Why do people have kinks? Where the evidence runs out

The oldest answer is the one still standing. Alfred Binet, writing in 1887, argued that these interests form out of chance associations made early in life, and most of what has been published since is a version of that claim, tested, narrowed or attacked. The field is not short of alternatives. Schippers, Hoogsteder and de Vogel's 2024 systematic review in Sexual Abuse collected 47 theories of how deviant sexual interests develop and found seven that met basic quality criteria for a theory.

The catalog is a place to hold those theories against something. Each of its 486 entries records what its own literature says about origin. Twenty-two invoke John Money's lovemaps. Ten name classical conditioning in those words. Several attach to a fibre or a plastic nobody had made before the twentieth century, which is a constraint on any theory that treats the object as arbitrary.

The association hypothesis

Every modern account of where a fetish comes from descends from a claim made in Paris in the 1880s.

In 1882 the neurologists Jean-Martin Charcot and Valentin Magnan described patients with fixed attachments to shoe nails, nightcaps and white aprons, and filed the pattern under degeneration theory without giving it a name. Five years later Alfred Binet named it. His essay Le fétichisme dans l'amour argued that a fetish begins in an accident of timing: an object that happened to be present at an early moment of arousal, and then held it. Binet's chance-association claim is the hypothesis every later theory either elaborates or refutes.

Krafft-Ebing had already catalogued the cases in Psychopathia Sexualis the year before, reading them as a pathology of association. Freud, in the Three Essays on the Theory of Sexuality of 1905, put the fetish on a continuum with ordinary erotic life rather than apart from it. The catalog inherits that vocabulary whole: fetishism is a DSM-5-TR disorder because Binet's word survived the journey from anthropology through neurology into the manuals, and partialism, the focus on a non-genital body part, is a nineteenth-century observation the DSM has split off and folded back twice.

What the laboratory produced

The conditioning account has been tested directly, and both times it worked.

In 1966 the behaviourist Stanley Rachman projected a slide of a pair of women's boots immediately before slides of nude women, and measured penile response. Three men reached his criterion of five successive responses to the boots alone, after 30, 65 and 24 trials. The response then extinguished when the pairing stopped, and recovered spontaneously afterwards, which is what a conditioned response does. Rachman's experiment and the replication he ran with Hodgson two years later remain the only human demonstrations of their kind. A laboratory built something with the shape of a fetish, in three men, and then took it away again.

The animal version is cleaner, because rats have no culture to borrow from. Pfaus, Erickson and Talianakis gave sexually naive male rats their first matings while the animals wore a small jacket, and reported the result in Physiology & Behavior in 2013. Males that had learned to mate in the jacket went on copulating normally in it. Males that had learned without it, then fitted with one, approached less, ejaculated less often and took longer about it. A garment had taken control of the behaviour. That is Binet's mechanism, running in an animal that has never seen a boot.

Rachman's own subjects are the reason the finding is quoted with care. The sample was three men, the boots were fashionable in 1966, and later readers have asked how much of the response was conditioning and how much was the decade.

Where the conditioning account thins

Conditioning explains how an interest could form. It does not explain who ends up with one.

Kevin Hsu and Michael Bailey put the objection in the Archives of Sexual Behavior in 2020, under the title The Poverty of Conditioning Explanations for Sexual Interests. Their points are the obvious ones, which is what makes them hard to answer. Nearly everyone encounters arousal alongside incidental objects, and nearly nobody comes away with a fetish. Fetishes are overwhelmingly male, and conditioning has nothing to say about that. The response Rachman produced faded within a few sessions, while the interests people report last decades.

James Pfaus, whose rats supply the other half of the evidence, replied in the same journal that year. His title carries the argument: conditioning in humans is difficult to see, close to impossible to test, and probably how it happens anyway. A person's sexual learning is thousands of unrecorded pairings, in private, across years, and no ethics committee will approve the experiment that would settle it. The argument ends in a design problem rather than in a finding.

The theories that survive a quality check

Of the 47 theories Schippers and colleagues collected in 2024, seven passed their quality assessment, and the survivors converge on four processes: excitation transfer between emotion and arousal, conditioning, difficulty with what the review calls normative sexuality, and social learning. No neurobiological theory of sexual deviance met the bar at all, because none of acceptable quality could be retrieved.

The cortical story about feet shows what that absence looks like. In Phantoms in the Brain (1998), V. S. Ramachandran noted that the foot and the genitals lie next to each other on the cortical body map and suggested that cross-activation between the two might predispose people to eroticise feet. The hypothesis is memorable and it is attached to the largest case in the catalog: foot fetish is the partialism people report most, estimated at 14 percent of adults and ranked fifth of 486 entries. When Turnbull and colleagues mapped erogenous intensity across the body in around 800 people for Cortex, feet came out low and adjacent sites did not cluster. The story is still repeated, and it has never been supported.

Arousal borrowed from somewhere else

One of the four surviving processes has nothing to do with objects at all, and it covers a family of interests the association account handles badly.

Schachter and Singer argued in 1962 that physiological arousal is ambiguous and gets labelled by context. Dutton and Aron demonstrated the consequence in 1974: men who had crossed a high swaying bridge were more likely to read their own racing hearts as attraction to the woman waiting on the far side than men who had crossed a solid one. Dolf Zillmann's excitation-transfer model describes the same borrowing across the minutes after a fright.

In the catalog it accounts for the interests whose object supplies a bodily state rather than a memory. Fear play is the explicit case, and its entry is built on that literature. Consensual non-consent, estimated at 30 percent of adults and ranked 40th, works on the same machinery with the fear scripted in advance. Brontophilia, arousal from thunderstorms, sits at 0.1 percent and 429th, and its entry is candid that nobody has tested the mechanism on it directly. The transfer account predicts a bodily state doing the work, which is why it fits the rarest entries as well as the common ones.

The story about damage

The account most people arrive with is that a kink comes from something that went wrong. It is the one claim in this area with population data standing against it.

Richters and colleagues put BDSM questions to 19,307 Australians in a national telephone survey published in 2008. Respondents who had taken part in the previous year were no more likely than anyone else to have been sexually coerced and no more likely to be unhappy or anxious, and the men among them scored lower on psychological distress. Wismeijer and van Assen compared 902 practitioners with 434 controls in 2013 and found the practitioners less neurotic, more extraverted, more open to experience and more secure in attachment. Neither survey found the damaged population the trauma account requires.

Money's Lovemaps (1986) supplied the version of the story the vocabulary still carries: the vandalised lovemap, an arousal template deformed by early shame or punishment. He named a great deal of what there was to name, which is why the catalog credits him with coining sixteen of its entry names, from autonepiophilia to somnophilia, and invokes lovemaps in twenty-two. The theory fits any case once you have the case, and forecasts none of them in advance.

What the objects are dated to

Whatever forms a kink, the object it settles on has to exist first.

The catalog files 30 entries under objects and materials and 32 under clothing and garments, and several of those attach to a substance with a birth certificate. Nobody could have eroticised nylon before DuPont announced it in 1938, or spandex before Joseph Shivers made it in 1958, or PVC clothing before André Courrèges put it on a Paris runway in 1964.

Entry	The material arrives	Rank of 486
Gas mask	Small Box Respirator issued, 1916	272
Nylon	DuPont announcement, 1938	207
Spandex	Shivers at DuPont, 1958	282
PVC	Courrèges collection, 1964	227

Age tracks rank, and the gap is wide. Leather and latex, the two oldest materials in the group, rank 39th and 49th; every synthetic that came after them sits outside the top two hundred. Pictophilia, arousal running primarily through images, is estimated at 15 percent and ranked 50th, and the earliest surviving erotic photographs date to around 1846. Robot fetish took its community name from a Usenet newsgroup founded in the 1990s.

None of that settles the mechanism. It does constrain the answer: a theory that makes the object arbitrary has to explain why the arbitrary objects arrive in the order the factories made them.