fmri studies (Arnow et al., 2002; Georgiadis et al., 2010; Mouras et al., 2008) have shown bilateral activation of the claustrum to correlate positively with penile tumescence. As the claustrum is a very thin structure close to the insula, two high resolution fmri studies (7 T) were useful in demonstrating claustral activation (Walter et al., 2008a; Metzger et al., 2010). Arnow et al. (2002) have suggested that the activation of the right insula/claustrum region is related to its role in cross-modal information transfer, a process that occurs for instance in tasks requiring subjects to visually identify objects that have been perceived by touching. Thus, Arnow et al. (2002) suggested that claustral activation may reflect cross-modal transfer of visual input to imagined tactile stimulation. Although speculative, this interpretation is consistent with the concept of somatosensory imagery that has been suggested to account for certain parietal activations (see above, Section ). Alternatively, the claustrum may be more directly involved in motivational processes. Thus, a claustral activation was found both in a study of the neural correlates of thirst (Denton et al., 1999a) and in a study on hunger (Tataranni et al., 1999). Additionally, there is further evidence for the role of the claustrum in emotional and motivational responses in rats (Hamamura et al., 1997) and in man (Reiman et al., 1989). Regarding the sexual specificity of claustral activation, in a neuroimaging study of hypogonadism (Redouté et al., 2005) the activation of the claustrum in response to VSS was lower in untreated patients than in healthy controls and activation increased under androgen treatment. That these claustral responses appear to depend on plasma testosterone level indicates that this region mediates SA and not only a process of general emotional or motivational arousal Orbitofrontal cortex. The OFC showed activation in response to VSS in 14 studies (37.8%) (Table 2; Fig. 1). Before discussing the functional role of OFC in SA, it is essential to analyze the participants subjective responses to VSS. Subjective SA included pleasure, perceived erection, appraisal of beauty of displayed characters, desire to perform sexual actions, and sexual imagery representing both desired persons and desired actions (Moulier et al., 2006; Mouras et al., 2008). Thus, some responses correspond to actual (perceived) reward, i.e., pleasure, while others correspond to potential (imagined) reward, i.e., desire. The function of the OFC in emotion and motivation has been mostly studied in domains other than sexual behavior. These studies concern primary reinforcers, such as odors and tactile stimuli (Rolls, 2000), or secondary reinforcers, such as pleasurable music (Blood and Zatorre, 2001), money or social praise (O Doherty, 2007). In these domains, there is considerable evidence that the OFC is firstly involved in coding for the value of received as opposed to anticipated reward (O Doherty, 2007). This role of the OFC may account for its activation in response to VSS as these stimuli are rewarding in themselves. Studies have explicitly used attractiveness of faces as a reward (Aharon et al., 2001; Ishai, 2007; O Doherty et al., 2003). For instance, in heterosexual men and homosexual women, attractive female faces evoked stronger activation in the medial OFC than did attractive male faces, whereas in heterosexual women and homosexual men, attractive male faces elicited stronger activation than did attractive female faces (Ishai, 2007). Secondly, neuroimaging studies have implicated the OFC in the prediction of future rewards. This function is commonly thought

15 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) to correspond to an internal representation of the incentive value of stimuli (Roesch and Olson, 2007), i.e., of their capacity to increase motivated behavior. This function has been investigated in one study of SA (Klucken et al., 2009); however, the OFC was not found to respond more to a conditioned stimulus predicting the presentation of VSS than to a CS predicting the presentation of control visual stimuli. Thirdly, the OFC probably represents the value of the expected outcomes of specific goal-directed actions (O Doherty, 2007). This has not been specifically studied in SA as paradigms have not used instrumental actions such as button presses to obtain delivery of rewards such as VSS. Although until now paradigms have not specifically targeted the desire aspects of SA, neuroimaging studies in other motivational domains suggest that those aspects could account for at least part of OFC activations. Is a part of the OFC specialized in processing sexual stimuli, with various parts of the OFC specialized in processing different categories of rewards, or are all types of rewards processed in the same region? Two spatial gradients have been proposed to answer this question (Kringelbach and Rolls, 2004). One is a mediolateral distinction, whereby medial OFC activity is related to monitoring the reward value of many different reinforcers, whereas lateral OFC activity is related to the evaluation of punishers. The second is a posterior anterior distinction with more complex or abstract reinforcers (such as monetary gains) represented more anteriorly in the OFC than simpler reinforcers such as taste. Fig. 1 shows that most foci lie in the posterior part of the OFC (posterior anterior axis) and not in the medial OFC (mediolateral gradient). In a test of the posterior anterior gradient hypothesis, Sescousse et al. (2010) have presented subjects with monetary rewards and with VSS and found that, while only monetary rewards activated bilaterally the anterior OFC (y = 51/54 mm, left/right), only VSS activated the posterior OFC (y = 33/33 mm). All foci were in an intermediary position along the mediolateral gradient (x = 30 mm). In addition, among areas activated only by VSS a cluster was also present in the medial OFC in an intermediate position along the posterior anterior gradient (y = 45 mm). In relation with the issue of regional specificity, the level of response to VSS in the right OFC was found to depend on testosterone, with higher responses in healthy subjects than in hypogonadal patients, and increased responses in hypogonadal patients in response to testosterone administration (Redouté et al., 2005). The location of this cluster was consistent with the Sescousse et al. s study (x = 30, y = 40 mm; Sescousse et al., 2010) Medial prefrontal cortex. Activation in the medial prefrontal cortex (mpfc), a region lying dorsal to the medial OFC, was reported in nine (24.3%) studies (Table 2; Fig. 1). In addition, the level of activation in the ventral mpfc correlated positively with ratings of pleasure (Sescousse et al., 2010), of self-relatedness of VSS (how much subjects perceived VSS as personally relevant; Heinzel et al., 2006), and with level of erectile response (Moulier et al., 2006). By contrast, in the dorsal mpfc, deactivation (Bocher et al., 2001; Tsujimura et al., 2006) and negative correlation with erectile response (Moulier et al., 2006; Mouras et al., 2008) were found. The activation of the mpfc has been related both to general emotional arousal (Karama et al., 2002; Walter et al., 2008b) and to a specific role in mediating the erectile response (Moulier et al., 2006). Sescousse et al. (2010) found that the degree of activation in the ventral mpfc correlated with the level of the participants hedonic experience, regardless of the type of reward presented, i.e., monetary gain or VSS. They related this finding with the proposition that, in order to evaluate and compare the relative value of different categories of rewards on a unique scale, the brain may use a common neural currency, likely to be implemented in integrative reward regions such as the ventral mpfc. Walter et al. (2008b) have suggested that dorsal mpfc activations are related to higher order processing of socially relevant stimuli (pictures of nude persons or sexual s), which could also be of particularly high self-relevance Insula. The insula was found activated by sexual stimuli in 23 studies (62.2%) (Table 2; Fig. 1). In addition, the level of activation of the insula was found correlated with the level of markers of SA penile tumescence or perceived SA (Arnow et al., 2002; Ferretti et al., 2005; Moulier et al., 2006; Mouras et al., 2008). Several functions have been related to these insular responses. Firstly, given the involvement of the insula in visceral sensory processing (Craig, 2003), the correlation between level of activation and penile response may reflect the role of the insula in the perceptual processing of penile inputs (Moulier et al., 2006) and possibly in the awareness of erection, a process that would be akin to the insular function of interoceptive awareness (Craig, 2003). This role of the insula in the sensory processing of penile inputs is consistent with the report that manual stimulation of the penis strongly activates the right posterior insula (Georgiadis and Holstege, 2005). In addition, the insula has been found to be a starting point of partial seizures with bilateral genital sensations (Stoffels et al., 1980). Furthermore, Moulier et al. (2006) focused on the temporal relationship between the BOLD signal from the insula and the penile response to photographic stimuli and found that the time courses of posterior and anterior insular responses were different with regard to penile tumescence. In the posterior insula, the BOLD signal changes were led by those of the plethysmographic signal, which is consistent with the idea that this region processes inputs from the penis. Conversely and secondly, the BOLD signal from both anterior insulae led penile responses by a 20-s lag. Similarly, in response to sexual video clips, the BOLD signal from both anterior insulae led penile responses (Mouras et al., 2008). Thus, both studies suggest a second role of the insulae in SA, i.e., that the anterior insulae play a role in mediating penile erection (Moulier et al., 2006; Mouras et al., 2008). This possibility is supported by studies where electrical stimulation in the anterior insula evoked marked and consistent autonomic responses, e.g., of diastolic blood pressure and of heart rate (Cechetto, 1994). Thirdly, the response of the right anterior insula to VSS was found to depend on the level of plasma testosterone (Park et al., 2001b; Redouté et al., 2005). Thus, activation was higher (i) in healthy men than in untreated hypogonadal patients and (ii) in hypogonadal patients under hormonal treatment than when untreated. In keeping with this modulatory function of testosterone, a study suggests that the insula is involved in the neural control of testosterone secretion (Banczerowski et al., 2001): in rats, right-sided lesions of the insular cortex resulted in a significant decrease in serum testosterone concentration. The existence of a neural connection between insula and testes not mediated by the hypothalamic-pituitary-testicular axis is supported by anatomical studies using the transsynaptic viral tracing technique, with insular neurons labeled after viral injection into the testis (Gerendai et al., 2000; Lee et al., 2002). Fourthly, it has been proposed that the anterior insula participates in determining the affective tone of experience and behavior (Mesulam, 1985). Damasio et al. (2000) have suggested that the insula holds a map of the internal milieu, viscera and musculoskeletal frame, from which it receives signals. They propose that this information is accessible to consciousness and contributes to generating conscious feelings. Accordingly, insular responses to VSS may mediate the emotional tone of SA. Finally, in domains other than sex, insular activity has been shown to be related to motivational level. Level of insular activity was correlated with hunger ratings (Morris and Dolan, 2001; Tataranni et al., 1999), thirst ratings (Denton et al., 1999b), and drug craving (Garavan, 2010). It has been hypothesized that the insula provides the conscious awareness of the aversive bodily feeling states that constitute the craving

16 1496 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) response (Garavan, 2010). Whether the insula plays a direct role in sexual desire in addition to the functions reviewed above remains an open question. A recent model may be helpful to account in a parsimonious way for the various insular functions reviewed above (Menon and Uddin, 2010). The anterior insula, along with the anterior cingulate cortex, is part of a salience network that functions to segregate the most relevant among internal and extrapersonal stimuli in order to initiate appropriate control signals to regulate behavior and homeostatic state Hypothalamus. Fourteen studies (37.8%) reported a hypothalamic response to VSS (Table 2; Fig. 1). Karama et al. (2002) and Paul et al. (2008) found that levels of perceived SA were correlated with the magnitude of hypothalamic BOLD signal. Similarly, Redouté et al. (2000) using PET, as well as Arnow et al. (2002) and Ferretti et al. (2005) using fmri, found that the hypothalamic response was correlated with the level of penile tumescence. In addition, Ferretti et al. (2005) studied the patterns of brain activation in different phases of the erectile response. The hypothalamus was more activated during the phase of rising erection than in the phase with no erection. By contrast, the hypothalamic activation was not significantly different when comparing the phase of sustained erection and the phase with no erection. However, the reverse findings were reported, with no hypothalamic response during rising erection (Tsujimura et al., 2006), but hypothalamic activation during maintained erection (Miyagawa et al., 2007). Finally, Savic et al. (2001) showed that, when smelling an estrogen-like substance, heterosexual men but neither women, nor homosexual men activated the hypothalamus, with clusters centered in the paraventricular and dorsomedial nuclei. The precise location of the response within the hypothalamus varies across studies (Fig. 1), which makes it likely that different hypothalamic regions are activated and that the hypothalamus contributes to different aspects of SA. However, specifying which hypothalamic nuclei are involved is tentative because of their small sizes and of the limitations of fmri spatial resolution related in particular to spatial smoothing. The activation may firstly be related to the process of erection. Specific nuclei where activation could be related to erection are the paraventricular and the dorsomedial nucleus. While the paraventricular nucleus has been repeatedly implicated in the control of erection in animals (e.g., Argiolas and Melis, 2005), in freely moving monkeys the stimulation of the dorsomedial hypothalamic nucleus elicits full erection and it is the only hypothalamic locus where radiostimulation leads not only to mounting and thrusting but also to ejaculation (Perachio et al., 1979). In several studies, the location of maximally activated voxels is compatible with activation in these nuclei (Fig. 1). Secondly, regarding the role played by the hypothalamus in sexual motivation, in rats the stimulation of the posterior hypothalamus induced copulatory behavior and bar pressing for the opportunity to copulate (Caggiula, 1970). In addition, the medial preoptic area has repeatedly been implicated in sexual motivation (e.g., Balthazart and Ball, 2007; Hull and Dominguez, 2006; Swanson, 2000). In an experiment performed in monkeys (Oomura et al., 1988), the rate of discharge of neurons in the medial preoptic area of an experimental male increased when he viewed a receptive female and performed button presses resulting in bringing the female close to him. Indeed, in many reviewed studies, the anterior location of activated clusters is compatible with an activation in the medial preoptic area (Fig. 1), which could reflect its role in the motivational component of SA. Finally, one study (Brunetti et al., 2008) has suggested that hypothalamic activation might be related to gender identity, in keeping with recent neuroanatomical studies (Savic et al., 2010) Dorsal striatum. Nineteen studies (51.4%) have reported an activation in the caudate nucleus (n = 15; 40.6%) and/or in the putamen (n = 11; 29.7%) in response to the presentation of sexual stimuli (Table 2; Fig. 1). While the dorsal striatum has been traditionally associated with motor processes, evidence for its role in motivated behavior is mounting (Delgado, 2007). In order to understand the functional significance of the activation of the dorsal striatum, it is important to note that in the experimental paradigms used in these studies, no overt motor response was possible. Therefore, it is the premotor aspects of responses, as well as responses of regions concerned with withholding behavior, that may have been revealed by these studies. According to a model of basal ganglia function in motivated behavior (Rolls, 1999), once the neurons in the OFC have decoded the motivational significance of stimuli (see above), it is essential that these reward-related signals should not be interfaced directly with motor behavior. Instead, what is required is that the signals enter an arbitration mechanism, which takes into account the cost of obtaining reward. It has been proposed that the basal ganglia participate in this function (Rolls, 1999). They receive inputs from numerous cortical areas that compete within the caudate nucleus for behavioral output, and this nucleus maps each particular type of input to the appropriate behavioral output, implemented via the return basal ganglia connections to premotor/prefrontal cortex. This model is consistent with animal studies showing that lesions of the dorsal striatum impair inhibitory control (e.g., Eagle and Robbins, 2003) and with neuroimaging studies, which demonstrate an activation of the putamen and/or the caudate nucleus in paradigms where the need for a motor response is conflicting with the need to withhold it (Pardo et al., 1990) and an activation of the caudate nucleus upon volitional tic suppression in Tourette Syndrome (Peterson et al., 1998). Finally, this model is supported by clinical evidence, such as hypersexuality in a patient with lesions circumscribed to the head of the caudate nuclei (Richfield et al., 1987). The above development shows that any model of the neural correlates of SA should include a component consisting in the control of the motor expression of SA (Bancroft, 1999). Secondly, motor imagery is commonly reported by subjects presented with VSS and is known to be related to striatal activation (Szameitat et al., 2007). Thirdly, the involvement of dorsal striatum in motivation and reward processing, shown in domains other than sexual behavior, might also account for its response to VSS (Delgado, 2007). VSS are experienced as rewards in themselves (Moulier et al., 2006; Redouté et al., 2000). Furthermore, the presentation of erotic pictures generates the desire to watch additional similar material, a desire that is indeed fulfilled in block presentations. Similarly, the dorsal striatum is activated during reward anticipation (Delgado, 2007), possibly reflecting the craving associated with such anticipation. Finally, the putamen has been repeatedly implicated in the mechanism of erection. A correlation was found between the level of activation recorded in the putamen and the magnitude of penile tumescence (Arnow et al., 2002; Mouras et al., 2008; Redouté et al., 2000). Similarly in non-human primates, electrical stimulation of the putamen evoked most often an erection and/or genital manipulation in Macaca Mulatta (Robinson and Mishkin, 1968) and prolonged erection was reported after micro-injection of bicuculline in the medial putamen (Worbe et al., 2009). By contrast, the right ventral putamen was found activated in the plateau phase of erection (Miyagawa et al., 2007) but not in the excitation phase with rising erection (Tsujimura et al., 2006) Ventral striatum. In response to VSS, the ventral striatum was found activated in 10 studies (27.0%) (Table 2; Fig. 1). In addition, the level of activation in the ventral striatum was found to be

17 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) correlated with the degree of perceived SA (Redouté et al., 2000; Walter et al., 2008b). Two main interpretations of this activation have been proposed. Firstly, it has been conceived as being the neural representation of received reward, as VSS are experienced as rewards in themselves (Sabatinelli et al., 2007; Sescousse et al., 2010; Walter et al., 2008b). Secondly, activation of the ventral striatum has been related to the incentive aspects of VSS and to anticipated reward (Ponseti et al., 2006). But what exactly does reward anticipation mean in terms of subjective experience? It may mean the cognitive operation whereby a subject knows that a specific rewarding outcome will occur in the future. It may also denote the craving or desire associated with such knowledge. Interestingly, activation of the ventral striatum increases proportionally to the magnitude of anticipated monetary reward (Haber and Knutson, 2010), which supports the notion that the desire associated with anticipation is represented in the ventral striatum Thalamus. Thalamic activation was recorded in response to VSS in 19 studies (51.4%) (Table 2; Fig. 1). Most of these activations lay in the medial part of the thalamus. In addition, a positive correlation between level of thalamic activation and magnitude of erection was reported in four studies (Ferretti et al., 2005; Moulier et al., 2006; Mouras et al., 2008; Redouté et al., 2000). Thalamic activation has been understood as a correlate of the general emotional arousal that accompanies SA (Karama et al., 2002; Walter et al., 2008b; Metzger et al., 2010). A review of fmri studies using monetary incentive delay tasks indicated that anticipation of reward vs. anticipation of punishment did not clearly elicit differential dorsomedial thalamic activation, which suggests that dorsomedial thalamic activation reflects general arousal to a greater extent than value, i.e., either rewarding or punishing (Knutson and Greer, 2008). Consistent with this view, the mediodorsal nucleus of the thalamus is part of basal gangliathalamo-cortical circuits (Alexander et al., 1990): the amygdala, the OFC, the hippocampus, the temporal association cortex and the anterior cingulate cortex project to the ventral striatum, which has connections through the ventral pallidum to the mediodorsal nucleus of the thalamus, which in turn projects to the prefrontal and cingulate cortices. The dynamic of these areas is believed to result in reward-dependent selection of particular actions (Doya, 2008). By contrast, some studies indicate that thalamic activation mediates specifically sexual responses. Firstly, in a high-field fmri study using emotional pictures as control stimuli, right paraventricular thalamic activation including mediodorsal, laterodorsal, and parataenial nuclei was induced by both expectancy and presentation of VSS (Metzger et al., 2010). Conversely, in hamsters lesions to the mediodorsal thalamic nucleus and paraventricular thalamic nucleus resulted in inappropriate and inefficient precopulatory and copulatory behavior (Sapolsky and Eichenbaum, 1980). Secondly, as the level of activation of foci in the ventrolateral part of the thalamus was correlated with level of erection (Moulier et al., 2006; Redouté et al., 2000), it has been suggested that this part of the thalamus may be involved in the perception of erection. Such involvement of the thalamus in erectile processes is consistent with studies in monkeys, where erection was elicited by electrical stimulation of the thalamic tubercle, of the rostral pole (MacLean and Ploog, 1962), of the mediodorsal nucleus (MacLean et al., 1963) and of midline thalamus (Robinson and Mishkin, 1968) Anterior cingulate cortex. A majority of papers (n = 25; 67.6%) have reported an activation of the anterior cingulate cortex (ACC) in response to sexual stimuli and/or a positive correlation of the BOLD level in the ACC with the degree of penile tumescence or with the rating of perceived SA (Table 2; Figs. 1 and 2). Responses were distributed along the full extent of the ACC (Fig. 2), suggesting that they mediate distinct functions. Hereunder, we discuss the possible implication of the ACC in the affective, premotor and autonomic components of SA. A useful parcellation of the ACC into a caudal cognitive and a rostral affective division has been proposed (Bush et al., 2000). The latter division is primarily involved in assessing the salience of emotional and motivational information and in the regulation of emotional responses. The rostral cingulate responses to VSS (Arnow et al., 2002; Ferretti et al., 2005; Hamann et al., 2004; Paul et al., 2008; Ponseti et al., 2006; Rauch et al., 1999; Redouté et al., 2000) lay in the affective division and were understood as correlates of emotionally or motivationally relevant information processing (Ferretti et al., 2005; Rauch et al., 1999; Redouté et al., 2000). Moreover, the level of activation of the perigenual ACC was correlated with the degree of self-relatedness of VSS, i.e., to how much subjects felt they associated with VSS (Heinzel et al., 2006). An activation in the affective division was found in a study aiming to identify regions associated with desire, irrespective of the category or object of desire (events, objects and persons) (Kawabata and Zeki, 2008). Furthermore, studies of thirst (Denton et al., 1999b) and hunger (Tataranni et al., 1999) also reported an activation in the rostral ACC. Premotor areas are located in the region identified as the cognitive division (Picard and Strick, 2001). The role of the caudal ACC in motor function is known to be similar to the role of premotor and supplementary motor area cortices (Dum, 1993). In the monkey, stimulation of the ACC elicits genital manipulation of a masturbatory character (Robinson and Mishkin, 1968). In human partial seizures with motor sexual manifestations (such as pelvic thrusting), the origin of discharge has been located in the ACC (Landré et al., 1993). The caudal ACC response to VSS (Kim et al., 2006; Moulier et al., 2006; Mouras et al., 2008; Redouté et al., 2000; Safron et al., 2007) is thus consistent with the previous general conclusion that the caudal ACC plays a crucial role in the initiation of goal-directed behaviors (Devinsky et al., 1995), which includes sexual behavior. However, in the neuroimaging paradigms used to study SA, where the urge to act conflicts with the instruction to withhold any overt behavior, there must be control mechanisms at least as powerful as the motivational processes. Activation in the caudal ACC was found in a study of a GO/NO-GO task (Kawashima et al., 1996). More generally, the ACC is activated when there is conflict between possible responses (Carter and van Veen, 2007). According to Picard and Strick (2001), a specific part of the cingulate sulcus [located at y = 24 mm ± 7 mm (mean ± standard deviation, in Talairach coordinates)] appears to be involved in conflict monitoring. Interestingly, several papers have reported activations in this area (Arnow et al., 2002; Karama et al., 2002; Moulier et al., 2006; Safron et al., 2007; Stoléru et al., 1999; Sundaram et al., 2010), which suggests that these activations may correspond to conflict monitoring. Accordingly, we propose that the activation of the caudal ACC in response to VSS results from conflicting inputs to this area: on the one hand, inputs of the GO type, correlated with the perceived urge to enact SA, and, on the other hand, inputs of the NO-GO type, correlated with the perceived need to withhold any overt sexual behavior in the current circumstances. Both types of inputs would be associated with activation, because it is the local synaptic activity that is energy consuming. In spite of the evidence for a role of the caudal ACC in the premotor component of SA, one may also ask whether activation of ACC was related to SA per se or to other psychological phenomena elicited by sexual stimuli. For instance, the caudal part of the ACC has been implicated in selective attention, working memory, semantic and episodic memory retrieval (Cabeza and Nyberg, 1997). Some of these cognitive processes may have participated in

18 1498 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) the processing of sexual stimuli without being specifically triggered by the sexual character of processed stimuli. In addition, Brodmann areas 24 and 32 of the ACC are involved in modulating autonomic functions, including erection (Dua and MacLean, 1964; Robinson and Mishkin, 1968). The magnitude of the erectile response was linearly correlated with the level of activation in both the caudal and the rostral ACC, including the subgenual part (Arnow et al., 2002; Ferretti et al., 2005; Moulier et al., 2006; Mouras et al., 2008; Redouté et al., 2000). Finally, a pharmacological study in healthy males has suggested that sexual dysfunction, a common side-effect of selective serotonin-reuptake inhibitors (SSRIs), could be related to decreased activation of ACC. Compared with placebo, while ratings of subjective sexual dysfunction increased under paroxetine, activation decreased in subgenual, pregenual and midcingulate cortices in response to VSS (Abler et al., 2011) Premotor areas. Intuitively, as SA is related to the desire of performing sexual actions, one would expect that premotor cortical areas (supplementary motor area, ventral premotor area) be activated. Indeed, 18 studies (48.6%) have reported an activation of premotor areas and/or a correlation between the level of activation of premotor areas and the degree of plethysmographic responses (Table 2; Fig. 1). By contrast, Miyagawa et al. (2007) reported a deactivation of the supplementary motor area. The ventral premotor area and the supplementary motor area are known to have distinct functions. A variety of experiments indicate that the lateral premotor cortex to which the ventral premotor area belongs uses information from other cortical regions to select movements appropriate to the context of the action (Purves et al., 2001). Thus, when a monkey is trained to reach in different directions in response to a visual cue, the appropriately tuned lateral premotor neurons begin to fire at the appearance of the cue, well before the monkey receives a signal to actually make the movement. Rather than directly commanding the initiation of a movement, these neurons appear to encode the monkey s intention to perform a particular movement; thus, they seem to be particularly involved in the selection of movements based on external events. In subjects presented with VSS, activation in the ventral premotor area may reflect externally triggered preparation of movements. The medial premotor cortex, to which the supplementary motor area belongs, also mediates the selection of movements. However, this region appears to be specialized for initiating movements specified by internal rather than external cues. In contrast to lesions in the lateral premotor area, removal of the medial premotor area in a monkey reduces the number of self-initiated or spontaneous movements the animal makes, whereas the ability to execute movements in response to external cues remains largely intact (Shima and Tanji, 1998). Imaging studies suggest that in humans this cortical region functions in much the same way. For example, PET scans show that the medial region of the premotor cortex is activated when the subjects perform motor sequences from memory (i.e., without relying on an external instruction) (Grafton et al., 1998). Alternatively, the activation of premotor areas has been interpreted as the neural correlate of sexual motor imagery experienced by subjects during the presentation of VSS (Moulier et al., 2006; Mouras et al., 2003; Stoléru et al., 2003). Finally, it has been proposed that the activation of the ventral part of the lateral premotor cortex reflects the increased activity of mirror neurons (Bocher et al., 2001; Mouras et al., 2008; Ponseti et al., 2006; Stoléru et al., 1999). Mirror neurons are a particular class of visuomotor neurons, originally discovered in area F5 of the monkey premotor cortex, that discharge both when the monkey performs a particular action and when the monkey observes the same action performed by another animal (monkey or human) (Rizzolatti and Craighero, 2004). The theory of the perception-action coupling mechanism based on mirror neurons has been expanded to the domain of emotion to account for emotion processing and empathy. Furthermore, it has been suggested that a similar perception action coupling mechanism mediated by the mirror-neuron system prompts the observers of VSS to resonate with the motivational state of other individuals appearing in visual depictions of sexual s, with observers activating motor representations associated with the observed depictions (Mouras et al., 2008). This interpretation is reinforced by the fact that the presentation of VSS induces also an activation of the inferior parietal lobule, another brain region that contains neurons belonging to the mirror-neuron system Cerebellum. Cerebellar activation and/or a correlation of cerebellar activation with a marker of SA were recorded in 14 studies (37.8%) (Table 2; Fig. 1). Although the cerebellum is essentially involved in motor control, many studies indicate that it also contributes to cognitive (Ivry and Fiez, 2000) as well as to emotional and motivational (Aalto et al., 2002; Parsons et al., 2000) processes. The precise location of responses within the cerebellum varied greatly across studies of SA, so that these responses likely correspond to several distinct functional processes. As stated by Hu et al. (2008), possible interpretations of the cerebellar response are that it might be related to the emotional, motivational, and/or motor imagery processes induced by sexual stimuli, three interpretations that we consider hereunder. It has been suggested that the cerebellar response could be related to the feeling experience associated with SA (Beauregard et al., 2001). According to a meta-analysis (Stoodley and Schmahmann, 2009), emotional processing is associated with the activation of three distinct parts of the cerebellum, i.e., left Crus I, right lobule VI, and left lobule VIIAt. Indeed, among reported cerebellar responses to VSS, several fell within these regions (Beauregard et al., 2001; Hu et al., 2008; Safron et al., 2007; Sundaram et al., 2010; Tsujimura et al., 2006). Regarding studies on motivation in domains other than sexuality, cerebellar activation has been reported as a neural correlate of the sensation of thirst, hypercapnia and hunger for air (Parsons et al., 2001), and hunger for food (Tataranni et al., 1999). There is considerable similarity among regions of cerebellar involvement across hunger, thirst, and air hunger, with a consistent response of the quadrangular lobules (Parsons et al., 2001). Many studies (Bühler et al., 2008; Hu et al., 2008; Meseguer et al., 2007; Mouras et al., 2008; Redouté et al., 2000; Safron et al., 2007; Tsujimura et al., 2006) did also find a response to VSS in the quandrangular lobules [lobules IV, V and VI in Schmahmann s anatomical nomenclature (Stoodley and Schmahmann, 2009)]. It has also been suggested that the cerebellar response was related to induction of erection (Tsujimura et al., 2006), which is consistent with the correlation between the level of cerebellar activation and the degree of erection (Moulier et al., 2006; Mouras et al., 2008; Redouté et al., 2000). Finally, the cerebellum has been shown to be part of the neural network mediating motor imagery (Buccino et al., 2006), with activations in Crus I. Subjects presented with VSS commonly report motor imagery with sexual content on debriefing (Moulier et al., 2006; Stoléru et al., 2003), a phenomenon which might be related with cerebellar activation. Crus I has been reported to respond to VSS in studies of SA (Beauregard et al., 2001; Sundaram et al., 2010). Because the lowest parts of the cerebellum are located very low under the bicommissural line, they may not have been covered by all studies. Among 25 studies reporting no cerebellar activation (Table 2), two scanned the uppermost part of the cerebellum and eight did not provide enough information to determine whether the cerebellum was covered. Therefore, activations in the lower part of the cerebellum may have been overlooked by available studies.

19 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Midbrain. Midbrain activation and/or a correlation of midbrain activation with a marker of SA were recorded in 10 studies (27.0%) (Table 2; Fig. 1). Bocher et al. (2001) reported that rcbf in the midbrain was positively correlated with the levels of perceived SA and perceived erection. They related this finding to ventral midbrain dopaminergic neurons, located in the substantia nigra and ventral tegmental area, which give rise to mesolimbic efferents and play a key role in motivation. In line with this interpretation, Abler et al. (2011) reported that healthy males under placebo treatment showed midbrain activation in response to VSS, which decreased under paroxetine, an SSRI drug. They suggested that the latter effect reflected an SSRI-induced inhibition of dopaminergic reward circuits. By contrast, compared with paroxetine, bupropion, a dopamine and norepinephrine reuptake inhibitor, increased midbrain activation (Abler et al., 2011). In the dorsal midbrain, activation was found related to general emotional arousal as opposed to specific SA (Walter et al., 2008b). This region is a station between the hypothalamus and the autonomic nervous system; in particular, the periaqueductal grey matter is involved in the autonomic component of SA in healthy subjects (Walter et al., 2007). To sum up, the most consistently activated set of regions found related to SA in healthy men were the lateral occipital and/or lateral temporal cortices (83.8% of studies), the ACC (67.6%), the insula (62.2%), the parietal cortex (56.8%), the inferotemporal cortex (54.1%), the thalamus (51.4%) and the frontal premotor areas (48.6%). Regions inconsistently found activated were the mpfc (24.3%), the claustrum (27.0%), the ventral striatum (27.0%), the midbrain (27.0%) and the putamen (29.7%). In between, were found the caudate nucleus (40.6%), the hypothalamus (37.8%), the OFC (37.8%), the cerebellum (37.8%), and the amygdala (35.1%) Potential control mechanisms. Control mechanisms have been directly studied (Beauregard et al., 2001) or inferred from patterns of response to VSS (Redouté et al., 2000). Individuals have to control SA both to refrain from expressing sexual behavior in inappropriate contexts and to express sexual behavior acceptable to a potential partner. Paradigms used to study SA include an element of control since subjects are explicitly instructed to refrain from moving. In addition, experimental evidence supports the notion that the non-sexually aroused state, in which subjects are at the start of the experiments, may result from an active sustained inhibition of brain regions which mediate the unfolding of SA. If this is true, paradigms suppose alleviating inhibition from these regions, making development of SA possible. Deactivation in a given region in response to VSS suggests that, in the absence of sexual stimulation, this region exerts a continuous, inhibitory control on SA and that the development of SA requires the release of such inhibition. Indeed, areas in the lateral occipito-temporal cortex and in the inferior temporal cortex have been found to respond to VSS by a deactivation (Table 2). Similarly, the inferotemporal cortex was found bilaterally deactivated in response to manual stimulation of the penis by the female partner (Georgiadis and Holstege, 2005). These results are consistent with the dramatic hypersexuality after the removal of temporal lobes in monkeys (Klüver and Bucy, 1939) and humans (Devinsky et al., 2010). Thus, whereas the dorsal striatum may withhold the behavioral expression of an already current state of SA (see above), some temporal regions could exert a tonic inhibition on the initiation and development of SA. These deactivated temporal regions (BA 20, BA 21, BA 22, BA 39) were distinct from the temporooccipital regions (BA 19, BA 37) found activated in response to VSS. Interestingly, regions proposed to mediate inhibition of SA have been found activated in tasks involving evaluative processes of guilt and embarrassment (Berthoz et al., 2002; Takahashi et al., 2004). The gyrus rectus has also been implicated in the inhibition of SA (Stoléru et al., 2003). Indeed, all deactivations of the OFC lay in the gyrus rectus (Bocher et al., 2001; Georgiadis et al., 2010; Redouté et al., 2000). By contrast, OFC activation was found in more lateral parts of the OFC, except in one study (Safron et al., 2007). Consistent with an inhibitory role of the gyrus rectus, in patients with hypoactive sexual desire disorder there was an abnormally maintained activity of this region in response to VSS (Stoléru et al., 2003) and, just as in the case of temporal regions, the gyrus rectus was activated by tasks requiring moral judgments (Moll et al., 2002). Clinically, patients with lesions in this area are often tactless, lacking in social restraints, and present with excessive pleasure-seeking behaviors, especially in the sexual domain (Miller et al., 1986). Finally, the lateral part of the left OFC has been proposed to play an inhibitory role on SA (Redouté et al., 2005). Whereas the right OFC is activated in response to VSS and is thought to play a role in the assessment of the sexual relevance of stimuli (see above), several studies found a deactivation of the lateral part of the left OFC (Table 2). Similarly, in this area the BOLD signal was negatively correlated with the concurrent penile plethysmographic signal (Moulier et al., 2006). In addition, ejaculation is associated with a decreased rcbf in this area (Georgiadis et al., 2007). Importantly, Beauregard et al. (2001) demonstrated an activation of this region in men instructed to inhibit any emotional reaction to VSS. Furthermore, in a study focused on healthy males (Redouté et al., 2000), the increased plasma testosterone in response to VSS was negatively correlated with rcbf in the left orbitofrontal gyrus (unpublished results). This finding suggested that increasing testosterone levels in response to VSS might enhance SA by reducing the inhibiting action of the left lateral OFC. Indeed, the deactivation of the left OFC in response to VSS was subsequently shown to depend on the level of plasma testosterone: whereas a deactivation was observed in healthy controls, rcbf did not change significantly in untreated hypogonadal patients (Redouté et al., 2005). Thus, in healthy men, VSS coupled with a normal or with an increasing level of testosterone could allow for the release of inhibitory control from this region, whereas in hypogonadal patients inhibitory control would persist. That administration of testosterone to hypogonadal patients restored deactivation in the left OFC in response to VSS (Redouté et al., 2005) is consistent with this negative neuromodulatory role. Thus, several findings suggest that the left OFC might exert a testosterone-dependent inhibitory tonic control on SA, and that this control decreases upon visual sexual stimulation. Finally, like temporal regions mentioned above, the left lateral OFC was activated in response to tasks involving evaluative processes of guilt and embarrassment (Berthoz et al., 2002; Takahashi et al., 2004) Comparison between homosexual and heterosexual men. Five fmri studies have investigated the neural correlates of SA in homosexual men (Table 2). Paul et al. (2008) used a block-design fmri paradigm to compare 12 heterosexual with 12 homosexual men. Only VSS corresponding to the subjects sexual orientation induced an activation pattern characteristic for SA in both groups, with activation in the lateral occipital cortex, inferior temporal cortex, parietal cortex, mpfc, caudate nucleus and hypothalamus. While the pattern of activation was similar across groups, only heterosexual subjects showed activation in the right insula, left caudate nucleus, and bilaterally in the ACC, whereas an activation in the left OFC (in a median location along the mediolateral gradient) was demonstrated only in homosexual participants. However, the statistical comparison of responses of heterosexual men to videos of heterosexual couples with the responses of homosexual men to videos of homosexual couples showed differential activation only when the threshold was decreased to a lenient p < 0.05, uncorrected for multiple comparisons. At this threshold, the only regions of interest that showed differential activation were the hypothalamus

20 1500 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) and right medial OFC in the contrast [heterosexuals homosexuals], as well as the parietal cortex and the left OFC in the contrast [homosexuals heterosexuals]. These findings suggest that the neuroanatomical correlates of SA are quite similar in hetero- and homosexual males, at least as shown by fmri. Ponseti et al. (2006) reached the same conclusion. Similarly, Hu et al. (2008) used clips depicting heterosexual and homosexual couples presented in a block-design paradigm to compare ten homosexual with ten heterosexual men. Clips corresponding to the subjects orientation induced in both groups an activation of the lateral temporal cortex, superior parietal lobule, insula, thalamus, anterior cingulate cortex, premotor areas and vermis (Table 2). Only in homosexual men were the left angular gyrus, left caudate nucleus, and right pallidum found activated. Conversely, only heterosexual men showed activation in the bilateral lingual gyrus, right hippocampus, and right parahippocampal gyrus. Authors concluded that different neural circuits may mediate SA in homosexual and heterosexual men. However, this conclusion is questionable as no whole-brain analysis of the Condition (Sexual Stimuli, Control Stimuli) by Group (Heterosexual, Homosexual) was performed. Moreover, in two regions of interest (amygdala and thalamus), a comparison of the fmri signal intensity change failed to show differences between homosexual and heterosexual men. Using an event-related paradigm and photographs instead of clips, Safron et al. (2007) compared 11 hetero- and 11 homosexual males on their brain responses to pictures corresponding or not to their sexual orientation. In the whole-brain analysis, differential neural activation was not found. In the analysis of regions of interest, homosexual men showed higher amygdalar activation to male than to female sexual stimuli, while heterosexual men showed similar levels of amygdalar activation across male and female stimuli. Ponseti et al. (2009) used fmri to evaluate whether the spatial response pattern to sexual stimuli could be used to predict heterosexual vs. homosexual orientation in individual subjects. In an event-related study, participants were exposed to pictures of samesex and opposite-sex aroused genitals. Two statistical classification methods performed well in predicting individual sexual orientation with a mean accuracy of >85%. Similarly, analysis of fmri data distinguished with high accuracy between pedophiles and healthy males based on their activation pattern in response to actual and expected presentation of VSS (Walter et al., 2010) Meta-analysis The 21 studies selected (N = 235 subjects; 715 foci) are marked with an asterisk in Table 1. Regions showing consistent activation across studies are presented in Table 3. In the right and left lateral occipital cortices, meta-analysis distinguished an anterolateral and a posteromedial cluster. Similarly, for both insulae it confirmed the distinction between an anteroventral cluster and a posterodorsal cluster. All regions found by the meta-analysis had been identified in the review conducted beforehand. Conversely, three regions identified in the review were not found in the meta-analysis: (i) the right OFC cluster (24 mm 3 ) did not pass the 200 mm 3 limit (see Section 2); (ii) although no voxel with peak ALE value lay in the putamen, a large part of cluster #4 did cover the right putamen; (iii) no voxel with statistically significant ALE value was found in the ventral striatum. Two reasons may explain this negative finding: among 13 studies reporting ventral striatal activation, seven did not meet inclusion criteria for meta-analysis; the remaining studies reported foci widely scattered across the nucleus accumbens and the ventral putamen. In addition to regions reported in a previous meta-analysis (Kühn and Gallinat, 2011), the present analysis found, on the right side, the claustrum, medial prefrontal cortex, and caudate body; bilaterally, the inferior frontal gyrus, substantia nigra, and, medially, the cerebellar vermis. These new findings in the current meta-analysis are likely due to the larger set of studies and to the conjoint meta-analyses of subtractive and correlational analyses. With recent developments of GingerALE, similar analyses on the same group can be run within one meta-analysis with no increased risk of false positive findings (Turkeltaub et al., 2012) Healthy female volunteers A few studies have targeted functional brain correlates of SA in healthy women (Tables 1 and 2) (Arnow et al., 2009; Georgiadis et al., 2006; Gizewski et al., 2006; Karama et al., 2002; Park et al., 2001a; Rupp et al., 2009; Zhu et al., 2010). Brain regions that respond to VSS in women are largely the same as those that respond in men (Table 2). The specificity of studies on women is the potential influence of the phase of the menstrual cycle on these brain responses. In three fmri studies, heterosexual women were scanned in a specific period of the cycle. Firstly, in 20 females scanned outside the ovulatory period, erotic clips induced activations bilaterally in the occipitotemporal, orbitofrontal, medial prefrontal, insular and anterior cingulate cortices, as well as in the ventral striatum and amygdalas (Karama et al., 2002). Secondly, Arnow et al. (2009) used fmri to study 20 females during the luteal phase or 7 20 days following the first day of last menses for those on oral contraceptives. Video stimuli included erotic, sports, and relaxing segments. Vaginal vasocongestion, an index of physical SA, was measured with a vaginal photoplethysmograph. In addition, subjects used a potentiometer to continuously rate, for each segment, their level of subjective SA. Arnow et al. (2009) replicated only part of the findings of Karama et al. s (2002) study (Table 2). Correlations between subjective ratings of SA and regional BOLD signal were found in the left anterior cingulate gyrus, left amygdala, right claustrum and bilaterally in the middle occipital, inferior occipital and fusiform gyri. Finally, in no region did the authors find a correlation between the vaginal photoplethysmograph measure and the regional BOLD signal. This negative finding was consistent with the low correlation between the vaginal photoplethysmograph measure and the subjective ratings of SA. Thirdly, in an investigation of putative human pheromones, Zhou and Chen (2008) showed that in the periovulatory phase the right OFC and right fusiform cortex were more activated in response to the sweat previously collected in heterosexual men experiencing VSS-induced SA than to the sweat from the same men being presented with emotionally neutral clips. Three fmri studies have compared brain responses across different phases of the menstrual cycle. Gizewski et al. (2006) compared the brain responses to sexual clips in the same women scanned in the ovulatory period and during their menses. Whereas in the ovulatory period higher activations were recorded in the right caudal ACC, left lateral OFC and left insula, no area was more activated during the menses. The higher responses in the ovulatory period could be related to the higher SA ratings recorded in that period (p = 0.054). Indeed, when the level of SA was used as a confounding variable in the comparison, the response of the right ACC was no longer different across the two cycle phases. However, responses of left OFC and insula remained significantly different. In the second study (Rupp et al., 2009), ten women were scanned in the late follicular phase and in the luteal phase. When presented with photographs of male faces, participants were requested to judge how likely they would be to have sex with the person represented on each picture. Pictures of houses were presented as control stimuli and women were asked to evaluate the houses for how likely they would be to rent them. The phase of the menstrual cycle had no significant effect on ratings of faces and houses. However, in response to photos of male faces activation was higher in the right medial OFC during the follicular phase. In addition, linear regression analyses demonstrated a positive correlation between activation

21 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Table 3 Results of meta-analysis of studies of heterosexual men. Cluster Brain region BA Side x y z Volume (mm 3 ) 1 Middle occipital g. BA19 R Middle occipital g. BA19 R Fusiform g. BA 37 R Inferior occipital g. BA 19 R Middle occipital g. BA19 L Inferior occipital g. BA 19 L Amygdala R Claustrum R Insula BA 13 R Insula BA 13 R Thalamus, dm R Thalamus, dm R Thalamus, vl R Middle occipital g. BA18 L Middle occipital g. BA19 L Middle occipital g. BA19 L Superior parietal g. BA 7 L Superior parietal g. BA 7 L Hypothalamus R Hypothalamus L Middle occipital g. BA19 R Amygdala L Superior parietal g. BA 7 R Inferior frontal g. BA 44 R Precentral g. BA 6 R Inferior parietal g. BA 40 L Thalamus L Ant. cingulate g. BA 32 R Ant. cingulate g. BA 24 L Ant. cingulate g. BA 24 R Insula BA 13 L Substantia Nigra R Fusiform g. BA 37 L Caudate body R Insula BA 13 L Inferior frontal g. BA 47 L Insula BA 13 R Cerebellum, vermis L Substantia Nigra L Postcentral g. BA 2 L Medial prefrontal g. BA 9 R Superior frontal g. BA 6 L Notes: Threshold: p < 0.05, FDR-corrected; x, y, z coordinates (MNI space) refer to the voxels with maximum ALE values in the clusters. Abbreviations: BA: Brodmann area; dm: dorsomedial nucleus; g.: gyrus; L: left; R: right; vl: ventrolateral nucleus. in the right medial OFC in response to male faces and the estradiol to progesterone ratios. Although the cycle phase effect was partly confounded with an order effect (for eight women the first scan took place in the luteal phase), this result suggests that ovarian steroids modulate the response of the medial OFC to sexually relevant stimuli. This result is consistent with reports that in men the level of response of the OFC to VSS was modulated by testosterone (see above). However, as there was no correlation between OFC activation and ratings of male faces, the psychological significance of the steroidal modulation of the OFC in women remains to be determined. Within this same study, attributes of faces were varied so that they appeared (i) more or less masculine (using morphing software) and (ii) linked with more or less risky partners, by associating faces with information about the men s supposed number of previous sexual partners and their use of condoms. Evaluations of likelihood of having sex with the men presented did not differ by menstrual cycle phase or masculinity of the faces. However, women demonstrated differential neural responses across the phases of the cycle: Whereas in the follicular phase activation was greater in response to masculinized than feminized faces in the bilateral ACC, bilateral inferior parietal lobule, and left precentral gyrus, in the luteal phase activation was greater only in the posterior cingulate cortex. While plasma progesterone levels were negatively correlated with activation in the right precentral gyrus, free testosterone levels positively predicted activation in the anterior cingulate cortex. Women reported that they would be more likely to have sex with low- compared to high-risk men. Compared with stimuli depicting high-risk men, those depicting low-risk men elicited stronger activation in the ACC. In the third study (Zhu et al., 2010), 17 women s responses to film excerpts were investigated with fmri three times in one menstrual cycle: during ovulation, menstruation, and at one other time at their convenience (follicular phase, n = 6; luteal phase, n = 11). No significant differences were found between brain responses in the menstruation phase and in the luteal or follicular phases. By contrast, compared with ovulation, higher responses were observed in the menstrual phase and in the rest of the cycle in the right inferior frontal gyrus, right lateral occipital cortex, and bilateral superior parietal lobules. These results differ from those of Gizewski et al. (2006) who found higher activation in the ovulatory period, albeit in other areas. According to Zhu et al. (2010), these discrepancies could be related to the various methods of ascertaining time of ovulation [luteinizing hormone urine test paper in Zhu et al. (2010) vs. calendar method in Gizewski et al. (2006)]. In a study grounded on the parental investment theory (Trivers, 1972), Roberts et al. (2008) investigated the influence of the phase of the cycle and of the gender of stimuli on the cerebral correlates of a response inhibition task. Women were requested to press a

22 1502 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) button ( Go ) for each presented picture of a series of attractive male or female faces, but to withhold button press ( No Go ) if the same stimulus was shown on two successive trials. There was an between cycle phase and gender of stimulus in the right inferior frontal gyrus, a region involved in inhibitory behavioral control: whereas in the follicular phase this region showed a lower No-Go related activation in response to male than to female faces, the reverse was true in the luteal phase. Authors suggested that this reduced activation might reflect greater efficiency in inhibitory brain function in the follicular phase, a potentially fertile period, and that such increased efficiency in response to male stimuli was adaptive in enabling females to be selective and cautious when committing to a sexual encounter in this period Comparison between healthy men and women A common presumption in society and the media is that men respond more strongly to VSS than do women (Rupp and Wallen, 2008). It is estimated that of the 40 million adults who visit pornography websites annually, 72% are male while only 28% are female ( accessed ). Eight fmri studies and two PET studies (Table 4) have compared females and males on their brain responses to sexual stimuli. A large number of brain regions displayed similar activation in men and women. This is the case bilaterally for the OFC, the mpfc, the ACC, the insula, the parietal cortex, the occipitotemporal cortex, and the ventral striatum (Hamann et al., 2004; Karama et al., 2002). In a study on male and female heterosexual and homosexual volunteers based on the presentation of photographs of sexually aroused genitals, Ponseti et al. (2006) searched for regions where the BOLD response to preferred and non-preferred sexual stimuli (pooled together) as compared with nonsexual stimuli (like landscapes, humans in sports, etc.) was influenced by the biological sex of the observer regardless of sexual orientation. No differential effects of the sex of the observer on the regional brain responses were found. However, several studies have found regions differentially activated across genders (Table 4). In Karama et al. s study, only in males were the hypothalamus and the thalami found activated by erotic clips (Karama et al., 2002). Moreover, when regional responses were compared across genders the hypothalamus was more activated in males. Importantly, however, the mean level of reported SA was also higher in males than in females. Thus, the higher hypothalamic activation might only have reflected that visual stimuli were more appropriate to induce SA in males. Indeed, when the rating of perceived SA was used as a confounding variable in the analysis, hypothalamic responses did not differ across genders. Similarly, in a study where females and males did not differ on their ratings of SA in response to photographs (Walter et al., 2008b), an analysis contrasting the regional responses to VSS and to nonsexual emotional stimuli did not reveal differential responses across genders. Opposite results were reported by Hamann et al. (2004): while levels of reported SA induced in men and women by photographs did not differ significantly, in an analysis contrasting responses to still pictures of heterosexual couples engaged in sexual activity and responses to a fixation cross, higher levels of activation were found for males in the hypothalamus, the right and left amygdalas, the right cerebellum and the right posterior thalamus. When regional activation was examined separately in each gender, only in males were the amygdalas and the hypothalamus found activated. Gizewski et al. (2006) used fmri to compare men s brain responses to film excerpts depicting heterosexual s with those of women scanned both in the periovulatory period and in the menstrual phase. SA ratings did not differ significantly between men and women in periovulatory period, but men gave higher ratings than women in menstrual phase. Compared with women in the periovulatory period and in the menstrual phase, higher activation was reported for men bilaterally in the amygdalas, the parahippocampal gyri, the OFCs, the insulas, the caudal part of the right anterior cingulate cortex, the left thalamus, but not in the hypothalamus. Similarly, Gizewski et al. (2009) compared men with women outside the menstrual phase. SA ratings in response to the film excerpts did not differ significantly across genders. Men showed a higher activation in the right and left amygdalas, OFCs, insulas, and in the left thalamus. In a study of men and women (phase of cycle not specified) where photographs representing sexual or neutral s between a man and a woman were paired with geometrical shapes (CS+ and CS ) (Klucken et al., 2009), the difference between the CS+ and the CS was more pronounced in men in the right amygdala, left brainstem, right thalamus, and in the occipital cortex (bilaterally). In addition, the comparison of brain responses to the photographs (unconditioned stimuli) showed that activation in response to erotic photographs was higher in men in the left brainstem, left insula, medial OFC (bilaterally), right occipital cortex, thalamus (bilaterally), and left ventral striatum. In all these five studies (Gizewski et al., 2006, 2009; Hamann et al., 2004; Karama et al., 2002; Klucken et al., 2009), no region was reported as more activated in females. In addition, in a study focusing exclusively on a large region of interest comprising several cortical areas involved in visual perception, i.e., striate, extrastriate, medial parietal and inferior temporal cortices, Sabatinelli et al. (2004) found that men showed greater activation than women in response to erotic photographs. However, no data on the participants subjective sexual responses were reported. Finally, in a near-infrared spectroscopy study Leon-Carrion et al. (2007) found a higher response to VSS in men, both for the on and the off after stimulus cessation periods. Does the visual nature of sexual stimuli used in the above studies bias the results toward higher responses in male participants? How would men s and women s neural responses compare if sexual stimuli were presented through other sensory modalities? Georgiadis et al. (2009) compared brain responses of women and men to tactile genital stimulation by their partner. In men, higher responses were found in the right claustrum, left part of cerebellar vermis, and right middle temporal gyrus. However, this is the only study where higher brain responses were also found in women, in the right and left inferior parietal lobules, left postcentral gyrus, left precentral gyrus, right middle fontal gyrus (premotor cortex), and left precuneus. No data about perceived SA were reported. In the only study using sexual stimuli that were exclusively auditory, i.e., words pronounced with an erotic prosody, the responses in the right superior temporal gyrus were similar in males and females (Ethofer et al., 2007). There was an between the gender of the participant and the gender of the actor pronouncing stimulus words in the right superior temporal gyrus, a region considered as associative auditory cortex, with similarly increased responses to the voice of opposite sex speakers in males as well as in females. In a PET study, Savic et al. (2001) purported to determine whether smelling sex hormone-like compounds caused sexdifferentiated activation in the human brain, similar to the brain activation associated with pheromones in other species. Indeed, in heterosexual women, smelling an androgen-like compound activated the hypothalamus, with the center of gravity in the preoptic and ventromedial nuclei. By contrast, when smelling an estrogenlike substance, heterosexual men activated the hypothalamus, with the center of gravity in paraventricular and dorsomedial nuclei. That these findings are relevant to sexual behavior is suggested by electrophysiological experiments in monkeys: increased neuronal activity in the ventromedial nucleus in a female monkey was synchronized with mating acts (Aou et al., 1988), whereas the stimulation of the dorsomedial hypothalamic nucleus in a male monkey elicited penile erection (Oomura et al., 1988). Subsequently, Savic

23 Table 4 Differential regional cerebral activation in healthy women and men in response to sexual stimuli. Reference Ns: M/F Phase of menstrual cycle Savic et al. (2001) Karama et al. (2002) Hamann et al. (2004) Sabatinelli et al. (2004) Berglund et al. (2006) Gizewski et al. (2006) Ponseti et al. (2006) Ethofer et al. (2007) Walter et al. (2008b) Klucken et al. (2009) Gizewski et al. (2009) 12/12 2nd to 3rd week of cycle 20/20 outside ovulatory period Technique Sexual stimuli PET Smelling EST or AND fmri V: M F sexual 14/14 fmri Ph: M F sexual ; nude Fs and Ms 14/14 fmri Ph: M F sexual 12/12 2nd to 3rd week of cycle PET Smelling EST or AND 22/22 Menses fmri V: M F sexual Periovulatory period 27/26 a fmri Ph: nude M and F aroused genitals 12/12 fmri Words, erotic prosody by actor of opposite gender 11/10 fmri Ph: nude Fs or Ms or heterosexual couples 20/20 fmri Ph: M F couples, sexual Conditioned stimuli predicting VSS 12/12 outside menstrual phase fmri V: M F sexual Georgiadis et al. 11/12 PET Tactile (2009) b genital stimulation SA ratings LOCC and/or LTC IT AMY Cl Lat. OFC or anterior orbital gyrus mofc INS Hy CN Pu VS Th ACC PM Cb None No F: R activation to AND; M: R activation to EST No No No No No F > M for AND; M > F for EST No No No No No No No M > F No No No No No No No L: M > F No No No No No No No No No No B: M > F No No No No B: M > F No No No B: M > F No No R: M > F None B: M > F B: M > F None No No No No No No No F > M for AND; M > F for EST No No No No No No No M > F No No B: M > F No No B: M > F B: M > F No No No No L: M > F R: M > F No No No No No B: M > F No R: M > F L: M > F B: M > F No No No No L: M > F R: M > F No No None No No No No No No No No No No No No No No No None R: F > M for M speaker R: M > F for F speaker No No No No No No No No No No No No No No No No No No ROI No ROI L: M > F R: M > F L: M > F ROI ROI ROI L: B: M > F No ROI ROI M > F M > F B: M > F ROI R: M > F ROI No No No ROI ROI ROI No R: M > F No ROI ROI No ROI ROI B: M > F No No B: M > F B: M > F No No No No L: M > F No ROI ROI None R: M > F No No R: M > FNo No No No No No No No No R: L: M > F F > M Abbreviations: : not covered by scan or not mentioned in article; ROI: not a region of interest; AND: androgen-like compound; B: bilateral difference; EST: estrogen-like compound; L: difference in left hemisphere; No: no significant difference; None: not collected; Ph: photographs; V: videos; VSS: visual sexual stimuli. Other abbreviations: see legend of Table 2. a Both for male and female participants, homosexual and heterosexual subjects are pooled. b Findings relative to tactile genital stimulation, not to orgasm, are shown here. c Midbrain column was omitted because of lack of between-gender differences. S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012)

24 1504 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) et al. (2005) found that both in homosexual men and heterosexual women the androgen-like compound activated the medial preoptic area and ventromedial hypothalamus, which indicated a link between sexual orientation to men and hypothalamic neuronal processes. However, both in homosexual and heterosexual men, the estrogen-like substance induced a greater activation in the dorsomedial and in the paraventricular nuclei than it did in heterosexual women, which suggested that biological sex may also be associated with the hypothalamic response to this compound. In a third experiment (Berglund et al., 2006), when smelling the estrogen-like substance, homosexual women partly shared an activation in the anterior hypothalamus with heterosexual control men, although this was shown only at a subsignificant level (corrected p = 0.06). In addition, in contrast to heterosexual control women, the androgen-like compound did not induce a hypothalamic activation in homosexual women. These studies (Ethofer et al., 2007; Georgiadis et al., 2009; Savic et al., 2001) suggest that gender differences in brain responses to VSS may not apply to other sensory modalities. 4. Brain areas involved in orgasm 4.1. Male ejaculation and orgasm The first neuroimaging study (Tiihonen et al., 1994) of male orgasm, based on single photon emission computed tomography (SPECT), found decreased rcbf in all cortical areas except in the right prefrontal cortex, where rcbf increased. In a first PET study of ejaculation (Holstege et al., 2003), rcbf during ejaculation was compared with rcbf recorded during manual penile stimulation performed by the volunteers female partners. In subcortical regions, activation was found in the mesodiencephalic transition zone (including the ventral tegmental area, lateral central tegmental field, zona incerta, subparafascicular nucleus, and the ventroposterior, midline and intralaminar thalamic nuclei), the claustrum and the lateral putamen. Neocortical activation was found almost exclusively on the right side, in the inferior occipital gyrus, lingual gyrus, inferior parietal lobule, precuneus, and inferior frontal gyrus. Strong activations were found bilaterally in the cerebellum. Conversely, in the left amygdala and adjacent left entorhinal cortex, deactivation was observed. Usually, however, PET data are acquired over 1- or 2-min frames, which makes PET inadequate to study relatively short events such as ejaculation. To analyze more precisely data corresponding to the moment of ejaculation, Holstege et al. (2003) acquired data over seven 10-s frames and one 50-second frame for the ejaculation condition, and over one 120-s frame for the control condition. Subsequently, the same group evaluated the validity of their previous method (Georgiadis et al., 2007) and reported that some findings (activation in cerebellum, putamen, claustrum, midbrain, thalamus, lingual gyrus, inferior parietal lobule, and inferior frontal gyrus) were artefactual and related to the different lengths of data acquisition used for the different conditions. In a new sample, the comparison of equal 120- s images of the ejaculation condition and of the penile stimulation condition showed rcbf decreases throughout the prefrontal cortex. Activations were shown in the left dentate cerebellar nucleus, left lateral midbrain, and right pons. A limitation of this second study is that the 120-s images of the ejaculation condition contained more than ejaculation per se, so that these new findings might also have been related to events occurring immediately before and/or after ejaculation, such as high SA or sexual satiety. In the only fmri study focusing on sexual satiety (Mallick et al., 2007), the amygdalas, temporal lobes and septal area were more active 3 min after ejaculation than after 30 min. As the latter duration was considered as longer than the postejaculatory refractory period, the above-mentioned activations were thought to be correlates of the refractory period Female orgasm Women diagnosed with complete spinal cord injury above T10, i.e., above the level of entry into the spinal cord of the genitospinal sensory nerves, have been reported to perceive mechano-stimulation of the vagina and/or the uterine cervix, and to respond with orgasms to such stimulation. Komisaruk et al. (2004) hypothesized that the afferent pathway for these perceptions was provided by the vagus nerves, which bypass the spinal cord. Out of the five patients they studied with fmri, three reported orgasm in response to cervical self-stimulation elicited by an ad hoc apparatus. The hypothalamic paraventricular nucleus, medial amygdala, nucleus accumbens, anterior cingulate, insular, frontal and parietal cortices, and the cerebellum were reported to be activated during orgasm. Georgiadis et al. (2006) studied orgasm induced by clitoral stimulation in 12 healthy women. They used PET to measure rcbf in four states: (1) a nonsexual resting state, (2) imitation of muscular contractions of orgasm (motor output control), (3) clitoral stimulation provided by male partner (SA control), and (4) orgasm induced by clitoral stimulation by male partner. Perceived level of SA and rectal pressure variability, which increases at the time of orgasm, were measured. Compared with the control conditions (clitoral stimulation and imitation of orgasm), orgasm was associated with markedly decreased rcbf in the left lateral OFC, fusiform gyrus and anterior temporal pole. Orgasm was characterized by activation of the left deep cerebellar nuclei when compared with the clitoral stimulation condition, but not when compared with the imitation condition. The deactivation of the left lateral OFC and of the left temporal lobe was interpreted as the neural correlate of the temporary sexual disinhibition necessary for orgasm to take place. Activation of left deep cerebellar nuclei was suggested to be involved in muscular contractions occurring during female orgasm. The fmri study by Ortigue et al. (2007) did not include an orgasm condition. Instead, on the one hand, brain responses of women to the subliminal presentation of their beloved partner s name were recorded, and, on the other hand, a composite measure of orgasm quality based on frequency, ease to reach orgasm, level of satisfaction was obtained through a questionnaire. In one region, the left anterior insula, the BOLD response to beloved partner s name was correlated with reported orgasm quality. Ortigue et al. (2007) proposed that the subliminal presentation of the beloved partner s name evoked sexual memories encoded during previous partnered orgasm experiences and that the left insula was involved in the memory of sexual experiences. As another interpretation, they suggested that the insular activation might code for the anticipation of future reward. Alternatively (see Section ), insular activation might be related to its role in detecting salient events (Menon and Uddin, 2010). Using PET, Georgiadis et al. (2009) compared brain responses of women and men associated with orgasm induced by tactile genital stimulation by their partners. Whereas in both genders many regions presented a deactivation (right medial OFC and, in the left hemisphere, inferior frontal, middle frontal, superior frontal, medial frontal, fusiform, superior temporal gyri) or an activation (cerebellum), in a few regions activation was higher in men (dorsal midbrain/periacqueductal gray matter, left lingual gyrus) or higher in women (right insula). Results of PET studies of orgasm should be taken with caution because of their low temporal resolution. Future studies of orgasm would likely benefit from the application of fmri. 5. Time course of brain responses to sexual stimuli Neuroimaging studies investigating SA induced by the presentation of erotic pictures or film excerpts have mainly used blocked designs with long stimulus presentation times (Bühler et al., 2008). To clarify how fmri study design affects findings,

25 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Bühler et al. (2008) compared brief (750 ms) event-related presentation of erotic vs. neutral visual stimuli with blocked presentation of the same stimuli (duration of one block = 19.8 s) in 10 men. Brain activation differed depending on design type in only 10% of the voxels. Regions where activation was more marked in the event-related paradigm were bilaterally the Rolandic opercula and the supplementary motor areas. Conversely, activation was more salient with the block-design presentation in occipital and temporal regions (left middle occipital and middle temporal gyri, right inferior and middle occipital gyri, right inferior and middle temporal gyri, and right fusiform gyrus), parietal regions (bilaterally inferior and superior parietal lobules and postcentral gyri; right supramarginal gyrus), and frontal regions (right middle frontal gyrus and precentral gyrus). Authors suggested that event-related designs might be a potential alternative when the core interest is the detection of networks associated with immediate processing of erotic stimuli. In a MEG study, evoked magnetic fields recorded in male subjects 126 ms after stimulus onset were greater in the right and left lateral occipital and temporal regions in response to pictures of male and female nudes than to neutral pictures (Costa et al., 2003). In another MEG study (Rudrauf et al., 2009), responses to 10 s-long sexual clips were recorded. Three main time periods could be distinguished. During T1, early bilateral responses to VSS ( ms) were found in the OFC and temporal pole, as well as in inferotemporal and insular regions. During T2 ( ms), most of the cortices corresponding to the ventral visual stream, as well as bilaterally the OFC, posterior cingulate cortex and temporal pole, showed greater activations for sexual than for neutral stimuli. During T3 ( ms), greater activations for VSS were observed bilaterally in the OFC, ventromedial prefrontal cortex, dorsal and perigenual ACC, left insula, left dorsolateral prefrontal cortex and lateral and inferotemporal cortices, as well as in the right somatosensory and somatomotor cortices, including all SI and SII. As regards the effect of stimulus cessation, in a functional nearinfrared spectroscopy study, Leon-Carrion et al. (2007) found that exposure to a sexually explicit scene produced bilaterally a strong activation in the dorsolateral prefrontal cortex, while exposure to a non-sexual scene did not. This activation, observed during the on period, became even more pronounced during the off period, after stimulus cessation. They interpreted this finding as indicating that the dorsolateral prefrontal cortex is actively involved in maintaining the representation of sexual information in working memory. Although electroencephalographic techniques are not the focus of this review, it is appropriate to mention that using EEG to record event-related potentials offers even greater time resolution than event-related fmri designs and shows how early differential responses to VSS occur. It has been shown that event-related cortical potentials were higher in response to VSS than to sport stimuli, even when statistical analysis controlled for small differences between the two categories of stimuli on valence and arousal ratings (van Lankveld and Smulders, 2008). These higher cortical potentials occurred between 300 and 500 ms after stimulus onset (P300) and on a positive slow wave (PSW) between 500 and 700 ms. In another EEG study, Ortigue and Bianchi-Demicheli (2008) investigated the responses of nine men and four women to photographic VSS. Estimation of the potential field revealed a current source density maximum in the posterior superior temporal sulcus extending to the temporoparietal junction. This study suggested that these brain areas play a crucial role for coding of desirability of VSS in a time interval extending from 142 to 187 ms after stimulus onset. Does the brain respond to VSS when they are unseen, i.e., presented in a way that prevents their conscious recognition, in the same manner as when VSS are consciously recognized? Childress et al. (2008) used event-related fmri to test the brain responses of male cocaine patients to cocaine cues and to sexual cues of 33 ms duration. In their visual backward-masking procedure, very brief targets were each immediately followed by much longer stimuli of different content. Subjects reported having seen the longer stimuli, but the brief targets were reported as unseen. Unseen sexual cues produced, however, robust limbic activations in the amygdala, ventral striatum/pallidum, OFC, anterior and posterior insula, temporal poles, and the hypothalamus/midbrain tegmentum. Overall, the brain s rapid responses to sexual cues presented outside awareness showed substantial anatomical overlap with its responses to longer cues. 6. A neurophenomenological model of sexual arousal How can these multiple regional brain responses be organized within a phenomenologically meaningful model, i.e., a model that would account for the multiple and varied facets comprising the experience of SA? A four-component neurophenomenological model, comprising cognitive, motivational, emotional and autonomic/neuroendocrine components, has been proposed (Redouté et al., 2000, 2005; Stoléru et al., 1999, 2003). In addition, the model includes inhibitory processes. Brain regions related to each phenomenological component have been specified on the basis of their functions in SA, as discussed in the above sections (Fig. 3). The cognitive component comprises (i) a process of appraisal through which stimuli are categorized as sexual incentives and quantitatively evaluated as such, (ii) increased attention to stimuli evaluated as sexual, and (iii) motor imagery in relation to sexual behavior. The activations of the right lateral OFC, of the right and the left inferior temporal cortices, of the superior parietal lobules, and of areas belonging to the neural network mediating motor imagery (inferior parietal lobules, left ventral premotor area, right and left supplementary motor areas, cerebellum) are conceived as the neural correlates of the cognitive component of the model. The process of appraisal is postulated as being the earliest one, with later processes depending on it. The emotional component includes the specific hedonic quality of SA, i.e., the pleasure associated with rising arousal and with the perception of specific bodily changes, such as penile tumescence. The activations of the left primary somatosensory cortex that receives inputs from the external genitalia the left secondary somatosensory cortex, the amygdalae, and of the right posterior insula are conceived as neural correlates of this emotional component. The motivational component comprises the processes that direct behavior to a sexual goal, including the perceived urge to express overt sexual behavior. Thus, the motivational component is conceptualized as including the experience of sexual desire but is not limited to this conscious experience. The model suggests that the ACC, claustrum, posterior parietal cortex, hypothalamus, substantia nigra and ventral striatum are neural correlates of this motivational component. The autonomic and neuroendocrine component includes various responses (e.g., genital, cardiovascular, respiratory, changes in hormonal plasma levels) leading subjects to a state of physiological readiness for sexual behavior. According to the model, activation in the ACC, anterior insulae, putamens and hypothalamus participates in the mediation of the autonomic and neuroendocrine responses of SA. These four components are conceived as closely coordinated. Finally, inhibitory processes comprise (i) processes that are active between periods of SA and that prevent its emergence; the model suggests that this type of inhibitory control is exerted by regions of the temporal lobes and by the left lateral OFC where activity decreases in response to VSS; (ii) cognitive processes that may decrease the sexual relevance of VSS; according to the model, this type of control is mediated by the medial OFC (gyrus rectus); and (iii) processes that control the overt behavioral expression of SA, once SA has begun to develop; the model suggests that the caudate nucleus and the caudal ACC participate in this function.

26 1506 S. Stoléru et al. / Neuroscience and Biobehavioral Reviews 36 (2012) Fig. 3. A four-component neurophenomenological model of visually induced sexual arousal. Abbreviations: ACC: anterior cingulate cortex; Ant.: anterior; Comp: component; Cx: cortex; Inf.: inferior; IPL/SPL: inferior/superior parietal lobule; OFC: orbitofrontal cortex; PMv: ventral premotor area; post.: posterior; SI/SII: primary/secondary somatosensory cortex; SMA: supplementary motor area; SN: substantia nigra. 7. Future directions for research Until now, almost all studies have relied on inducing a state of SA, generally by requesting participants to view VSS, without attempting to focus on the individual components of this complex state. To ascertain the functional role of the brain responses identified by these studies, it should be useful to shift from passive paradigms to active paradigms requesting subjects to perform specific tasks. For instance, that the right lateral OFC is involved in mediating the pleasure experienced in response to viewing VSS could be tested by requesting subjects to assess levels of pleasure in response to individual VSSs during or immediately after the scanning sessions. Furthermore, new paradigms could help distinguish the neural correlates of experienced reward (pleasure) and expected reward (desire). Similarly, whether the parietal lobules mediate sexual imagery could be ascertained by requesting subjects to produce sexual imagery during sessions. Secondly, the functional relationships between brain areas responding to sexual stimuli should be investigated using methods such as structural equation modeling or dynamic causal modeling (Huettel et al., 2009). Thirdly, as the awareness of SA is likely to induce inhibitory reactions that obscure initial responses, studies based on subliminal presentation of sexual stimuli should be helpful to better map brain responses to sexual stimuli as has been the case in other motivational domains (Pessiglione et al., 2007). References Aalto, S., Näätänen, P., Wallius, E., Metsähonkala, L., Stenman, H., Niem, P., Karlsson, H., Neuroanatomical substrata of amusement and sadness: a PET activation study using film stimuli. Neuroreport 13, Abler, B., Seeringer, A., Hartmann, A., Grön, G., Metzger, C., Walter, M., Stingl, J., Neural correlates of antidepressant-related sexual dysfunction: a placebocontrolled fmri study on healthy males under subchronic paroxetine and bupropion. Neuropsychopharmacology 36, Aharon, I., Etcoff, N., Ariely, D., Chabris, C., O Connor, E., Breiter, H., Beautiful faces have variable reward value: fmri and behavioral evidence. Neuron 32, Alexander, G.E., Crutcher, M.D., DeLong, M.R., Basal ganglia-thalamocortical circuits: parallel substrates for motor, oculomotor, prefrontal and limbic functions. Progress in Brain Research 85, Aou, S., Oomura, Y., Yoshimatsu, H., Neuron activity of the ventromedial hypothalamus and the medial preoptic area of the female monkey during sexual behavior. Brain Research 455, Argiolas, A., Melis, M., Central control of penile erection: role of the paraventricular nucleus of the hypothalamus. Progress in Neurobiology 76, Arnow, B., Desmond, J., Banner, L., Glover, G., Solomon, A., Polan, M., Lue, T., Atlas, S., Brain activation and sexual arousal in healthy, heterosexual males. Brain 125, Arnow, B., Millheiser, L., Garrett, A., Lake Polan, M., Glover, G., Hill, K., Lightbody, A., Watson, C., Banner, L., Smart, T., Buchanan, T., Desmond, J., Women with hypoactive sexual desire disorder compared to normal females: a functional magnetic resonance imaging study. Neuroscience 158, Baird, A., Wilson, S., Bladin, P., Saling, M., Reutens, D., Neurological control of human sexual behaviour: insights from lesion studies. Journal of Neurology, Neurosurgery and Psychiatry 78, Balthazart, J., Ball, G., Topography in the preoptic region: differential regulation of appetitive and consummatory male sexual behaviors. Frontiers in Neuroendocrinology 28, Bancroft, J., Human Sexuality and its Problems, second ed. Churchill Linvingstone, London. Bancroft, J., Central inhibition of sexual response in the male: a theoretical perspective. Neuroscience and Biobehavioral Reviews 23, Banczerowski, P., Csaba, Z., Csernus, V., Gerendai, I., Lesion of the insular cortex affects luteinizing hormone and testosterone secretion of rat. Lateralized effect. Brain Research 906, Barlow, D., Causes of sexual dysfunction: the role of anxiety and cognitive interference. Journal of Consulting and Clinical Psychology 54, Beauregard, M., Lévesque, J., Bourgouin, P., Neural correlates of conscious selfregulation of emotion. Journal of Neuroscience 21, RC165. Berglund, H., Lindström, P., Savic, I., Brain response to putative pheromones in lesbian women. Proceedings of the National Academy of Sciences of the United States of America 103, Berridge, K.C., Food reward: brain substrates of wanting and liking. Neuroscience and Biobehavioral Reviews 20, Berthoz, S., Armony, J., Blair, R., Dolan, R., An fmri study of intentional and unintentional (embarrassing) violations of social norms. Brain 125, Blood, A., Zatorre, R., Intensely pleasurable responses to music correlate with activity in brain regions implicated in reward and emotion. Proceedings of the National Academy of Sciences of the United States of America 98, Bocher, M., Chisin, R., Parag, Y., Freedman, N., Meir Weil, Y., Lester, H., Mishani, E., Bonne, O., Cerebral activation associated with sexual arousal in response to a pornographic clip: a 15 O-H 2O PET study in heterosexual men. Neuroimage 14, Brunetti, M., Babiloni, C., Ferretti, A., Del Gratta, C., Merla, A., Olivetti Belardinelli, M., Romani, G., Hypothalamus, sexual arousal and psychosexual identity in human males: a functional magnetic resonance imaging study. European Journal of Neuroscience 27, Buccino, G., Solodkin, A., Small, S., Functions of the mirror neuron system: implications for neurorehabilitation. Cognitive and Behavioral Neurology 19, Bühler, M., Vollstädt-Klein, S., Klemen, J., Smolka, M., Does erotic stimulus presentation design affect brain activation patterns? Event-related vs. blocked fmri designs. Behavioral and Brain Functions 4, Burns, J.M., Swerdlow, R.H., Right orbitofrontal tumor with pedophilia symptom and constructional apraxia sign. Archives of Neurology 60, Bush, G., Luu, P., Posner, M., Cognitive and emotional influences in anterior cingulate cortex. Trends in Cognitive Sciences 4, Cabeza, R., Nyberg, L., Imaging cognition: an empirical review of PET studies with normal subjects. Journal of Cognitive Neuroscience 9, 1 26.