Jumping to conclusions? The hidden shortcuts in “he-said–she-said” and “sexual harassment”

Susan and Richard work together at a law firm. One day when they’re alone in the conference room, Richard makes a vulgar sexual overture. Shocked, Susan refuses — to which he snarls, “Fine! Go home to your cats, then!” and storms angrily out of the room.

Incensed at the blatant sexual harassment, Susan complains to her manager and to HR. Following an investigation — one consisting entirely of asking Susan and Richard what happened — HR tells Susan that they’ve looked into her allegation, but weren’t able to substantiate it. Richard denied her allegation; and without any witnesses or other corroborating evidence, they explain, it’s unfortunately one of those he-said–she-said situations. Disciplining Richard for his alleged misconduct just wouldn’t be consistent with due process, they say.

This is a familiar story. One person accuses another of wrongdoing, and the accused denies it. It’s right there in the label: “he-said–she-said”.

But there’s more information packed into these four words — it’s not just that we apply them in cases of conflicting testimony; we also use them to signal something epistemic: if it’s a he-said–she-said situation, one will naturally think, we should suspend judgment about what really happened. The very idea of a “he-said–she-said situation” seems to imply that in cases like those we opened with, the fair thing to do is to remain agnostic about what really happened. (As we’ll explain below, we reject this idea.) The ideas around this label pave the way for certain normative or value-involving conclusions to be drawn about how (or even whether) to proceed in our collective normative deliberations. At the public level, he-said–she-said involves what we call a communal normative inference ticket. Just as a train ticket licenses its holder to ride the train from one station to another, an “inference ticket” provides permission for certain transitions in thought. Think of them as “mental shortcuts”— direct inferential paths to conclusions that would otherwise take a longer, more circuitous “route” to get to. In the case at hand, we think these inferences are licensed:

She accused him of misbehaviour, and he denied it —————– It’s a he-said–she-said situation

It’s a he-said–she-said situation —————– We should suspend judgment as to what really happened

As we (following Ryle and Prior) understand them, inference tickets have “input” conditions — under what circumstances can one infer that it is a he-said–she-said situation? — and “output” conditions — what does its being such a situation allow you to conclude? Our normative inference tickets have outputs with normative significance — either because they are explicitly normative (and so involve concepts like should and wrong), or otherwise feed obviously into outstanding normative or evaluative frameworks. So normative inference tickets give us “mental shortcuts” not just to certain (merely descriptive) conclusions like “Eddy is a cat”, but to conclusions about what is good, what is bad, what matters (or does not matter!), what does or does not “make sense”, and what we ought or ought not to do.

That the he-said–she-said inference ticket plays such a role in our collective deliberations about how to think — and that that role is open to interrogation and potential change — is a live idea in popular political discourse. “He-said–she-said” is a #metoo trope, illustrating entitlement, himpathy, institutional betrayal, and many other standard components of rape culture.

Speakers fluent in the contemporary sense of “he-said–she-said” might not notice just how much ideology has been built into the phrase. Before the 1990s, “he said, she said” was just a label for a kind of dialogue between a man and a woman: he said X to her, and she said Y back to him. (See for instance this 1948 hit, performed by Marion Hutton.) It wasn’t until the 1991 Clarence Thomas SCOTUS confirmation hearings, when Anita Hill described a pattern of sexual harassment from Thomas, that the label took on its now-familiar connotation in mainstream American English: “he” and “she” aren’t saying things to each other; they are each saying something to a third party, perhaps the public. Moreover, they’re not saying just anything —  they are giving opposing testimony about their interaction: she said he mistreated her, and he says that he didn’t. The label has come into wide currency since the early 90s — it was attached to Bill Clinton and Monica Lewinsky, to Brett Kavanaugh and Christine Blasey Ford, and even to Donald Trump and James Comey.

So it’s easy enough to recognize a “he-said–she-said” situation, as individuals and interpersonally. Speakers in our community quickly and easily communicate that they think there’s no point trying to decide what happened in a given situation by applying this label.

A conceptual repertoire containing these inference tickets carries substantive epistemic commitments — implicit ideas about what to believe or not believe. “He-said–she-said” provides an inference ticket from conflicting testimony to skepticism. But conflicting testimony of this kind doesn’t inevitably balance out; women rarely have incentive to make false sexual misconduct allegations (in part because of the hostile responses they often provoke). Men accused of wrongdoing, by contrast, will typically have incentive to deny allegations, even if they are true. Sometimes the preponderance of evidence leaves one unable to determine what to think. But sometimes the epistemic significance of a complainant’s testimony is enough to rationalize belief. So in these cases, the “he-said–she-said” inference ticket misleads: it gives thinkers a tendency to infer, falsely, that one should suspend judgment.

Inference tickets, then, can be epistemically unreliable, by yielding false beliefs. But they can also be morally pernicious.

A characteristic feature of inference tickets is that they are default and automatic. The “mental shortcuts” they provide are largely reflexive, highly intuitive, and, as far as cognitive processing power goes, efficient. To consider a mundane example, an inference ticket ordinarily takes us from x is a bird to x flies:

x is a bird —————- x flies

But inference tickets are also characteristically exception-allowing: thinking with them does not inevitably carry a full commitment to these default inferential patterns. Thus when we “board” the inferential train at “x is a penguin” we can ride express to x is a bird while getting off before “x flies”:

x is a penguin ————— x is a bird

x is a bird ——–X——- x flies

Inferences paved by inference tickets, in other words, are not analytic entailments. The conclusions they license do not necessarily follow, and there can be exceptional circumstances under which we may consistently reject them. Crucially, though, such circumstances are just that –– exceptional. Just as a commuter might hop off at an earlier station on mornings their usual stop’s out of service, we do not “have” to go where an inference ticket would ordinarily take us. But, like the commuter, we almost always do.

This makes inference tickets conceptually recalcitrant, in the sense that they resist counterexamples and prompt automatic (rather than deliberate) inference. This can make them a moral liability, as it makes them (and the stereotypes that underwrite them) incredibly hard to “unlearn”––even for those who want to.

This liability is especially pronounced in the case of bigoted thought and language. We think fluent usage of slurs and other derogatory language paradigmatically exploits widespread (communal) normative inference tickets, taking speakers from some conspicuous  (stereotypical) feature that they’ve observed about something to certain (stereotypical) negative conclusions or evaluations about that thing. 

For example, consider the expression ‘chick flick.’ We think that, among speakers who use this expression, there is a communal normative inference ticket that functions more or less like this:

x is a movie made for / about women ————— x is a chick flick

x is a chick flick —————- x is bad / not worth watching

The crucial point is that the ‘chick flick’ inference ticket makes the inferential slide from “x is a movie made for / about women” to “x is bad / not worth seeing” almost instant. And it can do so even in cases where the relevant negative judgment would otherwise be pretty surprising. Thus, when it was announced that the 2016 Ghostbusters reboot would feature all-female leads, suddenly an action comedy about hunting ghosts was being widely described as a “chick flick.” Countless men refused to see it, declaring their favorite childhood franchise “ruined”. Within six weeks of its release, the reboot’s trailer became the most disliked movie trailer in YouTube history. 

The “chick flick” label was controversial. Some wanted to reject it full-stop on the grounds that the very concept of a “chick flick” is inherently sexist. (In effect, this is a repudiation of the “chick flick” inference ticket.) Others simply denied that the label was apt in this particular case. They believed in chick flicks, but said that even though it starred women, Ghostbusters isn’t a chick flick. They made an exception to the generalization that movies like that are chick flicks.

x is a movie made for / about women ———X—— x is a chick flick

x is a chick flick —————- x is bad / not worth watching

Finally, some were happy to call Ghostbusters a chick flick, but denied that that was a bad thing. It was, as it were, “one of the good ones.” They made an exception to the generalization that chick flicks aren’t worth seeing. This exception-tolerating feature of inference tickets, we think, is key to the persistent and pernicious character of slurs in bigoted thought and language — one can allow for exceptions, but they do not challenge the generalization.

Here’s another example to illustrate how stereotypes and inference tickets interact. Think back again to Susan, the victim in our sexual harassment thought experiment. When she rebuffed Richard, he angrily responded by telling her to “go home to your cats”. This is a recognizable gendered insult. Richard did not pick a random feature of Susan’s life that he happened to know about. (No one would angrily retort, “go home to your kitchen chairs, then!”) Invoking cats does something specific.

Richard is invoking the cat lady stereotype, and an associated set of sexist normative inference tickets. The public imagination includes a set of stereotypical associations that provide inference tickets that license these inferences:

x is a woman who lives alone with cats ————— x is a cat lady

x is a cat lady —————- x is romantically undesirable

This pattern of licensed inferences, supported by communal stereotypes, paves the way for an automatic, unreflective inference from having cats to being romantically undesirable, at least relative to a certain mainstream heterosexual standard. Different communities associate these ideas in different ways: many feminist women have embraced the ‘cat lady’ label for instance, and some people do associate it with attractiveness. (Likewise, some people like chick flicks, and describe them as such.) This is another illustration of the contingency and contextualization of inference tickets.

Generally, though, the normative outputs of (in these cases quite problematic) inference tickets are more-or-less reflexively embraced. And the more commonly they’re embraced, the bigger the problems will be. We have identified two kinds of problems: moral problems concerning treating people fairly, and epistemic problems having to do with doing a good job forming accurate beliefs. Consider “he-said–she-said” again. A default tendency towards skepticism in cases of testimony about sexual harassment will create a tendency to perpetuate testimonial injustices. Moreover, since the kinds of inference tickets we are interested in are communal, as opposed to individual, their effects are often cultural. The he-said–she-said inference ticket leads to broad social misunderstanding, and contributes to rape culture, by serving to protect perpetrators of sexual misconduct.

At the same time, however, these very features make good inference tickets good: such “mental shortcuts”, we argue, can constitute genuine moral and hermeneutical advances.

We think, for example, that some of the social progress Miranda Fricker describes in her discussion of hermeneutical injustice is best understood in terms of good normative inference tickets.

One of Fricker’s central examples in Epistemic Injustice is the historical development of the term “sexual harassment”. Fricker emphasizes the epistemic challenges faced by victims before it was widely used and understood. Our character Susan could easily recognize, and communicate to HR, her experience of sexual harassment. But earlier victims of sexual harassment were not so lucky. As Fricker puts it, these women “suffered (among other things) an acute cognitive disadvantage from a gap in the collective hermeneutical resource” (2007, 151). Moreover, she says, in such cases “harasser and harassee alike are cognitively handicapped by the hermeneutical lacuna—neither has a proper understanding of how he is treating her” (2007, 151). When this kind of hermeneutical disadvantage is the product of a prejudice, Fricker says, someone suffers a hermeneutical injustice.

Many philosophers have interpreted Fricker’s notion of hermeneutical injustice as primarily a matter of concept availability. But we doubt this is the best way to explain Fricker’s point about the harms suffered by earlier victims of sexual harassment. For one thing, as philosophers like Kristie Dotson and Gaile Pohlhaus, Jr. have pointed out, many sexual harassment victims could understand the harms they were suffering; there is a diversity of sets of hermeneutical resources, available in different communities, that Fricker failed to acknowledge or adequately emphasize. We agree –– it’s not just a matter of a missing concept. Still, we think Fricker is right that key hermeneutical resources were missing. Rather than concepts, though, we think they were communal normative inference tickets.

What the early victims of sexual harassment were missing was not a concept of what was happening to them. What they (and their harassers) lacked, we submit, was an inference ticket: an automatic, reflexive mental shortcut  that could take them, not just in their own heads but in their discussions with each other, and at the public level at large, from “this is happening” to “this is wrong.”

Because inference tickets make inferences easy and automatic, the hermeneutical advantage of such a ticket in this instance is profound. It is of course true that, even in the absence of the concept sexual harassment, the inference

x is repeatedly asking their co-worker on a date
————————————————————
x is doing wrong

is available. But getting there requires a significant amount of social knowledge and moral imagination. A “sexual harassment” inference ticket makes it much easier:

x is repeatedly asking their co-worker y on a date ————— x is sexually harassing y

x is sexually harassing y —————- x is doing wrong

These inferences, we think, are central to the stereotypes attached to “sexual harassment”; a community with such stereotypes in it will be one in which certain important social knowledge — like the fact that it’s wrong to repeatedly ask your co-worker on dates — will be much more widespread. Indeed, it will feel near-tautological.

Inference tickets can be specific to local social contexts, or in wide public currency. Consciousness-raising can be a way of developing inference tickets within a small community, as in the feminist groups described in Fricker’s discussion; but when the hermeneutical resources become firmly-enough established there, they can be brought into to the broader culture as well. The sexual harassment inference tickets are now quite secure, even outside feminist spaces — due in part to awkward but ubiquitous PSAs like this one:

This is an important component to epistemic justice.

Those automatic associations in thought that amount to normative inference tickets can have a profound impact — whether helpful, as in the case of ‘sexual harassment’, or harmful, as in the case of ‘he-said–she-said’. But they are not inevitable. Like all social phenomena, communal normative inference tickets can with time be changed, augmented, or replaced. Recognizing and understanding inference tickets and their operation is an important first step towards responsible social decision-making about what kinds of inference tickets to encourage.

The Women in Philosophy series publishes posts on women in the history of philosophy, posts on issues of concern to women in the field of philosophy, and posts that put philosophy to work to address issues of concern to women in the wider world. If you are interested in writing for the series, please contact the Series Editor Adriel M. Trott or Associate Editor Julinna Oxley.