Rape and Signal Detection

Research blogged at yesmeansyes indicates that about 6% of males are rapists (from a military sample). Most of these rapists also know their victims, and use alcohol and drinking to perform and get away with rape.

Phaedra Starling gives an example called Schroedinger’s rapist: that (all) unknown males are potential rapists. She describes that potential in terms that would remind physicists of signal detection (all emphases hers):

To begin with, you must accept that I set my own risk tolerance. When you approach me, I will begin to evaluate the possibility you will do me harm. That possibility is never 0%...

The second important point: you must be aware of what signals you are sending by your appearance and the environment. We are going to be paying close attention to your appearance and behavior and matching those signs to our idea of a threat.

I’m going to focus on these points because I think they’re the most intellectually interesting that she raises. I don’t know if she purposely cast the question in scientific terms in order to appeal to male sensibilities, but I think it works, and I would like to extend it.

When working in signal detection, we aim to detect signals and ignore noise by setting a threshold for what will be considered as a signal.

The two errors that can occur are false negatives (signal is under threshold, not detected) and false positives (noise is over threshold, false ‘signal’ detected).

In any but the most clear-cut situations, noise is going to be a problem and these errors are going to occur. What is interesting, though, is that the person interpreting the signal can decide whether they’d rather receive false positives or false negatives by changing the signal threshold.

You have to choose your poison: In the case of medical tests, we move the threshold down, as false negatives are what we want to avoid (tests can be repeated in the event of a false positive); in case of a trial by jury, we set the threshold for evidence fairly high, following the legal maxim that it is better to let 10 guilty walk than imprison a single innocent.

Phaedra Starling’s signal detection problem, Schroedinger’s rapist, is likely to have a low threshold. In order to keep the danger of rape (a false negative) as low as possible, males who does not communicate well must be turned away (the false positive- a normal person mistaken for a rapist). This is the only way in which this signal detection can work- a low threshold is essential to its function.

I think that it is perfectly reasonable to be uncomfortable and suspicious of a person that propositions you late in the evening in an enclosed space. Skepchick was right in taking all precautions to avoid a potential sexual assault.

When he was turned away and nothing happened, though, we should be intellectually honest enough to admit that it this was a false positive.

Unwanted attention is unpleasant, and we have all had it, whether male or female, and have been in situations where we have been scared of the person. When nothing goes wrong, there are two possibilities- you were lucky, or perhaps the person was not as bad as you thought they were.

Watson’s comment on the issue was arrogant in not acknowledging this possibility. I would not expect this attitude from a reason-based individual. We’re supposed to be able to revise our views based upon evidence. If we were free to speculate without resort to reason, one might come up with the following counter-narrative:

In the youtube clip, she talks, as if to him, but in a generalized manner. She is talking down to him, as she knows that she occupies a position of higher status. What is profoundly bizarre, though, is that she seeks to prolong their relationship at all by communicating with him. This ultimately undermines her claim that he was scary in the first place. Also, she hates sexualisation of women but sells calendars of sexy skepchicks on her website.