Monday, December 11, 2017

Evidence of sexual harassment and assault

The epistemology of testimony has received some attention lately, though not by that name. The epistemology of testimony is the part of philosophy in which we study when you should believe what people say (and why). Recent high-profile reports of sexual harassment and assault have resulted in a lot of discussion about whether and when we should believe these reports, which is really just a case of applied epistemology. So, if anyone ever tells you that philosophy doesn’t matter, here’s a good example of when it does.

I’m an evidentialist, which means that I think that what you should believe is determined entirely by your evidence. ‘Evidence’ is a tricky word, but as I use it, it refers to a specific kind of reason for believing something. We can have all kinds of reasons for believing things: I can believe that unicorns exist because it’s fun, believe what Maria says because she’s my friend, or believe that a job interview will go well because that will make me more confident. None of those reasons are evidence for those beliefs, though. Evidence consists of reasons that indicate the truth of something. Seeing a bunch of unicorns would give me evidence that unicorns exist, for example, because that experience would indicate that unicorns really do exist. Having experience with Maria being very reliable would give me evidence that what she says is true, but just the fact that she is my friend would not. And, sadly, if my last dozen job interviews have gone poorly, that gives me evidence that this next one will also go poorly. According to Evidentialism, you should believe something just in case you have these kinds of truth-indicating reasons for doing so.

There’s an argument for initially believing reports of sexual harassment and assault, though, that relies on a different kind of reason. It goes like this:

When we first hear a report of sexual harassment or assault, we must either believe it or disbelieve it. Disbelieving these reports has very bad consequences. Among other things, it sends the message that these reports are not taken seriously or treated as credible, which makes future victims less likely to report assault or harassment. So, we should initially believe reports of sexual harassment or assault.

The conclusion here does not require that we always believe reports of sexual assault or harassment, regardless of any other information we might discover. That would be unreasonable, as, on investigation, some (low) percentage of reports will turn out to be false. The conclusion is only that, when we first hear a report, prior to any further investigation, we should believe it.

If Evidentialism is true, though, this is a bad argument. The fact that believing or disbelieving something has very bad consequences is not evidence for or against that thing. The consequences of a belief do not indicate whether it is true. So, the premises here do not support the conclusion.

And even if Evidentialism is false, this is still a bad argument, because the first premise is false. On hearing a report, we aren’t forced to choose between believing it and disbelieving it. We also have the option of suspending judgment, of forming no belief either way, which does not obviously have the same bad consequences as disbelief. Disbelieving a report requires believing that the reporter is either lying or mistaken about their own experience, but suspending judgment does not. And suspending judgment while looking for further evidence seems to, at least in some important respects, take the report seriously.

As is often the case, though, this is a bad argument for a true conclusion. We should initially believe reports of sexual assault or harassment, but not because failing to do so would have bad consequences. We should believe them because a report of harassment or assault is evidence—it is a truth-indicating reason to believe that the harassment or assault occurred. It takes no special insight to know this, just the familiar principle that when someone says that something happened, that is evidence that it happened. We appeal to this principle all the time. It’s how I have evidence that Abraham Lincoln was shot, that Taylor Swift has received ten Grammys, and that my grandfather went to the gym last week. It’s also very often how we have our first evidence that any kind of crime has occurred.

Of course, this evidence can be defeated by further evidence. We might uncover a reason to doubt a particular reporter’s reliability or sincerity, or we might have a general reason to doubt the credibility of a certain kind of report. Absent this evidence, though, failing to initially believe reports of harassment or assault is failing to believe what is supported by our evidence. It is, also, believing that a report is not credible without any evidence for that belief—a further violation of Evidentialism and an (epistemic) injustice to the reporter. So, according to Evidentialism, we should believe reports of sexual harassment or assault, unless we have some other evidence to doubt them.

We might worry, though, that believing reports of sexual assault or harassment by default would also have bad consequences, raising another kind of challenge to Evidentialism. Those believed to have committed sexual assault or harassment face a range of possible consequences, including loss of employment and prison time, and when the reports are false, these consequences will be unjust. Shouldn’t we at most suspend judgment until a thorough investigation is completed, so as to avoid these unjust consequences?

But this worry confuses the evidence required for belief with the evidence required for action. It’s true that we shouldn’t terminate or imprison people without first conducting a thorough investigation, but it doesn’t follow that we shouldn’t believe the reports that lead to those investigations. If Evidentialism is true, then if you have a good evidence to believe something, you should. This is consistent with saying that you should seek more evidence before taking action.

Brandon Carey
Department of Philosophy
Sacramento State


  1. Great post, Brandon. Let me ask a quick question that struck me as I made it to the end of your post here: if, as you say, the statement by an accuser is itself evidence that the report is true, do you think the same thing can be said about the denials of the accused? Most of these reports are against people who are still alive and able to speak up for themselves. Some of the accused have indeed spoken up and admitted fault; others have remained silent in the face of these accusations; and still others have denied the reports. (Of course, there are actual and possible situations where an individual will calibrate their responses accordingly if there are multiple accusations, perhaps by multiple accusers, against them.)

    I guess my question is this: does your principle of epistemic charity about an initial report also apply, and to the same degree, to an initial denial of said report by the accused? And, if so, does that lead to a kind of epistemic stalemate in some cases?

    I don't take my comment here as a critique, just a clarification.

    1. Hi Brandon, thanks a lot for this post. Two questions:

      1. When you say that what we believe should be determined entirely by our evidence, what sense of 'should' are you appealing to? Is it general, or are you appealing to a specifically epistemic sense of normativity? Personally, I think if you are appealing to a general sort of normativity, especially one that encompasses moral normativity, there are some quite obvious exceptions. But if you are appealing to an epistemic sense, then its implications for real life may be quite limited.

      2. I agree with you that testimony that P counts, ceteris paribus, as evidence for P. But I think you are (purposely?) overstating your point quite a bit when you say that "we should initially believe reports of sexual harrassment..." The truth is that reports of sexual harassment only raise the probability that it has occurred, and if the prior probability is very low, then it may raise the probability to well below the threshold required for belief. For example, knowing what I do about my friend Russell, I would not initially believe a student I don't know who claimed that he or she had been harassed by him. I would say that it raises the probability, and that it certainly needs to be investigated, etc., but I would say the probability is very low.

    2. Thanks, Russell. The answers are yes and yes. When someone says that something didn't happen, that is also evidence that it didn't happen. And, if we have one person's testimony that something happened, another person's testimony that it didn't, and no other evidence, we should suspend judgment.

      There's a general complexity in this kind of case, as we might think that the accused is likely to deny the accusation, even if it's true, and so the denial is less reliable than standard testimony. This, though, is just other evidence that we have that we're bringing to bear on this case which breaks the stalemate.

    3. Hi, Randy. Both good questions.

      1. When I say I'm committed to Evidentialism, I'm using an epistemic sense of 'should'. I think what we should do in a general sense is somehow determined from what we should do epistemically, morally, prudentially, etc., and so there will be some cases where we should not believe according to our evidence (e.g. because it would be wrong or impractical to do so). With different senses of 'should' at our disposal, we can put my point here another way: this issue is sometimes framed as one where there is a conflict between what we epistemically should do (follow the evidence) and what we morally should do (believe victims), such that we have to choose between being epistemically rational and morally good. There are cases like that, but I don't think this is one of them; I think our epistemic and moral obligations recommend the same belief here.

      2. I should be more careful about 'initially'. What I meant by that is something like 'prior to considering other evidence', but of course we could have evidence that defeats the evidence from a report before we even hear the report, as in your case. I think that case is only chronologically different from one in which e.g. you discover that the student routinely makes false accusations some time after you hear the report of harassment. But yes, what I mean to say is that we have a prima facie reason to believe reports of sexual harassment or assault. This reason can be defeated either by new evidence we later discover or by evidence we already possess.

    4. Thanks Brandon. Here are a couple of quick thoughts on your responses:

      1. You may be right about the coincidence of epistemic and moral obligations in this case, but I don’t think it is obviously right. I may, e.g., have an obligation to help someone and know that my having certain beliefs about them would weaken my resolve to do so. In such cases it may be morally preferable for me not to have those beliefs even though there is strong evidence for them. If this is so, an independent argument is needed to show that this is not so in the case of sexual harassment.

      2. I think you may be right about this, too, but only if we accept the premise that sexual harassment claims are more likely to be true than false, or perhaps a principle more general than that This, of course, is an empirical claim, and I think a plausible one, but we should be prepared to provide evidence for it.

  2. Brandon, I found this post very clarifying. You're right, philosophy is useful. No society needs all that many philosophers; but our society clearly does not have all the philosophers it needs.

    Either that, or we're not doing our jobs (except for you).

    Just one question: Does the semantic concept of 'presupposition' have application here?
    Looking at a very expensive item in a store, the presupposition of my asking the question,
    "How much does this cost?"
    is that, more likely than not, I can't afford it.

    The consequences of an accusation of sexual harassment for the accuser - more downside than up - make it more likely than not, given just the accusation, something out of the ordinary took place.

    There would be lots of defeaters to the presupposition, of oourse, but just by itself it's not nothing.

    1. Thanks, Tom! I do think that when someone testifies to something, that gives us evidence not just for the asserted proposition, but also everything that is presupposed by the the assertion (or question, imperative, etc.). I think in the kind of case you're describing, though, we're bringing in background evidence (about the consequences for the accuser) that is doing some additional work. This background evidence is also subject to defeat, as you say, but yes I think it gives some additional support to believing the accusation.

  3. Hey Brandon, great post! Very timely :).

    I am glad that you presented my own belief, that reports are evidence, however I am concerned with the distinction you made in the final paragraph between evidence required for action and evidence required for belief.

    To me, if the evidence required for action and belief are separate then I can believe X but act as if ~X. It seems to me that you walk the line between saying 'belief doesn't imply the level of action we think it does' and 'belief is not a criteria for action'. Is this what you mean? Or am I missing something...

    To me at least, the grounds for belief and action have to be the same in order to maintain consistency between what we believe and how we act.