Friday, February 25, 2011

The Subjective Experience of Pornographic Images

With the utmost appreciation to the participants in this survey...


The Survey Instrument
The Survey Population


Overall evaluations: Sexy vs. Disturbing
Reasons for Disturbance
Ambiguous Readings
Ambiguous Readings: Gender
Ambiguous Readings: Power Roles
Ambiguous Readings: Attitudes
Role in the Narrative
  • Not There / No role Mentioned
  • Jumping the Frame (JtF)
  • Basic Surrogacy
  • Personalization
  • Complex Surrogacy
  • Off-stage Roles
  • Marginal Roles and Meta-fantasy
  • Unclear (JtF or Simple Surrogacy)
  • Split Surrogacy
  • Other

Other Thematic Content of Narratives
  • Incorporation of Lovers
  • Coercion and Consent
  • Salvation Fantasies

Areas for Further Research


Since the 1960s, scholars, activists, and congressional committees have compiled a large, polemic, and yet fairly inconclusive literature on pornography. The bulk of this scholarship, both empirical and theoretical, focuses on two distinct arenas of disagreement. First, there is the question of the behavioral effects of pornography on its audience. Second, and independently, there are ethical questions around the production and use of pornography, or around censoring it in a society that values free speech.

It would be hard to overstate the degree to which these arguments are polarized between explicitly self-identified critics and defenders of pornography. Nevertheless, both critics and defenders of pornography have typically agreed on the subjective mental experience of the pornographic consumer. That is to say, it has generally been assumed that pornography was used by heterosexual men who inserted themselves, directly or indirectly, into a fantasy of heterosexual male domination indicated by the pornographic image. This assumption is generally referenced as the “male gaze,” a term loosely derived from Laura Mulvey's 1973 paper Visual Pleasure and Narrative Cinema.

Visual Pleasure actually used the term “male look” and was focused not on pornography per se but on a Freudian and Lacanian reading of cinema. Mulvey's argued that the sexualized female figure in films is generally presented for the pleasure of heterosexual men, and sexualized male figures are presented as "surrogates" for the male viewer. Both of these versions of the gaze help alleviate male castration anxiety.

Mulvey cautioned that "this complex interaction of looks is specific to film," but the terminology has been adopted and simplified to analyze more static texts, and especially to analyze and critique pornography. The term does not seem to be especially common in recent scholarship. Garrity has described the “male gaze” as a useful cliché, and Paglia (1998) has called on Mulvey to retract or clarify her original thesis, quoting Mulvey herself as saying Visual Pleasures was written “polemically and without regard for context or nuances of argument.”

Nevertheless, the idea of the male gaze is commonplace in popular feminist writing, where it serves as shorthand for an ostensibly thorough-going analysis of the embodied motivations and prejudices of all kinds of images, not simply film. The crucial components of the gaze, in this version, are the inevitable subordination of the female subject to the male viewer, and the use of male subjects as surrogates for the male viewer to relay their subordinating and misogynistic perspective.

The most immediate challenges to this conception have been the existence of a female and queer audience for pornography, as well as pornographic images that portray explicitly dominant (and thus, plausibly, non-subordinated) female figures. These categorical challenges have generally been ignored, and discussions of pornography have often taken pains to define them as irrelevant or marginal. Most broadly, since the mid-20th century there has been a shift in usage by legislators and activists towards a consensus that writing is not pornographic. This removes at a stroke the largest and oldest corpus of pornography aimed at a female audience: the romance novel.

A few authors (e.g. Jeffreys 1990) have addressed these challenges to the assumption of the male gaze more directly, arguing in effect that the narrative of misogyny is so deeply embedded in pornography as a medium that it asserts itself in spite of the image's objective contents.

Throughout this literature, though, there is very little inquiry into the subjective expereince of pornography. Researchers have measured the effect of pornography on penile engorgement and heart rate and political opinions and indices of sexism and so forth, but they have rarely ever asked the subjects to describe the pornography itself. In the few studies that do provide information about subjective experience, it tends to be auxiliary to the main purpose of the study. But these are revealing all the same.

At least three studies (Kant 1971; Rachman and Hodgson 1968; Marshall 1988) have noted in passing that participants tend to re-interpret the content of the pornography they use or were shown. Kant and Marshall both noted that sex offenders of various categories in his study did not have a history of exposure to, or current preference for, pornography with content reflecting the nature of their sex offenses. (Kant's 1971 study seems to have been expanded and republished several times.) Marshall describes how rapists, in interviews, often preferred pornography depicting consensual sex, which they might then re-interpret as a rape fantasy. Rachman and Hodgson noted that their subjects described sexual fantasies that were not consistent with the content of the pornography they were being shown.
While these findings have been ancillary to the studies they appeared in, they raise larger questions about the subjective experience of pornography. This study was designed to address those questions.


In October of 2010, I conducted a survey using a Tumblr website, and aimed primarily at people who already viewed pornographic Tumblr websites. I selected 70 images, mainly from Tumblr accounts, with the aim of creating a relatively diverse selection of objective content, within a single medium (still photography), and a fairly consistent level of photographic style. Several standard pinup photos were also taken from Playboy and Playgirl, respectively: all the other images were 21st century. I avoided images that seemed dated (as estimated by hairstyles, clothing, and photo quality). I tried to select images with models that had some degree of racial diversity, but were otherwise fairly typical of the body types and age range of models in mainstream pornography. In general, I aimed to minimize the variability between images to the themes I was considering: gender, activity, and D/s content.

I focused on themes of oral, anal and vaginal intercourse, nudity, exposure, bondage, and light to moderate sadomasochistic acts. Within those parameters, I selected a number of images that were ambiguous either in terms of the activity or the gender of the models. I avoided photos that had a thematic focus on other fetishes, such as animal role-play, uniforms, urine play, or the like. I also avoided distinctly transexual models. While these themes would all have been very interesting to incorporate, they would have increased the size of the survey to an untenable degree.

To the best of my ability, I removed website names and other text, as well as URLs, so as not to prejudice the viewer with information they might infer or know based on the source of the image. This was not entirely effective. Three respondents mentioned that they identified one of the models (Faye
Reagan) and one respondent identified the production company of another image.

The respondents were asked to provide basic demographic information, using an open-ended question format. They were asked whether or not they identified as feminists, and asked two questions to form an index of propensity for sexual violence.

The images were arranged into groups, each of which had thematically similar content, but varied by the genders of the models. They were then randomized. Participants were instructed to find five or more images that they had a “fairly strong reaction to” if possible. They were then given the following instructions:

-Note if you find the image “sexy,” “disturbing,” “both,” or “neither.”
-Describe in two to three sentences what you imagine or fantasize to be happening in this image.
-If you imagine yourself being in the image, or being in the story you have outlined above, explain who it is you imagine yourself to be. Otherwise, write “I'm not there.”

The Survey Population

The survey was advertised largely on pornographic Tumblr sites, so it can be presumed that the population is self-selected for exposure to pornography. It is especially noteworthy to have a sample of female respondents who regularly use pornography, since there is relatively little information about this group.

I received responses from 96 people, who collectively provided 929 evaluations of images. Respondents were asked to answer a series of preliminary questions, all open-ended, providing us with a profile of the respondents.

There were two transmen, two men who identifed as genderqueer males, and one woman who qualified her response as female. Noting this, I have collapsed the gender information into a binary format for the rest of the survey: 44 women and 52 men.

For formatting reasons, the demographic statistics for this population will be linked on a separate page**.

I categorized pornography use as low (weekly or less), high (daily or more), or intermediate. In passing, I should note that six men (12%) and 16 women (32%) who looked at porn daily (or more often) offered some type of explanation for this. The most common justifications were that the pornography they used was of an unusually acceptable variety, that they liked to view pornography but did not find it arousing, or that they did not seek out pornography, but it happened to be on websites they liked to visit. In all events, it seems safe to conclude that even in this population, women attach more stigma to viewing pornography than men do, and feel a greater need to justify their behavior.

Following a number of previous studies, I asked two questions on the survey with the hope of generating a measure of propensity to sexual violence. Unfortunately, given the small sample size, I had too few positive results to utilize in this analysis. The question did glean some results that might be worth pursuing. A number of respondents who identified themselves as kinky answered yes to questions about having pressured or coerced someone into a sex act, but qualified their answer by saying they had done so within a consensual framework. Unless we agree a priori that consensual BDSM is equivalent to rape or other sexual violence, these yes answers are misleading. Yet there is a distinct possibility that they affect the results of surveys and inventories that do not allow open-ended questions.


Overall evaluations: Sexy vs. Disturbing

Of all the images evaluated, 65% were seen as sexy, 8% as disturbing, 22% as neither, 3% as both, and 2% as other. There were no significant differences by gender, though women who reported low pornography usage were somewhat more likely to rate images as disturbing. Most of the respondents were quite familiar with pornography, and seven of them volunteered that they did not find any of the images disturbing, often citing their experience of pornography with much more extreme content. A number of other respondents commented that individual images, or the whole collection of images, became less disturbing and more sexy over time.

All of this suggests that the evaluation of pornography is influenced by familiarity with pornography as a genre. One respondent, a 31-year-old woman who uses pornography weekly, categorized image #57 as “some BDSM stuff”: neither sexy nor disturbing. It seems likely that someone less familiar with the tropes and themes of BDSM pornography would be more apt to find the image disturbing. This calls into question findings based on samples of people (typically women) with little prior experience of pornography, whose responses might be viewed as a first impression. For instance, Senn and Radtke (1986) use such a sample to argue that women can distinguish consistently between erotic and disturbing images.

Notably, no such consistent discrimination appears in this survey. Every image was rated as sexy at least once. Just over half the images (37) were also rated as disturbing. Moreover, there was not a continuum between images more likely to be rated as sexy and those more likely to be rated as disturbing. In fact, there was a slight positive correlation between those evaluations (r = 0.11). Two images (#1 and #40) were rated in the top tier in both categories. Notably, gender was not a significant factor in determining whether people found these images sexy or disturbing.

At the same time, the 33 images that no respondent considered disturbing were also ranked as sexy much less frequently. Of the nine lowest-ranked images for sexiness, six were also not rated as disturbing by any respondent. (These were #30, #34, #43, #46, #61, and #68. Two of these images, interestingly, are traditional Playboy nudes: this iconic genre of pornography had little interest for the respondents, with the exception of older heterosexual males.

The images considered sexy by the highest number of respondents were #3 and #32, both of them depicting heterosexual sex acts with a fairly close focus on the woman's face.

Taken together, this suggests that viewers with some experience of pornography consistently categorize images as having more or less impact, with high-impact images being regarded as either sexy, disturbing, or both.

Reasons for Disturbance

The most common reasons that respondents gave for disturbance with given images were dislike for a model's appearance; discomfort with gay and lesbian themes; a sense that the image was stereotypical; concern about whether the model was enjoying the experience; and concern about whether the model was underage. These cut across both genders without significant variation.

There were two major differences by the respondent's gender. Men were significantly more likely to be disturbed by homosexual themes, and often construed solo male images as homosexual per se. Moreover, men were much more specific, and much more varied, in being disturbed by the models' appearances. Women tended to complain about female models being stereotypical or “porny” (sic) and male models having too much body hair. Men had a much wider range of complaints, including breast size, implants, body hair and lack of body hair. Men were equally critical of male bodies, perhaps in part because of the concern about homosexuality. One 55-year old male respondent wrote: “Fit young men are generally disturbing to overweight middle-aged men like me.” Another man complained that the penis in image #52 was too large; a third man questioned whether the penis in image #21 is real. Still other comments focused on musculature; one respondent argued that the shape of a model's buttocks suggested he was gay.

These reactions only occur in a small minority of the responses: roughly 6% of all male responses to images of male models rated the image as disturbing, and invoked distaste for homosexuality (or transexuality) as a rationale for that disturbance. Still, this is a high enough figure that it raises a serious question about Mulvey's Freudian mechanism for the gaze. In contrast to the Freudian concept of men experiencing castration anxiety on seeing the naked female form, all the evidence from this survey would suggest that men in 2010 are more apt to experience anxiety on seeing other men's penises.

Many of the other reasons stated for disturbance were quite thoughtful and particular. One woman, who identified herself as a virgin, wrote (of image #3): “The woman's face is discomfiting because she seems enthusiastic and I am anxious about Having Sex because I haven't...” Several men mentioned that image #40 disturbed them because, in their fantasies, it gave them too much control, and they weren't sure what to do.

Ambiguous Readings

If someone had never seen the image gallery, and were attempting to reconstruct it from the responses, they would certainly conclude that there were several hundred images, not a mere seventy. Not only do the fantasy elements of each narrative differ, but the ostensibly objective content varies. For instance, the lower model in image #51 is variously described as a cis-man, a trans-man, a woman wearing a strap-on dildo, or a woman holding a dildo that is not strapped in place. Several respondents viewed image #56 as depicting a couple engaged in anal sex with a strap-on dildo, a reading that I as a viewer simply cannot reconcile with the positions of the model's bodies.

I want to make two contentions about the way that people “resolve” these ambiguities.

First, there is a tendency for people to resolve ambiguous images in ways that produce the narrative that they expect to find (or hope to find, or, potentially, are afraid they'll find.) One woman stated this explicitly, saying that the model on the right in image #62 was female, and noted that she made this assumption in part because of “my own interest in filling her spot.”

Second, it appears that once people have resolved an image in a particular way, it is difficult for them to re-interpret the image, even if they change modes from fantasy to objective analysis. The survey was not constructed to ask for objective analysis, but respondents volunteered it in a few cases. For instance, in one narrative of image #44, the respondent read the male model as wearing a penis-extending device, an idea that they said aroused them. They then argued that the image had racist undertones, since it implied that “black chicks are used to big dicks.” But no other respondent saw anything here except a man wearing a condom.

There is considerable external support for these two hypotheses. To take one recent instance, American Apparel advertisements have been frequent targets of public outrage over their erotic imagery. Both the critics and the defenders of these images frequently mis-read the images in concrete ways. A recent AA advertisement featured a half-naked model holding a polo mallet. Critics interpreted this as a woman with a riding crop, or a woman in bondage. Meanwhile, the editors defending the decision to run the advertisement in a Vermont magazine described it as a woman holding a ski pole (skiing presumably being a large part of the local economy) (Polston et al. 2010). In each case, someone sat down specifically to write a description of an erotic image, and in each case they saw something quite different, objectively incorrect, and in keeping with their preconceived narratives.

I want to consider these hypotheses in regard to gender, power role, and facial expressions.

Ambiguous Readings: Gender

There are nine images in which the respondents disagreed on a model's gender. (These are #2, #10, #27, #31, #34, #51, #55, #62, and #70.) Of the 119 responses to these images, only six explicitly noted the ambiguity of a model's gender. In three other images (#14, #23, and #69) respondents noted an initial mis-perception of gender which they subsequently revised.

I collected the responses with gender ambiguities, and divided them into three categories for which we can test the first hypothesis, as follows.

Category I – Respondent Identifies with the model whose gender is in dispute. There were ten instances. In seven of the ten instances, the respondent identified the model with their own gender. (In 6 of those cases this provided a match with their sexuality, as well.) The three exceptions were all male respondents.

Category II – Respondent identifies with a different model, and has an exclusive sexual orientation. There were 13 instances. In 11 of these, the gender ascribed to the model fulfilled the dyad for the respondent's sexuality. For instance, heterosexuals looking at an image of a woman and a model of ambiguous gender tended to view the second model as male. Once again, the two exceptions came from male respondents.

Category III – Respondent does not identify with any model, and has an exclusive sexual orientation. There were 33 instances. In only 14 of those instances did the ascribed gender fulfill the sexual dyad. Of the 17 other respondents, 14 were male.

It seems, then, respondents who insert themselves into the image, tended to resolve ambiguities of gender in favor of their own fantasies. They were much less apt to do this if they didn't insert ourselves into the image. And in all cases, women were more likely to do this than men.

Ambiguous Readings: Power Roles

I categorized the power dynamics in each respondent's descriptions of the images, and then compared these descriptions for each image. In nine cases, all respondents described an image in neutral terms, without mentioning a power dynamic. In 14 cases, all respondents described an image using the same power dynamic, and in a further 22 cases, all respondents described an image either in neutral terms or in terms of a single dynamic. For instance, in image #37, the stories either focus on dominating the model, or make no references to power dynamics. Again, in image #12, most respondents do not suggest any power dynamic, but those who do agree that the woman is the “top”.

For the remaining 24 images, however, respondents did not simply disagree about whether there is a power dynamic, they disagree over what type of power dynamic might be present. For instance, in image #1, some respondents see the woman as a submissive who is about to be willingly or unwillingly spanked. Other respondents see the woman as a dominant figure who is teasing a submissive character with a view of her vulva. Still others do not read any power dynamics in the image.

In 12 different images, one or more respondents envisioned “complex” dynamics. Just how complex can be suggested by the respondent whose story for image #11 involved the addition of six off-stage characters, including himself, his wife, and a lesbian couple. These storylines were often very idiosyncratic and had little to do with the apparent content of the image. Some images generated as many as three distinct complex readings.

To test the hypothesis above, I want to ignore complex readings, most of which are difficult to classify. We will look at the 13 images for which there are directly conflicting simple power dynamics: some respondents think X has power over Y, and some respondents think Y has power over X. There are 37 descriptions of these images by respondents who describe themselves as dominant or submissive (or similar terms), and mention a specific power dynamic within the image. In 35 of the 37 instances, the respondent's interpretation of the power dynamic in the image corresponds with their own preferences based on gender and D/s role. In a few cases, the resulting narrative does not match the respondent's stated sexual orientation, and this is occasionally commented on. For instance, a submissive lesbian respondent writes (of image #55): “I want to be said pretty curly-haired sub. I like this image because female dominance is so frequently fetishized into this disgusting (to me) dominatrix thing, which I am not into at all. So I tend to turn to male dominance in porn, which is less interesting given my limited interest in men.”

Ambiguous Readings: Attitudes

Respondents' impressions of the model's attitudes were highly diverse. It was commonplace for respondents to describe the same facial expressions in diametrically opposed terms. The model in image #9, for instance, is described as coy, seductive, terrified, apologetic, or bored. The model in #29 is “sensitive,” “very submissive,” “drunken,” “has been abused.” The female model in #32 is “frustrated,” “enjoying it,” “doesn't look like she's enjoying it,” “nervous,” “hungry,” “lost in...multiple orgasms.” The model in image #1 is described as “having fun”; “not having a good time”; “shameless”; “not vulnerable”; “cold”; “excited”; “nervous”; “crying with pleasure.” This is spite of the fact that no part of the model's face is visible in the photograph!

In almost every instance, the facial expressions are resolved in a way that is consistent with the respondent's fantasy narrative. If this resolution is not possible, it is often singled out. The female model in image #44 is smiling and elevating her eyeballs, an expression that many respondents interpreted as non-sexy, and focused on this in their responses.

One reading of facial expressions in image #30 gives a good illustration of the second hypothesis. A woman describes the right-hand model in this image as “clenching his teeth and trying not to yell and reveal weakness.” The teeth-clenching fits nicely into her BDSM fantasy, but it is an additon: the model's mouth is obscured in the image, and clearly we cannot know if he is making noise or not. It is equally plausible that he is being photographed in the act of screaming. The respondent followed her fantasy with an ostensibly objective analysis of the image, arguing the two models are “on closer inspection” probably performing for an audience. One of the pieces of evidence offered for this conclusion is that the model is “noiseless.” This wording betrays the idea that even a careful, factual, reading is colored by the way the respondent initially saw (and heard!) the image.

Role in the Narrative

For each image they addressed, I asked respondents to explain the role that they saw themselves taking in their own fantasy. The tension of these possibilities has been noted in erotic art for a very long time, but I am not familiar with any efforts to study it empirically. One female respondent, writing of image #14, says:

“It reminds me of that poster that college boys have- "The Kiss" with the two girls. But this somehow feels so much more real, less like for a male fantasy and more like a private moment. Lesbian imagery always begs the question: which one am I in this image, or am I a third party about to join?”

To look at the way people answer that question, I divided the responses into eleven categories, described below. Such a breakdown is necessarily subjective, but the vast bulk of the responses fell easily into one of these categories. Nine responses offered multiple possible interpretations of their role in the narrative, and I coded each possibility separately, for a total of 939.

All the correlations and statistical comparisons below are significant to 0.05 or better.

Not There / No role Mentioned

I had prompted the respondents to write “I'm not there” if they did not envision themselves within the image. This response, in various forms, accounted for 22% of the responses. It is quite common for images that respondents found disturbing and not sexy.  One respondent mentioned that she rarely ever saw herself as part of her fantasies when using pornography.

Another 28% of responses do not state any role, but do not explicitly disavow a role, either. This response was significantly more common from women, from kinky people (except submissives) and from heterosexuals; it was uncommon from bisexuals and homosexuals. It was negatively correlated with pornography usage, and most common with images that were found neither sexy nor disturbing; it is very uncommon in images that were regarded as both sexy and disturbing.

Virtually all the remaining role categories were strongly associated with sexy images, and dissassociated from “disturbing” or “neither” images.

Jumping the Frame (JtF)

In these responses, which account for 16% of the total, the respondent simply describes how they are interacting with the model(s). There is no segue or explanation for how or why they exist in this narrative: the hallmark of jumping the frame is that the viewer seems to takes it for granted that he or she can “step into” the picture.

The typical version of JtF is an older, heterosexual, male respondent who uses pornography frequently, looking at an image of a lone female model. This corresponds very well with Mulvey's description of the male gaze. As noted earlier, it also corresponds to a type of pornography (the “pin-up girl”) that seems to be losing prominence, and did not attract much interest from the younger respondents to this survey. JtF respondents are, on average, 47.5 years old; the survey mean is 38.4.

Numerous versions of JtF occur that do not meet the above description, though. These include female viewers, viewers entering the image to interact with groups, and in 10 instances, viewers imagining an interaction contrary to their stated sexual orientation.

Basic Surrogacy

In this pattern, the respondent equates themselves with a model whom they match in (percieved) gender and sexual orientation. They enter the narrative by “becoming” the character they assign to that model. A number of comments suggest that it is also important for the model to roughly match the respondent in terms of race and body type, and even hair color and handedness. This seems mainly to be a preoccupation of male respondents. Despite searching for a matching body, the respondent typically describes the story as though they are inhabiting the model's body and reality, not “exchanging places”. Typical phrases are “I would be that guy” or “I am her.”

This pattern, which corresponds well with Mulvey's concept of surrogacy, accounts for 15% of all responses. It is more common with submissives than dominants, and homosexuals than heterosexuals. It was more common among people who use pornography frequently and gave a qualified response to whether or not they were feminist.


These are responses that are primarily focused on discussing the respondent's real-life recollections, anecdotes, or anxieties, as reflected by the image. A succinct but typical example would be this response to image #44: “Looks like the Ethiopian girl I slept with over the weekend” Again, a 20-year-old woman responds to #50: “I love that this beautiful woman looks close to my own age/comfort level - and I am fascinated by the idea that her novice experimentation is akin to my own.”

Personalization responses account for 4% of the total. They tend to come from respondents who use pornography frequently, are less apt to identify as feminists, and did not find the images in question disturbing.

Complex Surrogacy

In this pattern, the respondent identifies with a model as a surrogate (as above), but the model does not match the respondent in terms of perceived gender and/or sexual orientation. This accounts for 3% of all responses.

Complex surrogacy is substantially more common among submissives and homosexuals, and less common among heterosexuals. Respondents using this pattern tend to be younger than average (32.2 as opposed to 38.4).

Offstage Roles

In this category, which accounts for 3% of the total, the respondent envisions themselves in the narrative, but identifies with a character who does not appear in the image. Sometimes this character is identified within the implications of the image or narrative: for instance, a man discussing #64 writes “I'm the limo driver,” and a number of responses to images with bondage elements state that the respondent is the person who has bound the model. Respondents who create a “slave auction” narrative for an image might identify as an off-stage buyer.

In any event, the respondent's character is not envisioned as interacting with the models in the image, at least directly or immediately. In some cases, they are quite remote: in one response to image #39, the respondent identified themselves as the model's master, who was away on a business trip. (This distance is even more notable in the next category of images).

There is some indication that off-stage roles allow people to resolve images that they can't imagine themselves in. For instances, in five stories based on #14, #53 and #66, heterosexual respondents (or homosexuals of the other gender) used this pattern to identify the same-sex encounters in the image as a performance for their benefit.

Offstage roles are disproportionately likely to be used by respondents who identify as dominant, heteroflexible, and use pornography frequently.

Marginal Roles and Meta-fantasy

As mentioned above, offstage roles generally distance the respondent's surrogate character from the activity in the image. The extreme version of this pattern deserves its own category, as it is quite consistently described and offers some insight on the pornographic experience. In this pattern, the respondents describe themselves being inside their fantasy narrative, but they are explicitly only there to watch, and quite often are only allowed to watch.

For instance, one response to #49: “I'm there, but I'm just watching.” Or to #53: “I sit quietly in the room, careful not to break the tension by moving.” This pattern is striking in that it retains essentially the same detached, voyeuristic quality as the actual act of looking at the photograph. For instance, a male respondent places himself offstage in image #47 thusly: “I am observing his predicament, and wondering how it would be if I was in his place.” But the respondent is, in fact, observing the model's predicament and considering what it would be like. What is achieved by creating a fictional surrogate to duplicate the experience of one's actual self?

This question is amplified when the fantasy based on the picture involves the picture itself. Photography-as-such is a common theme in all responses, and especially common for marginal-role responses, which often identify the respondent as the photographer. In a small but telling number of instances, the photograph appears in the story based on the photograph. For instance, a story for image #64 involves the image itself being lost and seen by people it was not intended for. More paradoxically, a story for image #35 identifies it “as a photo of a story a friend is telling me over drinks.” This implies a kind of meta-fantasy: the respondent is (in fact) being shown a photograph, and fantasizing about it. In his fantasy, he is also being shown the photograph and fantasizing about it. What, one has to wonder, is his surrogate persona fantasizing about? Still another iteration of being presented with the photograph?

Marginal roles account for a further 3% of the total responses. They are more common among male respondents, and never occurred in respondents that had a low rate of pornography usage.

Unclear (JtF or Simple Surrogacy)

In the system of categorization I am using, there is a some overlap between clear-cut JtF stories and clear-cut surrogacy. At the boundary between these two imaginaries, the viewer may imagine themselves jumping the frame and then physically replacing (rather than becoming) one of the models. An excellent example of this is a response to #48, where one man wrote “Here, hand me that cane, I'll finish this. Go get a beer, dude.” In other cases, though, it is ambiguous whether the respondent has become the model, or entered the scene and functionally replaced the model (who is not them). This is made still more ambiguous when respondents ignore one of the models in their description.

I have classified the ambiguous responses as a separate category, which accounts for 2% of the total. Demographically, they closely resemble the responses classified as JtF.

Split Surrogacy

In these cases, respondents identified themselves as both or any of the models in images that had more than one model. In one instance, a man discussed the possibility of being both the models simultaneously. Of all the categories, this was the most likely to place the respondent in a foreign gender role or sexual orientation. It accounted for 2% of all responses, and was negatively associated with heterosexuality, and strongly associated with qualified responses about feminism.


The final category, consisting of 1% of the total, includes responses in the second person, analyses of the image that didn't contain any information about the respondent's role in the image, questions of complaints about the purpose of the survey, and fantasies too confusing to categorize.

Other Thematic Content of Narratives

Incorporation of Lovers

A number of narratives involved the respondent's actual lovers, or former lovers. While this was most common in the category of personal reflections, lovers were also added to the image narratives, either off-stage or using models as proxy surrogates. In a few cases, respondents' fantasies placed their lover in an image but not themselves.

Coercion and Consent

Coercion was a common theme in many of the responses, and appeared in a variety of roles.

Some respondents explicitly fantasized about “consensual non-consent”: that is, they envisioned the persons in their fantasy as consenting to role-playing some type of coercion. This again has an aspect of meta-fantasy, since it suggests that the respondent's fantasy incorporates a secondary set of fantasies.
Other respondents explicitly fantasized about coercion, rape, sexual slavery, mind-control, and even murder. They may or may not emphasize that the victims derive some kind of enjoyment from this coercion (a common trope of D/s pornography) but the distinctly portray the fantasy as non-consensual. For instance, on respondent writes of image #41:

This has been a fantasy of mine since before I can actually remember. I remember enacting complicated auction sequences with a friend of mine in the backyard in fourth grade that were very similar to this. In my mind, these women have been born into slavery or sold into it from a young age. They have been groomed particularly for sex slavery, and are both terrified to be chosen and desperate to be pleasing enough to catch someone’s eye. It’s a matter of both ego and the total abnegation of ego. - I’m there. In this image, specifically, I’m the one in the blurry foreground, because that’s the one that most looks like me.”

Finally, a considerable number of the entries discussed the possibility that the models in the photographs were being coerced or pressured into modeling, or that they were underage or intoxicated.

In many of the fantasy depictions of coercion in this survey, though, it is unclear whether the coercion being described is envisioned as real, fantasy, or meta-fantasy. Even terms like “slave” or “rape” are often used by BDSM enthusiasts to refer to consensual roles and activities, so they do not necessarily help to clarify this point. Yet it was frequently evident that respondents were contemplating multiple levels of fantasy where the issue of consent was concerned. This layering is apparent in descriptions like the following, written of image #40:

“In this image, as a sexual fantasy, I see a girl, who has been forced to be a sexslave. Her eyes are so sad, but on some level I also know, that of course her participation in taking this image is consensual and therefore a bit sexy, because she can let herself to be submissive like this. I'm not there.”

Given the degree of ambiguity cataloged earlier, it is perhaps no great surprise that coercion themes are hard to predict based on the objective content of an image. In image #57, for instance, one respondent views the image as consensual self-bondage, noting that this is physically impossible. Another respondent fantasizes that the model is being prepared for “a butt plug or suppository that contains a mind-control drug”. Similarly, the elements in image #41 are objectively so innocuous that it is arguably non-pornographic by several common definitions. Yet six of the twenty narratives involved the models being slaves or prisoners, a much higher rate than for some of the images that show women in bondage or being beaten. Meanwhile, image #48 shows a woman bound and being beaten. Welts are visible on her buttocks. Yet no respondent describes this image as coercive.

Three images produced fantasies of life-threatening peril with the possibility of a character dying.  One respondent describes image #2 as depicting a man who has beaten a woman unconscious and is about to throw her in the sea. Another respondent describes #38 as a spy who is about to be frozen or buried alive and "left to die" by a mad scientist.  A third image, #35, also generated a somewhat ambiguous drowning narrative.  Indeed, none of these three stories are entirely explicit about the character's death, and the story for  #38 includes the possibility of rescue.  The major element of these narratives seems to be peril, rather than definite "snuff fantasies."

Salvation Fantasies

Another recurring theme in the fantasies offered by respondents was that they envisioned themselves “saving” the model from some type of distress, and generally being rewarded with sex. This salvation and gratitude fantasy occurred in a number of images that contain bondage or D/s elements: #25, #30, #35, #40, #57, #59, #60. It is used almost exclusively by heterosexual men who do not identify as kinky. In most cases, though, the respondents find the image sexy; their narrative does not ignore the bondage and D/s themes in the image, but rather is framed around negating them.

In some images that produced salvation narratives, there is no explicit danger present. For instance, one male respondent views image #61 as a sort of lost-in-the-wilderness story:

“I imagine him being found and nursed back to health by a heroine in romance novel fashion to spark a sexual scenario, either he is subservient out of gratitude, or else he overpowers his saviour. If I am in the story, then I am the saviour/heroine figure.”

Image #2, which most respondents did not see as containing any violent themes, produced both a salvation fantasy and a snuff fantasy.  The same is true of image #35, if we interpret the drowning story as a snuff fantasy.  And image  #38 contains both a fantasy of leaving the model to die (from the villain's point of view) and a possibility of rescue.  Apparently, content that lends itself to peril fantasies also lends itself to rescue fantasies.


In an argument loosely based on Mulvey's concept of the male gaze, it has long been assumed that the subjective experience of pornography involved either the viewer's sexual domination of the subject, or else the viewer adopting a like surrogate for the same end. In fact, these modes of encountering pornography appear to account for slightly less than one-third of responses to this survey. Even if we look only at responses in which the viewer is present in their narrative of the image, these simple modes account for only 63% of all responses. More surprisingly, from the traditional viewpoint, there is almost no difference in that rate by gender.

It appears that the experience of translating the objective elements of a pornographic image into an erotic fantasy is much more complex than has often been imagined.  Viewers exercise considerable creativity in engaging the image, and often engage the image on multiple levels, producing both fantasy and analysis, or even fantasy and meta-fantasy.  Viewers add, remove, and modify elements, impose storylines that bear no relationship to the image, freely change the genders of models, and insert themselves into the images in every imaginable way, or stay removed from it. Moreover, the objective content of the images does little to predict these interpretations. Polo mallets become riding crops; quiet conversations become anal sex. The most extreme themes in the respondent's fantasies, such as actual enslavement, mind control, and murder, are often found with images whose elements are quite innocuous. In several cases, an image with only two models produced five or more different combinations of power-role and gender interpretation.

Kipnis (1996) has argued that acknowledging these complexities undermines the use of pornography as a “political rallying point” for social conservatives. I am not sure I agree. In the last decade, anti-pornography legislation has increasingly focused on litanies of specific acts, body parts, and types of media, rather than abstract litmus tests for obscenity. Certainly, it is hard to reconcile such checklists with the idea that the content of images is highly subjective. But the acknowledgment of these ambiguities could shift legislators and activists in a more cautious direction. Image #41 is fairly tame in objective terms. One respondent saw it as a sexless “an ad for red hair dye.” But many of the other respondents read it as a slave auction or similarly coercive narrative. The existence of such readings could plausibly be a rallying point for concern about a much broader range of images.

Perhaps the most interesting result of this survey is the impression that there are cohort effects at work in how people experience pornography. The generation of heterosexual men over 30 are disproportionately likely to enjoy Playboy images and breast-focused single-model shots. They are also the group that most consistently “jumps the frame” or uses simple surrogacy to engage images. By contrast, younger respondents and non-heterosexuals are much more likely to engage the images in other ways, especially if they are frequent users of pornography.

These conclusions are problematic. It is much easier, for instance, to measure whether someone is looking at an image of a kiss, or a rape, than it is to measure if they are seeing a kiss, or a rape. Legislators, in particular, have to deal with the objective. Yet any understanding of pornography's behavioral effects has to focus on the subjective, and it would appear that that the subjectivities of pornography are as complex and varied as human desire.

Areas for Further Research

This study focused on a population that was already familiar with pornography; it would be instructive to compare results with a group that was not. Similarly, in this study I focused on a relatively narrow spectrum of body types, ages of models, and fetish elements. I focused on a single medium: photography. It seems possible that a narrative medium such as text or video would reduce the ability of the viewer to manufacture their own storyline. Again, images with more unusual themes might inhibit the creativity that was apparent in this survey. Or perhaps not: this is certainly an arena where further research would be rewarding.

From a policy standpoint, it would be quite useful to have a comparison with a population of known sex offenders, or a population that could be given attitudinal inventories of some type.

Finally, I believe this study demonstrates that Tumblr is an effective medium for implementing and advertising a survey of this type.


  1. I'm the person with the "mad scientist" fantasy for image #38. I think it is wrong to characterise this as a definite snuff fantasy, in my mind it is a peril fantasy. My description said:

    "she is going to be ... left to die like that... Rescue may or may not come." "Left to die" is not the same as "she dies", and the key point is that a rescue from this anticipated death is possible but uncertain. After all, in mainstream movies the hero always arrives in the nick of time!

  2. This is a good point, and I'll change the wording in the article. Given the popularity of peril tropes in the media, I expect this is a fairly common fantasy, and I think that (as you suggest) part of the fantasy is often the uncertainty. Finishing the story, in either direction, might be less exciting than the moment of not knowing. Yes?

  3. (This is the same person)

    Yes, indeed! Certainly with that particular story - and I think it's interesting that some of my earliest "kink" memories are of mainstream media serials with cliffhanger "peril" situations (even in media aimed at children!).

  4. Part of the drama of cliffhangers is that you can spend a whole week (or whatever) obsessing about the image. How will it resolve?

  5. I am a bit surprised that only one person out of 96 participants said, that she rarely ever saw herself as part of her fantasies when using pornography. I feel the same way, too, but it didn't occur me to mention it when I took this survey.

    Huh, I just thought that it was really common.

  6. I, three, never see myself as part of my fantasies when watching pornography. The concepts of jumping the frame and either complex or simple surrogacy seem alien to me. My mental process is more on the meta-fantasy level, I am watching, and I am *in fact* watching...I am not present in the image as a voyeur, but somehow retain the knowledge that I am watching outside the frame.

    This survey is so interesting, thank you Orlando, I wish I could have taken part!