Some writers have suggested that rapists are not merely especially interested in sex, but have a particular sexual interest in rape, or some other quirk of their sexual preferences that makes rape especially appealing to them.

This sort of question is a natural opportunity for the use of phallometry, the measurement of penile erection, usually by fitting a strain gauge around the subject's penis and recording changes in circumference. Greater increases in penis size are interpreted as more sexual interest. The value of phallometry, in theory, is that it is more difficult for men to control their erectile response than to lie about their subjectively experienced sexual preferences, and deceit is an obvious concern when we are trying to measure such socially charged variables such as sexual interest in children (subjects may even fear, e.g., that parole could be denied on the basis of their sexual preferences). This is not to say that phallometry is immune to fakery, but steps can be taken to reduce subjects' voluntary control over their erections, as by requiring them to press a button every time they hear a sex-related word in a story (Seto, 2008, p. 34). Phallometry does have the weakness of being inapplicable to female subjects. Measurement of genital arousal in women is possible, but genital arousal is not as clearly related to self-reported subjective arousal in women as in men (Chivers, Seto, Lalumière, Laan, & Grimbos, 2010), suggesting trouble for use of such measurement to assess sexual preferences.

Phallometric studies have indeed found that sexual abusers, compared to non-abusers, have greater erectile response to abuse-relevant stimuli. For example, a meta-analysis by Hall, Shondrick, and Hirschman (1993) found that abusers of women exhibit "slightly more" arousal to stimuli depicting rape than comparison subjects. A meta-analysis by Lalumière and Quinsey (1994), using men known to have forcibly raped an adolescent or adult woman, found a larger effect in the same direction. Lalumière, Quinsey, Harris, Rice, and Trautrimas (2003), in a review of five studies conducted since these two meta-analyses, argued that the conclusion of rapists having greater erectile response to rape than non-rapists did was upheld. Similarly, studies of child abusers (Freund & Blanchard, 1989; Blanchard et al., 2006; Barbaree & Marshall, 1989) find that child abusers have greater erectile response to pictures of nude prepubescents or stories of adult–child sex than comparison subjects (either nonoffenders or offenders against adult women).

A potential problem with the phallometric literature is the emphasis on documented and often convicted sex offenders as opposed to ordinary people, who, I argued in the previous chapter, are responsible for the majority of abuse incidents. But there have been a handful of studies on people drawn from ordinary subject pools (e.g., college undergraduates) who admit to abusing or being willing to abuse, and the findings are generally congruent (Malamuth, 1986; Lohr, Adams, & Davis, 1997; Bernat, Calhoun, & Adams, 1999). Also, in keeping with the lower rate of abuse victimization among prepubescent children than adults, substantial sexual interest in prepubescents seems to be uncommon; Seto (2008) (p. 6–8), from a review of the few relevant studies, estimates that less than 5% of men, and still less women, have such interest. A small degree of arousal to prepubescent girls, greater than that to males of any age, may be reasonably common in men (e.g., Hall, Hirschman, & Oliver, 1995).

You may now be suspecting that abusers' lust for abusive sex is a chief cause of abuse. But the evidence I've discussed so far is also compatible with another theme. The question, to be precise, is whether abusers prefer abusing others to consensual sex (perhaps also having little to no interest in consensual sex), or whether they are merely less deterred by lack of consent than non-abusers while still preferring consensual sex. This latter idea, which I think can be credited to Blader and Marshall (1989), appears to be better supported. For example, in Bernat et al. (1999) and Harris, Lalumière, Seto, Rice, and Chaplin (2012), abusers, like non-abusers, had fuller erections to consensual than non-consensual stimuli. The latter study, which orthogonally manipulated violent, non-consensual, and sexual themes in the stories read to subjects, found similarly that abusers' arousal was inhibited by violence and injury, albeit less so than in non-abusers. Outside phallometry, Heilbrun and Loftus (1986) had male undergraduates rate the sexual attractiveness of women posing six facial expressions of emotion. The bulk of subjects (44 out of 50) found happiness most attractive, but abusers found fear, anger, disgust, and sadness less unattractive than non-abusers did.

There are some studies (e.g., 9 of the 16 reviewed by Lalumière & Quinsey, 1994) that do find greater erectile response among rapists to rape than consensual stimuli. Without doing my own meta-analysis, my impression is that this is not the rule, and I would expect it to be even less common among abusive ordinary people than documented sex offenders. Unfortunately, a lot of phallometric studies (e.g., Freund & Blanchard, 1989; Blanchard et al., 2006) only report data that has been processed and standardized in such a way that positive interest in abuse and reduced inhibition can't be distinguished. There seems to be more faith in the positive-interest idea for the case of child abusers than adult abusers (see, e.g., the discussion section of Harris et al., 2012), but I don't see a lot of evidence for it. In Barbaree and Marshall (1989), for example, only 35% of subjects who had abused unrelated children, and no subjects who had abused their own children, had a phallometric profile suggesting a preference for prepubescents over adults.

I described earlier how abusers are more sexually promiscuous in general. This dovetails with the idea that they have a greater breadth of sexual interest than non-abusers rather than their interest requiring abusive themes. On that note, consider Seto and Lalumière (2010), a meta-analysis of studies comparing male adolescent sex offenders to male adolescents who had committed other crimes. The biggest estimated difference, although there were relatively few studies on the topic, was in "atypical sexual interests", a heterogeneous group of variables ranging from interest in rape to general hypersexuality to interest in cross-dressing. Not mentioned in Seto and Lalumière (2010) is that Daleiden, Kaufman, Hilliker, and O'Neil (1998) found that abusers, compared to non-abusers, were more likely to identify as bisexual and less to identify as heterosexual.