Mandatory treatment for generic plurals?

« previous post | next post »

Neurocriminology is a hot topic. From Isabella Bannerman, recently published in the Six Chix series:

From Peter Nichols, "Body of Evidence: Neurocriminologist Probes the Biology of Crime", recently published in Penn Arts & Sciences magazine:

In the mid-19th century, Italian physician Cesare Lombroso was doing an autopsy on Giuseppe Villella, a notorious brigand who’d spent years in the prisons of Pavia. Peering into the dead criminal’s skull case, Lombroso thought it resembled the crania of “inferior animals,” particularly rodents. “At the sight of that skull,” he wrote, “I seemed to see all of a sudden, lighted up as a vast plain under a flaming sky, the problem of the nature of the criminal.”

Often credited as the father of criminology, Lombroso hypothesized that violent behavior could be explained by cranial, skeletal or neurological deformities. Some people were just “born criminal,” he reasoned. Biological malformations—“stigmata” he called them — suggested that lawbreakers were throwbacks to an earlier, more brutish stage in human evolution.

“He was fascinated by the idea that there was a biological brain difference between criminals and the rest of us,” notes neurocriminologist Adrian Raine. From that day on, Lombroso took careful measurements of faces, jaws, heights, weights and other physical traits to gather data in support of what he called his “revelation.” […]

“Lombroso’s theories sound a bit ridiculous to us,” Raine comments, “but in a way he was right.” With the emergence of new and powerful imaging technologies, scientists can see detailed pictures of the brain and trace activity along its neural networks. “The brain was forgotten until neuroscience techniques evolved to a level where we could, for the first time, really look at brain structure and function,” he says. “And from then on, we found that there’s certainly a brain basis to crime—that the brains of violent criminals are physically and functionally different from the rest of us.”

We're a long way from thinking-of-parking tickets.  But if it's really true that "the brains of violent criminals are physically and functionally different from the rest of us", it's logical to ask whether some sort of diagnosis and mandatory treatment is appropriate.

As I understand it, the American legal system doesn't in general permit us to directly criminalize a mere statistical disposition towards criminal behavior.  However, there seem to be several cases where something very much like this has happened in matters of public health (like mandatory treatment of tuberculosis or the involuntary commitment of lepers) and in sentencing for certain types of crime (like "Megan's Law" registration and tracking of sex offenders).

I don't know much about this, and would appreciate historical, legal and philosophical instruction from those who do.  But the topic of this blog is linguistics, not legal theory or moral philosophy, and so my point is a linguistic one.  I propose a voluntary ban on the use of generic plurals to express statistical differences, especially in talking to the general public about scientific results in areas with public policy implications.

In other words, when we're looking at some property P of two groups X and Y, and a study shows that the distribution of P in X is different from the distribution of P in Y to an extent that is unlikely to be entirely the result of chance, we should avoid explaining this to the general public by saying "X's have more P than Y's", or "X's and Y's differ in P", or any other form of expression that uses generic plurals to describe a generic difference.

This would lead us to avoid statements like "men are happier than women", or "boys don't respond to sounds as rapidly as do girls", or "Asians have a more collectivist mentality than Europeans do" — or "the brains of violent criminals are physically and functionally different from the rest of us".  At least, we should avoid this way of talking about the results of scientific investigations.

The reason? Most members of the general public don't understand statistical-distribution talk, and instead tend to  interpret such statements as expressing general (and essential) properties of the groups involved. This is especially true when the statements express the conclusions of an apparently authoritative scientific study, rather than merely someone's personal opinion, which is easy to discount.

Now, there are obviously cases where group differences rise to the level where generic plurals are appropriate. Is one of those cases the distribution of anatomical and physiological differences in the brains of criminals? I invite you to read Adrian Raine, "The biological basis of crime" (in Wilson and Petersilia (Eds), Crime: Public Policies for Crime Control, 2002) and decide for yourself. I'll give you my opinion in another post.


  1. Garrett Wollman said,

    September 13, 2009 @ 10:55 am

    Yes, absolutely, right on!

    I would note that S.J. Gould was tilting at this particular windmill twenty-five years ago:

    All versions written for nonscientists speak of fused males as the curious tale of the anglerfish—just as we so often hear about the monkey swinging through the trees, or the worm burrowing through soil. But if nature teaches us any lesson, it loudly proclaims life's diversity. There ain't no such abstraction as the clam, the fly, or the anglerfish. Ceratioid anglerfishes come in nearly 100 species, and each has its own peculiarity.

    ("Big Fish, Little Fish", as republished in Hen's Teeth and Horse's Toes (1983))

  2. Neil Dolinger said,

    September 13, 2009 @ 11:28 am

    Have there been any studies done to gauge how well newspaper and periodical readers understand the statistics presented to them? It's certainly clear that many journalists oversimplify the meaning of statistical results, and it *feels* to me like many people take these misleading reports and run with them, but do we have any, hmmm, statistically based proof?

    [(myl) I'm not aware of any experiments of exactly this kind. There have been many experimental investigations of quantifier interpretation, but I don't know any that focus on generic plurals in the context of statements about the results of scientific research.

    I once suggested an approach based on gambling games, in "Betting on the poor boy: Whorf strikes back" (4/19/2009). There are plenty of alternative paradigms, both direct (ask people what a sentence means, or which situations it's true in) and indirect (reaction times; physiological measures of surprise; willingness to gamble; eye-tracking effects). You could also look at subjects' propensity to use generic plurals — though I would predict an asymmetry, whereby people who use generic plurals to describe marginal tendencies will nevertheless interpret others' generic plurals as closer to categorical, especially in science-reporting contexts.

    One recent study, not directly relevant but worthwhile as background, is Amanda Brandone and Susan Gelman, "Differences in preschoolers’ and adults’ use of generics about novel animals and artifacts: A window onto a conceptual divide", Cognition 110(1): 1-22, 2009:

    Generic readings cannot be equated easily with readings associated with any one quantifier. Unlike statements using “some” (e.g., “Some boys play with trucks”), generics (e.g., “Boys play with trucks”) invoke the entire category. However, unlike statements using universal quantifiers such as “all”, “every”, or “each”, generic statements allow for exceptions (Lawler, 1973). For example, whereas the statement “All boys play with trucks” is certifiably false, the generic statement “Boys play with trucks” is considered true, despite the fact that some boys do not in fact play with trucks. Because of their generalizability and resilience against counterexamples, generic constructions have been proposed to be an especially powerful and robust way to express properties that are characteristic of a kind.

    Another place to look for background would be S.A. Gelman, The essential child: Origins of essentialism in everyday thought,2003. From the book overview:

    Essentialism is the idea that certain categories, such as "dog," "man," or "intelligence," have an underlying reality or true nature that gives objects their identity. Where does this idea come from? In this book, Susan Gelman argues that essentialism is an early cognitive bias. Young children's concepts reflect a deep commitment to essentialism, and this commitment leads children to look beyond the obvious in many converging ways: when learning words, generalizing knowledge to new category members, reasoning about the insides of things, contemplating the role of nature versus nurture, and constructing causal explanations. Gelman argues against the standard view of children as concrete or focused on the obvious, instead claiming that children have an early, powerful tendency to search for hidden, non-obvious features of things. She also attacks claims that children build up their knowledge of the world based on simple, associative learning strategies, arguing that children's concepts are embedded in rich folk theories. Parents don't explicitly teach children to essentialize; instead, during the preschool years, children spontaneously construct concepts and beliefs that reflect an essentialist bias.

    I believe that this tendency persists to a variable extent into adulthood, and that generic statements summarizing scientific results — even statistically marginal ones — tap into these residues in a powerful way. But this is only an opinion, not the result of any experiments that I know of. ]

  3. John Lawler said,

    September 13, 2009 @ 12:24 pm

    Well, Tufte's books are a good place to start on the question of popular understanding of statistics, especially when they're presented via PowerPoint.

    As for generics, there are of course, three different generic NP constructions in English, of which the plural generic "refers to the Norm of the species over its individuals, as perceived, of course, by the speaker, who is unlikely to have conducted […] surveys, so the 'statistics' here are very vague and impressional."

    In other words, Garbage In, Garbage Out. So I'd support the ban, in principle, at least; though the ACLU might go either way.

  4. bianca steele said,

    September 13, 2009 @ 1:15 pm

    Does your proposal include a ban on adjectives, or only on collective nouns? In other words, are misunderstandings less likely for statements like "brown-eyed third-graders score lower on tests of reading comprehension than blue-eyed third-graders," than for "blue-eyedness is correlated with verbal intelligence"?

    [(myl) I'm (half-seriously) suggesting a ban (in popular science writing) on generic plurals as a way to express statistically-significant group differences with substantial overlap of distributions. This ban would cover "brown-eyed third graders score lower…" but would say nothing about "blue-eyedness is correlated with …"

    More generally, the idea is to avoid promoting essentialist thinking about human sub-groups, especially in cases where the reported facts involve group means that differ by well under a standard deviation, or have been estimated from small and perhaps-unrepresentative samples, or both. Avoiding unquantified generic plurals is neither a necessary or a sufficient condition for reaching this goal. But it's a useful slogan to start with, I think. ]

  5. Andrew said,

    September 13, 2009 @ 1:37 pm

    My own impression is that a large number of people (especially in the social sciences) don't really understand the meaning of statistics in academic publications either. Just ask yourself when the last time you saw someone use p>.05 to affirm the null hypothesis–even when they're working with a pretty small sample size that doesn't have much statistical power? It's so painfully commonsplace that I've heared (in lectures, even) terms like "statistically identical" to express p>0.05.

  6. Larry Goldsmith said,

    September 13, 2009 @ 6:10 pm

    Philip Kerr's novel _A Philosophical Investigation_ offers an interesting perspective on some of the ethical issues involved in dealing with the putative biological bases of crime. In a fictional future, scientists have discovered a physiological basis that predisposes a person to become a serial killer–though it is only a predisposition, not an inevitable result. The government tests everyone and keeps the results in a database, carefully guarded to protect the rights of innocents and available only to law enforcement for leads when there is a serial killer on the loose. But of course someone breaks into the database, and the police find themselves faced with a serial killer targeting only people predisposed to be serial killers. As the serial-killer-of-serial-killers is of a philosophical bent, there is a lot of Wittgenstein and more than a little linguistics along the way.

  7. J. W. Brewer said,

    September 14, 2009 @ 1:09 am

    Can you give us an alternative stock form of sentence using X, Y, & P that you would consider suitable for use by journalists attempting to convey the sort of research result you're talking about while avoiding the risks of misunderstanding you perceive?

    Violent crimes are overwhelmingly committed by possessors of a microscopic genetic marker commonly called the "Y chromosome," and this statistical correlation seems pretty robust across times and cultures. But we've generally avoided "diagnosis and mandatory treatment" for the criminal propensities thus implied. It could actually be a fun exam question for a law school evidence class to ask the students whether the prosecution ought to be able, under the applicable rules of relevance etc etc etc, to first introduce evidence tending to establish that the defendant is male and then put on expert testimony tending to establish that the type of crime in question is committed, for example, 91% of the time by males.

  8. Graeme said,

    September 14, 2009 @ 6:43 am

    A little off topic, but the cartoon reminded me of my Civil Rights law professor telling the class that whilst the International Covenant on Civil and Political Rights enshrined 'freedom of thought', it was purely salutary. There were not – could not be – any thought crimes.

    Though I felt he was being excessively behaviouralist, I couldn't conjure any examples to contradict him. Until I re-read our stock high school play, 'A Man For All Seasons', Bolt's rendition of the end of Sir Thomas More.

  9. Graeme said,

    September 14, 2009 @ 6:58 am

    Jurisdictions, internationally, have moved beyond Megan's Laws to incarcerating according to 'scientific' propensity to offend (sexually). Offenders can be subject to ongoing, annual, extensions of incarceration, despite having completed their original sentence for an actual serious sexual offence.

    The courts have justified this, despite breaching traditional rule of law/innocent till proven guilty principles, as 'community protection' rather than punishment. In truth it is both. In the annual hearings the courts expect psychologists to predict the future.

  10. Chris said,

    September 14, 2009 @ 9:30 am

    Violent crimes are overwhelmingly committed by possessors of a microscopic genetic marker commonly called the "Y chromosome," and this statistical correlation seems pretty robust across times and cultures. But we've generally avoided "diagnosis and mandatory treatment" for the criminal propensities thus implied.

    The usefulness of the Y chromosome test for violent crime is seriously impaired by a lot of false positives. Many people have Y chromosomes but do not commit any violent crime – in fact, this group may outnumber all violent criminals, both with and without the Y chromosome.

  11. Acilius said,

    September 14, 2009 @ 10:55 am

    "Most members of the general public don't understand statistical-distribution talk, and instead tend to interpret such statements as expressing general (and essential) properties of the groups involved."

    That's undoubtedly true, and it would of course be extremely difficult to create a situation where most members of the general public reflexively analyzed generic plurals in statistical terms. But would your proposal really be any easier to follow? Wouldn't an ethic that required us to avoid those sorts of statements be likely to collapse into a general unwillingness to make any definite statements at all?

    Second, what reason do we have to believe that a society where it was not the favored mode to use generic plurals to describe generic differences, even a society where no one used generic plurals to describe generic differences, would contain any less bigotry or any more cooperativeness than would a society where people commonly use generic plurals to describe generic differences? Chris' example above of the y-chromosome test for violent behavior shows that where a group is not already the target of discrimination, we are not likely to draw invidious conclusions from generic statements about that group. And certainly it is possible to be violently prejudiced against a group without making statements about that group using generic plurals to describe generic differences.

    Third, your reference to essentialism raises a serious problem. Wouldn't an ethic that forbade the use of generic plurals to describe a generic difference threaten to make it difficult to criticize nominalism? Nominalism may be true, essentialism may be false; certainly it is fashionable at the moment to think so. But are we so dogmatically certain of the truth of nominalism that we want to reconfigure our habits of language so that we can barely express its forbidden contrary?

    Finally, wouldn't we take more pride in an effort to raise the level of statistical literacy in the general population than in an effort to keep from troubling the uninformed with habits of speech that might confuse them? For my part, regardless of the outcome of these projects, I would certainly consider the first to be far more capable of honor than the second.

    Nor am I convinced that the outcome of an effort to inure the general population to statistical habits of thought would be bad. There may be some members of the general population who cannot or will not acquire the knowledge and habits necessary to hear a sentence like "Men are taller than women" without interpreting it to mean "Every man is taller than every woman." There may even be some who will never be able to hear such a sentence without concluding "Women's sports are a waste of time." Surely, however, such unfortunates are in the minority. And among the majority who would benefit from statistical education, who can doubt that most would benefit greatly?

  12. J. W. Brewer said,

    September 14, 2009 @ 11:44 am

    Actually, perhaps because they typically have considerable independent knowledge of the relevant underlying data, I would think that very few people think the statement "Men are taller than women" means the shortest man is taller than the tallest woman and/or that pointing to a particular 5'2" man or 6'2" woman makes the statement untrue. If you showed them a graph with overlapping bell curves of the male and female height distributions in a given population with a fairly simple explanation, many would probably get it and say "yeah, that's pretty much what I meant by 'Men are taller than women.'" The question is then whether they will systematically misinterpret other statements in the same syntactic form where they don't have a good prior sense of what the data likely looks like, perhaps by implicitly assuming a larger distance between the peaks of the respective bell curves than the data, if graphed, would show. (If the problem with the study is that the sample size is too small or poorly selected to scale up reliably to the group it's supposed to represent, that seems like a separate scientific/journalistic integrity issue than how to describe overlapping bell curves with a difference betweeen means that is deemed statistically significant but much less than 1 SD.)

    But maybe there are other syntactically identical statements that due to context and the listener's prior knowledge of the world are interpreted differently. Take "men can run faster than women." Is this more likely to be understood as a claim about means or as a claim about the rightward tails, i.e. that at any given level of track competition from a county high school championship to the Olympics the winning female runner in any given event will usually post a time slower than the winning male runner and often a time slower than the slowest male runner who qualified for the male-only finals? And really, is there any obvious non-essentialist justification for segregating athletic competitions by sex ("women's sports are a waste of time" could just mean "integrate now and may the fastest runner regardless of sex win")? I certainly don't object to the practice myself, but I'm not on an anti-essentialist crusade.

  13. Jens Fiederer said,

    September 14, 2009 @ 5:32 pm

    I'd be happy to support that (not for any moral reason, but because the usage is communicates poorly), but I doubt proposing it is going to make it happen. And limiting the restriction to statistical properties leaves statements like "Men have an X and a Y chromosome, women have two Xs" still valid.

  14. Acilius said,

    September 15, 2009 @ 8:19 am

    @Jens Fiederer: "And limiting the restriction to statistical properties leaves statements like "Men have an X and a Y chromosome, women have two Xs" still valid." You might want to ask Caister Semenya about that one…

  15. James said,

    September 15, 2009 @ 11:50 am

    Why do people interepret bare plurals that way? Do they always? I'm sure that the predicate or context or real-world knowledge must interfere in someway:

    Knights are brave. (They all are.)
    Lions have manes. (Only half of them actually do.)
    Owls hunt at night. (Not all owls hunt every night.)

  16. Women’s happiness, part one: Background « economic woman said,

    September 23, 2009 @ 12:39 am

    […] overlap is not what the Huffington Post et al imply, and that's important. Language Log has some great suggestions on how to better describe this sort of research: When we're looking at some property P of two […]

  17. Neuroskeptic said,

    October 1, 2009 @ 3:18 pm

    There was an excellent post over at Cognition and Culture last month that argues that this is not just a problem for lay people.

    Psychologists who study cultural differences, just like the rest of us, use the p<0.05 = significant rule. So if British people (or Europeans, or Asians, or whatever) differ from French (or African, or whatever) people on some measure at p<0.05 that's a "cultural difference".

    And once you have a cultural difference you can start to talk about the causes and the implications etc. Even if the difference is small and set against the background of overwhelming cross-cultural similarity.

    A good example is a study I wrote about finding that Asian and British people differ, slightly, in their accuracy at recognising certain emotions in facial expressions. But both groups were able to recognise every emotion at much better than chance. To me, that's evidence of essential universality, plus maybe some cultural quirks. But this was not how it was interpreted…

RSS feed for comments on this post