Friday 9 October 2015

Mass killings as 15 minutes of performance art

Some years ago, as part of the very broad range of work which has come my way by virtue of being a psychologist, or being thought to be a psychologist, two Police officers came to my office with a complicated problem.

An armed man had taken two women hostage. I will draw a veil over the details of the case, for fear of anyone being identified. A siege ensued, and a policeman was given the task of building up a negotiating relationship with the hostage taker, by chatting to him on a telephone link. Although the policeman had probably received some training, he was not experienced in these matters (protracted sieges are relatively rare in the UK). After some days of companionable conversation the perpetrator asked his police contact what was the record length of a domestic siege. Without much thought the policeman gave him the then current record, which led to the man holding the women for an extended record number of days, during which they were badly abused by him.

An anthropology professorial colleague later questioned me about it, and was glad to find that his theories were confirmed: criminals are as much in need of recognition and as bound by theatrical traditions as any other performer, particularly the duller ones without any other options and incapable of innovation.

The hostage taker knows that he has to keep things going until the TV vans arrive. Then he has to make protestations of innocence, describe a grievance, request special privileges, have specific people brought in to plead with him (priest, schoolteacher, celebrity, former spouse), require special foods and finally demand that his manifesto be published. Then he can concede and be led away to a police van, shouting his final demands, or repeatedly fire his weapons out the window and then storm out of the building into a hail of bullets, according to preference and accepted conventions.

Mass shootings in the US now have a gristly ritual to them. One is under way at North Arizona University as I write this, and there is no need for news outlets to explain what “lockdown” means.  On later investigation it will probably turn out that long justifications and opaque warnings were made on social media, with coded farewells, then ritual dressing for combat (in a variant of the kamikaze tradition), gathering the arsenal of fearsome weapons, the highly symbolic choice of target, and in the US cultural setting, the hail of bullets finale, leading to posthumous publication of The Statement.

I quail from giving a list of these mass killings but some of the worst are: 21 killed, 19 wounded; 22 killed, 20 wounded; 32 killed, 17 injured. 7 killed 7 injured seems to be a rough median, and they would not have been achieved without guns.  

Can one detect such killers beforehand? The usual answer from psychiatry is “not without locking up much of the population”. I have some sympathy with this view, though it is more of a judgment about psychiatry/psychology than it is about criminality. Murder is rare, and mass murder much rarer. (The likelihood of a subset should not be greater than the likelihood of the set from which the subset has been derived.)   Rarity might help in detection, but usually doesn’t, unless the rare behaviour has a strong link with something, like narcolepsy often has with orexin. Steve Sailer wonders what the credit scores of mass shooters are. This could be fertile ground. Credit card companies claim to be able to predict divorce better than spouses (but it probably takes only a few hotel bills in out of the way places to reveal an affair). Certainly the credit card records of Islamic terrorists are often very illuminating. As a rule of thumb, it is very unusual for extreme behaviour not to have a precursor, though some come out too late to be of much use. A sudden change in appearance (change of hairstyle; change of clothing from casual to devout, or less often the reverse; or from functionally sociable to taciturn) in the weeks and days before an event are usually seen only in retrospect. In UK policing a good way of catching criminals is to have spot checks on cars. Those who haven’t bothered to insure them contain a higher than usual proportion of people who haven’t bothered to return “borrowed” property, to attend court hearings, and to observe other, more major, laws. The sieve of minor transgressions reveals the richer soup of criminal noodles.

An impossibly disorganised credit history might be an indicator of a disorganised life, but the regression equation would probably have to include 5 or 6 predictors. What would they be if one was trying to predict US mass shooters? Guns might be a poor predictor (too widespread); multiple arms purchases perhaps somewhat better; internet viewing and postings a good predictor; social isolation a  fairly good predictor, but by far the largest category seem to be workplace grievances, followed by school/college arguments and marital disputes involving prior threatened or actual violence. Vengeance is served hot.

Each of the indicators might have a hit rate of barely 1 in 50, but carefully combined they might might have some predictive value. What then?

However, one general facilitator of these crimes might be a general un-willingness on the part of neighbours, work mates and fellow students to regard threatening and violent behaviour as reprehensible if it is classified as “mental health issues” and psychologists are involved, and to be too embarrassed to say: that person scares me, and I want them kept at a distance.

Perhaps if some behaviours go beyond the usual norms they need to be stigmatised again.

 

16 comments:

  1. Or it could turn out to be an argument between drunk frat boys.

    ReplyDelete
    Replies
    1. To be fair, the dude did have an Instagram history of posing with firearms.

      Delete
  2. I find it interesting that they often accumulate far more guns than they need for their massacres. How many weapons can you usefully take with you? Perhaps two pistols and an automatic rifle? They seem often to have far more at home.

    As for correlates, how about medications?

    ReplyDelete
    Replies
    1. SSRI meds are a common thread.

      Delete
    2. Interesting. Is there data on the frequency of such usage?

      Delete
  3. Posing with firearms is perfectly normal and common in the American context. Buying a collection of firearms is likewise perfectly normal in America. Neither is associated with criminality.

    ReplyDelete
    Replies
    1. Is there data on this? Had assumed that there were big regional and urban/rural differences, and that half the population at least had no guns at all.

      Delete
    2. The map in this study shows regional differences in gun laws: http://archinte.jamanetwork.com/article.aspx?articleid=1661390.

      In our area, posing with firearms in social media is not normal. So the answer is, "it depends."

      Delete
    3. can't get the link to work. Any demographics on the gun holding people?

      Delete
    4. Yes, there are big differences along a rural-suburban-urban gradient. Pew has a report on the demographics of gun-owning households that hits several demographic notes.

      I grew up in a rural household with a lot of guns, mostly shotguns for hunting. Guns always seemed normal stuff for a wall rack to me.

      Delete
  4. As with all tests for rare conditions, the problem is the false positive rate compared to the prevalence of the condition. For any realistic specificity and sensitivity of a mass-shooting predictor, virtually everybody identified as a threat will actually be harmless. Let's say the test identifies 100% of mass shooters, has a false positive rate of just 1 in 10,000, and the yearly prevalence of mass shooters is 1 in 30 million (10 mass shootings per year in the US).

    Intuitively, the test will identify about 30,010 potential mass shooters per year in the US (300M/10,000 = 30,000 false positives + 10 real positives), so there is about a 1 in 3,000 chance that an identified potential shooter really will be a mass shooter. No real test will be anywhere near that good, so the odds will be even lower than that.

    Tests predicting more prevalent risks than mass shootings have a much better potential to save lives. The most prevalent risks are drivers (especially with vehicles) and physicians (especially with prescription pads), each group having hundreds of times the death toll of mass shooters each year.

    ReplyDelete
  5. Dear EH, thanks for you comments, which made me smile. It reminds me of the way I used to greet medical students in their first year of studying psychology applied to medicince: Nothing you do in your career will be as cost effective as a condom.
    Yes, when a condition is very rare false positives defeat the purpose of any screening. The more fruitful approach to predicting violence is probably to look at high risk of offending populations: those with previous convictions, drug users and psychosis.

    ReplyDelete
    Replies
    1. Link to some of the issues in psychosis

      http://drjamesthompson.blogspot.co.uk/2013/03/schizophrenia-and-violence-delusional.html

      Delete
    2. Suitable censorship of the media would put a stop to much of this nonsense. Normally I would strongly oppose censorship but the media self-censors so much these days, particularly in the US, that a little bit of benign censorship would seem small beer.

      Delete
    3. Thanks, Dr. Thompson.
      I think the principal predictors of violence don't line up neatly with the DSM categories. The main predictors, I speculate are sustained rage together with a feeling of being personally wronged and a means of suppressing the usual inhibitions against violence, which may happen in several ways: a delusional narrative, psychopathic personality, or drugs/alcohol. The feeling of being personally wronged will, I think happen more often with personality disorders than with schizophrenics. Their frequent inflated self-worth and disdain for others have to be big risk factors. Paranoid schizophrenics usually imagine that their enemies are virtually all-powerful, so their delusion doesn't usually make them think that they can succeed in secret planning of a massacre. Also schizophrenics usually just aren't organized enough to bring off a mass murder.

      Use of amphetamines or some of the SSRIs such as Effexor (which may cause irritability or even manic episodes) in people with preexisting personality disorders or who have feelings of rage for other reasons could be a precipitating factor in mass shootings.

      Delete
  6. Dear EH, I think that very few things line up neatly with DSM categories! I probably haven't made it clear that, currently, I think it would be very hard to predict mass murders. Predicting single murders would also be hard, but predicting violence might be more manageable. It would certainly be worth looking at personality disorders. From memory, it seems that drug takers are 10 times more likely to be violent than ordinary citizens, so perhaps better screening and management there might be worth it.

    ReplyDelete