So here’s a recent New Scientist title: “Bad Boys Can Blame Their Behaviour on Hormones.”
All I can think is: New Scientist, Old School. Old, as in nature-nurture old and biological determinism old. Old as in moldy, rusted, failing ideas old.
But it’s not just New Scientist. Discover matches New Scientist with, “Teenage Hoodlums Can Blame Bad Behavior on Hormones.” And The Daily Mail delivers “Now Teenage Thugs Can Blame Their Hormones for Bad Behaviour.”
So what’s the problem? Well, it’s two-fold. First are journalists playing out a cultural script just like they subscribe to old-school cultural determinism. And second is some bad research that, not coincidentally, helps the journalists act like cultural automatons.
The cultural model goes like this: stereotypes, then blame, then biology. Take a stereotype we fear (“we” meaning journalists and readers alike). Bring in the politics and ideology of blame – hey, there’s a reason they are not like us, and why they threaten us. Invoke a cause, generally biological (though cultural causes come up too), outside of our particular realm of control. Hormones, nothing we can do about that, it means they were bad from the get-go. So we’re right to fear them and better make sure they don’t hurt us, whatever it takes.
Don’t believe me? Just look at the photos that accompany the articles. At the Daily Mail, a hooded guy point his hand like a gun at us the reader. Over at Discover, a crazed man with a clenched fist yells in our faces.
We all know journalists will play to stereotypes and will get research wrong and so forth. But in this case, like in most of the biologically-oriented research about complex human phenomena, the research only feeds into journalists typing out the normal crap.
The article in question is “Cortisol Diurnal Rhythm and Stress Reactivity in Male Adolescents with Early-Onset or Adolescence-Onset Conduct Disorder” (full access) by Graeme Fairchild, Stephanie van Goozen et al. and appears in the October 2008 issue of Biological Psychiatry. Neurocritic gives us the overview of the article if you don’t want to read the whole thing. (While I liked the Bad Boys music, I could have done with some more criticism in this particular Neurocritic post – but that’s okay, I’m going to play the bad boy this time.) Here’s the popular take from New Scientist on the article:
Out-of-control boys facing spells in detention or anti-social behaviour orders can now blame it all on their hormones. The “stress hormone” cortisol – or low levels of it – may be responsible for male aggressive antisocial behaviour, according to new research. The work suggests that the hormone may restrain aggression in stressful situations. Researchers found that levels of cortisol fell when delinquent boys played a stressful video game, the opposite of what was seen in control volunteers playing the same game.
The basic result of the study is that boys diagnosed with conduct disorder don’t have the same rise in cortisol in response to a stressful challenge, in this case a rigged game where the opponent taunts the player and it is impossible for the player to succeed. As the article says:
The competition began between 1 and 2 pm with a task involving confrontation, the Prisoner’s Dilemma Game (PDG), in which the opponent always failed to cooperate and sent antagonistic messages. Frustration was induced by having the participant perform a difficult, computer-based manual precision task (MPT) under time pressure while the video opponent and experimenter watched. By design, all participants failed to achieve their target score and received negative evaluations of their performance from the opponent.
So here is the fancy graph from Biological Psychiatry showing how the conduct disorder boys stayed close to controls who weren’t stressed while boys without conduct disorder has their cortisol go up in response to the challenge.
So what is so bad about this? Seems like a pretty clear case of a biological difference. End of story. The blame can start, since there’s obviously something wrong with those kids.
Just for the record, here’s the summary of the article:
BACKGROUND: Previous studies have reported lower basal cortisol levels and reduced cortisol responses to stress in children and adolescents with conduct disorder (CD). It is not known whether these findings are specific to early-onset CD. This study investigated basal and stress-induced cortisol secretion in male participants with early-onset and adolescence-onset forms of CD. METHODS: Forty-two participants with early-onset CD, 28 with adolescence-onset CD, and 95 control subjects participated in the study. They collected saliva across the day to assess their cortisol awakening response and diurnal rhythm. Subsequently, salivary cortisol was measured before, during, and after a psychosocial stress procedure designed to elicit frustration. Cardiovascular activity and subjective mood states were also assessed during stress exposure. RESULTS: There were no group differences in morning cortisol levels or the size of the cortisol awakening response. Basal cortisol levels in the evening and at 11 am during the laboratory visit were higher in both CD subgroups relative to control subjects. In contrast, cortisol and cardiovascular responses to psychosocial stress were reduced in both CD subgroups compared with control subjects. All groups reported similar increases in negative mood states during stress. CONCLUSIONS: Our findings suggest that group differences in cortisol secretion are most pronounced during stress exposure, when participants with CD show cortisol hyporeactivity compared with control subjects. There was no evidence for reduced basal cortisol secretion in participants with CD, but rather increased secretion at specific time points. The results do not support developmentally sensitive differences in cortisol secretion between CD subtypes.
So what’s wrong? Let’s start with the research design itself, before we move onto the overall approach and how that links into the supposed implications (promoted, of course) of this research.
Even just focused on the experimental methods, the study is suspect. The video game protocol was reported in a previous study, Salivary cortisol and cardiovascular activity during stress in oppositional-defiant disorder boys and normal controls, cited by the authors. The major conclusion of that work was: “Cortisol increase due to stress exposure was strongest in highly externalizing and highly anxious subjects.” Remember, this is the study to which they refer us to understand their main experimental protocol. It’s even done by the same main researcher, Stephanie van Goozen. But there is no direct measure of anxiety induced by the experimental protocol in this new paper.
Even worse, the protocol is designed to make them feel frustrated. No measurement of that. The researchers are proposing a link between low cortisol reactivity and aggression or conduct disorder. No measurement of aggressive impulses either.
It’s really breathtaking. They are working with individuals who have repeatedly been involved in aggressive situations and have dealt with their fair share of frustration in social situations where “conduct disorder” is a convenient label for being uppity (ah, did I mention that the conduct disorder kids came from a lower SES and had a greater percentage of minorities?). So, these kids are likely used to dealing with antagonistic opponents and negative evaluations. Why should that get a stress reaction out of them?
It’s absurd. And it could have been corrected by one small thing (well, besides measurements that actually get at the things that researchers are claiming). Say, a short interview after the experimental protocol to understand how the boys interpreted the test. But the researchers have absolutely no awareness of that need, and thus no sense of whether their measures capture anything realistic other than these kids have less cortisol reactions to situations that are likely similar to situations they have faced many times before.
The kicker is that Graeme Fairchild, the lead author, is aware of some of these problems and still did not address them in the research design. Here are some quotes from the press interviews: “It could be that they’re used to provocative situations and habituated to stress,” he says. Or: Researcher Dr Graeme Fairchild said: ‘The game was rigged to be impossible. The whole point is to make them feel angry and annoyed, and as if they were being socially evaluated.’
If the researchers want to make the claims that they do, then they should at least consider a more robust methodology. Besides actual measurements of the variables of interest, it would have been helpful to have at least two challenges in the study. An antagonistic, frustrating video game protocol could have been complemented with something, say a standard public speaking protocol, that would have provided some point of comparison.
Enough about the bad research design. Let us turn to the overall approach.
Describing something as “conduct disorder”, what a surprise, means it’s a mental illness, and thus should be treated as such. Did you realize that the researchers pulled that fast one right from the beginning?
Put differently, conduct disorder is simply something that psychiatrists have defined in the ways they see fit. In this case, it’s about behaviors like aggression and vandalism. The same thing happens with the other measures – subjective emotions, personality traits and the like. The only thing that really matters with these measurements is that they are reliable, that they give you the same scores more-or-less each time they are used. At times their connection to actual reality is quite tenuous.
So here we are dealing with a house of cards held together by statistics and the agreement of the research community – it’s their own cultural phenomenon, their own set of stereotypes and assumed causes and assigned blame. Think I sound absurd? Well, certainly I am pushing my analysis here, just the same as the researchers – this one cortisol drop in an experimental paradigm explains why these kids have problems.
Another thing that annoys me about this style of research is the lack of consistency, really of being thoughtful, by the majority of these types of researchers. They approach problems not as problems but as a chance to get data and get a publication. I already mentioned that they ignored one of the main results about anxiety from their previous research. But it’s worse than that. In a recent study, these same researchers showed that cortisol generally went up among antisocial/conduct disorder individuals: “reactive aggression was strongly correlated with elevated cortisol.” So one study it goes down, the other it goes up; one study it’s anxiety, the next study is that they have an emotional disconnect. That’s just being a bad scientist. I’d prefer someone who at least tries to think through problems alongside publishing research results and getting more funding.
The emphasis on biological causes also twists the accepted understanding of conduct disorder in the scientific literature, that it is a “biosocial” problem. A 2002 review by Adrain Raine, one of the major researchers in the field, states, “[W]hen biological and social factors are grouping variables and when antisocial behavior is the outcome, then the presence of both risk factors exponentially increases the rates of antisocial and violent behavior (pdf).”
These researchers’ own synthetic theoretical proposal goes “It is argued that serotonergic functioning and stress-regulating mechanisms are important in explaining individual differences in antisocial behavior. Moreover, low fear of punishment and physiological underactivity may predispose antisocial individuals to seek out stimulation or take risks and may help to explain poor conditioning and socialization.”
Serotonin and stress mechanisms are both sensitive to environmental influences; lower fear of punishment indicates an experience with punishment; socialization is a cause as well as a consequence. At least that’s how this neuroanthropologist sees it. They go back to nature vs. nurture, to a focus on “neurobiological deficits” not just as the mediators but often the assumed cause of antisocial behavior. It’s a safe bet, I am guessing, that these researchers rarely spend time outside their safe lab to see what life is really like in a rough neighborhood.
No, I am not saying it’s all social! That review by Raine, mentioned above, indicates that biological function can play an important role; for example, in well-to-do families, the link between neurobiological function and antisocial behavior are clearest, precisely since the social causes of antisocial behavior are not in the picture. But all this means is that the biology depends on context, a decidedly non-determinist view.
Conduct disorder and antisocial behavior are about behavior, types of social interactions, and violation of social norms. Here’s the American Academy of Child and Adolescent Psychiatry: “Children and adolescents with this disorder have great difficulty following rules and behaving in a socially acceptable way.” But these researchers went looking for biological differences, and gosh darn it, they found them. There is little sense that their research examined “why” in a holistic and integrative sense (biological and social) and in a way that takes into account the life experiences and perspectives of the individuals most affected by determinations of “antisocial disorder,” the boys themselves.
Rather, the researchers’ finding of biological difference then becomes the cause of behavior. Bad boys, whatcha gonna do?
That leads to the final and most scary part. Sure, journalists can get in wrong in selling a story. Researchers use bad methods and incomplete theoretical frameworks. It happens all the time. But in this case, the two add up to some frightening and completely wrong ideas about social policy and intervention.
Here’s the ending to the New Scientist piece:
The results also raise the possibility of finding biological markers in the blood of infants that identify those most likely to develop conduct disorders. Families and children could then be given help to manage and refocus their behaviour before it degenerates into the usual habits of lying, stealing, violence, malevolence and lack of concern for other people.
Alternatively, the research might lead to new drugs that have the same effect. It’s too soon to say whether extra cortisol would help. But Fairchild cites earlier experiments showing extreme violence in rats unable to make corticosterone, the rat equivalent of human cortisol. When the rats received extra corticosterone to compensate, it calmed them down.
Rather than people, we now have degenerate lab rats in need of injections and Big Brother re-training.
Or over at Discover, Graeme Fairchild is quoted as saying, “These findings basically indicate that antisocial behavior is probably more biologically based than many people recognize and is similar to conditions like depression and anxiety.” The Daily Mail actually opens with this point: “Teenage thugs could be suffering from a mental illness caused by a hormonal imbalance, scientists suggest today… Its findings point to the possibility of drugs being used in the future to control teenagers’ behaviour.”
So now antisocial behavior is a mental illness, rather than a social problem. Rather than tackling the social conditions that might bring about antisocial behavior, we are told that they have a mental imbalance and need to take drugs.
Indeed, the researchers aim to redefine the problem into biological terms and point to highly profitable ways for companies to extend their pharmaceutical reach and for governments to control their populations. It is a scary prospect, given the power of drug companies and the increasing reach of jails and other institutions for criminals and, now, “pre-criminals.”
Besides increasing those types of social control, disturbing questions arise, for example, if an “antisocial” child – determined by a blood test – is not given his drugs by the family, will that family lose the child? Will their taxes go up?
Biological determinism plays right into company profits and governmental oversight. If you like those things in your life, by all means, encourage one-dimensional research. But I for one think it would be quite nice if good research could help us get beyond letting other people determine my life.
Here’s a question along a very different line: If we think of antisocial disorder as related to neural processing, might it make more sense to think in terms of specific language deficits? Here early training programs and educational support can make a significant difference in making sure that specific deficits do not affect overall linguistic ability. Similarly, early interventions to help children interact in socially skilled ways could make a large difference for these kids. But no, it’s easier to say they have a biological problem, then drug them and walk away.
When Fairchild says, “A possible treatment for this disorder offers the chance to improve the lives of both the adolescents who are afflicted and the communities in which they live,” he has it backwards. How about improving the lives of those adolescents and the communities in which they live first?
Stereotypical reporting, bad research design, biological determinism, and social control – those are a very, very bad mix. Rather than helping us explore ways to work with these children, their families, their schools and their communities, this approach encourages us to stomp down, through harsh laws or harsh drugs, that which we don’t understand. In the end, it is much easier to blame their biology than to question ourselves.
14 thoughts on “Bad Boys or Bad Science”
Great post, you dissed them on a lot of points that I missed. Case in point, Table 1 and the failure to control for IQ and SES, among other things (I imagine the drug and alcohol use of the CD groups are considered part of their disorder).
I do have a question, though, about your class assumptions:
No, I am not saying it’s all social! That review by Raine, mentioned above, indicates that biological function can play an important role; for example, in well-to-do families, the link between neurobiological function and antisocial behavior are clearest, precisely since the social causes of antisocial behavior are not in the picture.
Are you saying that children who grow up in wealthy families cannot experience abusive treatment and toxic home environments that can lead to CD? In fact, for the adolescence-onset conduct disorder group, the largest proportion of participants were from high SES families. So is the cause purely biological in these boys?
While of course you’re entitled to your views and I believe that some of your criticisms are valid, I’d like to correct a few errors in your analysis:
You say: ‘there is no direct measure of anxiety induced by the experimental protocol in this new paper.’
In fact, the study did measure subjective states of the participants during stress, including feelings of anxiety, anger, embarrassment, and loss of control over the situation. These were not reported separately due to space limitations but the fact that these subjective feelings were measured is described in both the Methods and Results sections of the paper. It was striking that despite similar changes in negative mood states in all three groups (as noted in the abstract you copied and pasted into your critique), the typically-developing adolescents were the only ones to show increases in salivary cortisol levels following stress. The previous article you cited examined the relationship between presence of externalising and anxiety symptoms and cortisol responses to stress, not the relationship between changes in feelings of anxiety and changes in cortisol level, so your comment is based on a misunderstanding of these earlier results.
‘Even worse, the protocol is designed to make them feel frustrated. No measurement of that.’
As noted above, this is factually incorrect. Subjective states of anger were measured throughout the experiment (and afterwards, to make sure there were no lasting negative effects of the stress induction procedure).
‘The researchers are proposing a link between low cortisol reactivity and aggression or conduct disorder. No measurement of aggressive impulses either.’
Whether you agree or disagree with the DSM-IV diagnosis of Conduct Disorder, one thing is clear. Many of the symptoms of Conduct Disorder listed in the DSM-IV involve aggression and impulses, including initiating physical fights, use of weapons in such fights, physical cruelty to others, and forced sex. Most of the adolescents with a CD diagnosis endorsed at least one of these ‘symptoms’. For personal safety and ethical reasons, trying to cause individuals to act out aggressively during the session was not an aim of the study.
‘It’s really breathtaking. They are working with individuals who have repeatedly been involved in aggressive situations and have dealt with their fair share of frustration in social situations where “conduct disorder” is a convenient label for being uppity (ah, did I mention that the conduct disorder kids came from a lower SES and had a greater percentage of minorities?). So, these kids are likely used to dealing with antagonistic opponents and negative evaluations. Why should that get a stress reaction out of them?
It’s absurd. And it could have been corrected by one small thing (well, besides measurements that actually get at the things that researchers are claiming). Say, a short interview after the experimental protocol to understand how the boys interpreted the test.’
As noted above, subjective states of anger, embarrassment, etc. were monitored throughout the test. These data showed very clearly that a majority of the individuals from all three groups experienced the situation as stressful and challenging.
‘But the researchers have absolutely no awareness of that need, and thus no sense of whether their measures capture anything realistic other than these kids have less cortisol reactions to situations that are likely similar to situations they have faced many times before.
The kicker is that Graeme Fairchild, the lead author, is aware of some of these problems and still did not address them in the research design.’
I don’t know whether you are a researcher yourself, but it is not possible to address all possibilities and interpretations in a single piece of research. Rather, the progress of science is incremental and arriving at an understanding of a behaviour pattern as complex as aggressive or criminal behaviour will involve many studies. One potential limitation of the study was acknowledged with the following quote: “It could be that they’re used to provocative situations and habituated to stress.” But equally the reduced response to stress could be due to many other factors, such as polymorphisms in genes that determine stress reactivity, or the presence of early life adversity, rather than exposure to conflict in the participants’ current lives.
‘If the researchers want to make the claims that they do, then they should at least consider a more robust methodology. Besides actual measurements of the variables of interest, it would have been helpful to have at least two challenges in the study. An antagonistic, frustrating video game protocol could have been complemented with something, say a standard public speaking protocol, that would have provided some point of comparison.’
While this would undoubtedly have been helpful, it’s worth noting that other studies have used public speaking-based stress tests and found similar results (e.g. Popma et al., 2006, Psychoneuroendocrinology). I guess you could argue in this latter case, that they don’t care about authority figures’ impression of them, and this is why they don’t show a stress response in a public speaking situation. In addition, reduced stress responses to mild fear-inducing situations such as venipuncture have also been reported in children with Conduct Disorder. Perhaps you would argue here that they’re used to getting beaten around and therefore they’re less sensitive to pain.
The last thing worth pointing out is that despite the press hype, no claims were made in the article about causality. The design of the study means that it is not equipped to tell whether reduced physiological responses to stress cause antisocial behaviour. The only way to address these issues would be to run longitudinal studies examining these same issues, and to experimentally increase cortisol levels during conflict situations to assess whether such changes improve an individual’s ability to control their temper or aggressive impulses.
Neurocritic, thanks for the comment, always appreciated. That Raine point took me awhile to figure out myself, and I’ll get to that. But first to answer your questions. For me, certainly children in wealthy families can experience abusive treatment and toxic home environments, and that can still be a precipitating factor for CD in those kids. I could also see CD (or acting out more generally) as linked to attention getting, ineffective rule enforcement, an environment where bad behavior is spoiled, and so forth. So, surely, any biological predispositions in this case need context and take form in social interaction. And Raine indicated that birth complications are one well-known risk factor for CD, and those can also happen in wealthy environments. So for me, I would still take an interactionist view.
In the Raine paper, he spends a considerable amount of time on the class issue. Let me quote, “A number of studies have shown that psychophsyiological factors show stronger relationships to antisocial behavior in those from benign social backgrounds that lack the classic psychosocial risk factors for crime. For example, although in general resting heart rate is lower in antisocial individuals, it is a particularly strong characteristic of antisocial individual from higher social classes (Raine and Venables 1984b) [and other studies]… One explanation for this pattern of results is the ‘social push’ hypothesis. Under this perspective, where an antisocial child lacks social factors that ‘push’ or predispose him/her to antisocial behavior, then biological factors are more likely to explain antisocial behavior (Mednick 1977, Raine & Venables 1981). In contrast, social causes of criminal behavior may be more important explanations of antisociality in those exposed to adverse early home conditions. This is not to say that antisocial children from adverse home backgrounds will never evidence biological risk factors for antisocial and violent behavior – they clearly will. Instead, the argument is that in such situations the link between antisocial behavior and biological risk factors will be weaker (relative to antisocial children from benign social backgrounds) because the social causes of crime camouflage the biological contribution. Conversely, in the case of antisocial children from benign social backgrounds, the ‘noise’ created by social influences on antisocial behavior are minimized, allowing the biology-antisocial behavior relationship to shine through (314).”
So that’s the Raine argument. Certainly we can’t assume that benign-adverse map onto social class exactly; I’ve known plenty of rich kids from bad homes and poor kids with great parents. But in the case of criminal behavior, particularly the blue-collar crime we judge more harshly than white-collar crime, kids from lower SES are (all things held equal) more likely to have peers engaged in vandalism and petty theft and to also have more motive opportunity where “antisocial behavior” can provide social and financial rewards (e.g., entry into a gang, money in your pocket or shoplifting). When your friends are doing it, and there are important social and economic reasons favoring a behavior, then it’s not too surprising to see somebody engage in that behavior.
Credit Crunch, thanks for the lengthy comment. It’s always good to get differing opinions and critical responses on a site like this, so I appreciate that. And obviously this is an issue that has me worked up, so I was edging towards the anti-social in the tone of the post. I don’t know if my cortisol was up or down, but I do understand that it is impossible to get at all possibilities and interpretations in an experimental design, and that this is an approach that aims to take steps forward. An interactive experimental protocol; repeated measurement of mood and cortisol; the inclusion of a baseline control, two types of CD kids, and an experimental control – I applaud these things.
Still, I think the research could have been done better, as I will address again below. In retrospect, I recognize that I could have framed the blog post more along the lines of constructive criticism, with a less antagonistic tone. But precisely because the media claims and social policy implications are so high about these kids’ lives, and the experimental results are often taken as indisputable facts about these kids’ lives, the limitations of the study got me worked up.
So let’s go back to the study, where we disagree about the measurement issue. The subjective states were measured according to the following protocol:
“Participants rated their feelings eight times using an adaptation of a clinical self-rating scale (31). The scale contained 11 items (happy/gloomy, well/sick, cheerful/not cheerful, good/bad, liked/not liked, satisfied/not satisfied, worried/not worried, embarrassed/not embarrassed, ashamed/not ashamed, afraid/not afraid, and angry/not angry), which participants rated using 9-point ordinal scales. They also rated their feelings of control and confidence about winning the competition. Subjective ratings occurred at similar times as saliva collection, apart from the second rating, which was completed after the PDG and the fifth and sixth ratings (before and after the opponent played the MPT).”
One measurement problem here is that these are global measurements of mood, not specific measurements related to the experiences of playing the video game. Take frustration, for example, which Graeme Fairchild highlighted in the press as a main effect of the video game protocol. In my mind, frustration and anger are not the same thing, so measuring subjective states of anger does not get at what might be stressful or not for the game.
In this context, it is telling that feelings of control and uncertainty (linked to stress) about the game did not differ between the groups. To quote from the paper, “There were effects of time on controllability (p < .001) and confidence about winning (p < .001), with participants feeling less in control (all p < .01) and less confident (p < .05) at all time points during the stressor relative to baseline. However, there were no group effects on these variables or group × time interactions.” So the specific measurements linked to game showed no statistically significant difference.
Obviously the game protocol aroused negative emotions in all the players. Here’s what the paper reports about specific differences in subjective experience: “Group effects were seen for feelings of fear, worry, embarrassment, and shame, with post hoc tests showing lower levels of these emotions in EO-CD participants relative to control subjects.”
In my mind, “fear, worry, embarrassment, and shame” shows concern about performance, both with respect to the antagonistic opponent and to the scientists there judging them. For the “normal” kids, that was stressful (elevated cortisol). For the EO-CD kids, who have experienced plenty of antagonistic interactions and negative social judgments in their lives, they still felt negative during the game – who wouldn’t, with a rigged game. But concerns about their performance did not set off a stress reaction. The CD kids have “been there, done that” so why get stressed?
Getting a stress response when concerned about performance from “normal” controls is not the same thing as a lack of stress response being a main motive for the CD kids’ antagonistic behaviors.
But this issue could be partly resolved in future research by specific measurements that get at previous experiences of antagonistic interactions and negative social judgments, rather than just a diagnosis of CD. That would control for one possible confound. As you note, other factors such as gene polymorphisms and early life adversity could also prove useful.
But there is something even more important. Asking about aggressive impulses towards the antagonistic player during the game protocol, that would have been a robust measurement. Questions like “are you mad at the other player” or “did you want to hurt him,” those could have proven useful. That is the sort of specific measurement, rather than global mood assessment, that would provide clearer indications of how the CD kids are reacting. There is no need to try to “cause individuals to act out aggressively.” If you can measure global mood, certainly you can measure feelings of antagonism. That is crucial, because of what is at stake – claims about what drives criminal behavior.
Still, all these criticisms and counter-criticisms are speculation. As I wrote, an effective way to know what sorts of reactions the kids are having is to ask them in an interview to describe their experiences about playing the game. With this approach, the kids can speak for themselves instead of answering structured questions set by others. It would also be an effective way to see how to better develop measurements and to make sure that the protocol is ecologically valid.
Finally, about causality. In your comment, I agree with you about the importance of longitudinal research. But I do find it telling that you write, “The design of the study means that it is not equipped to tell whether reduced physiological responses to stress cause antisocial behavior.” Biological determinism is already there in the way you phrased it – reduced stress responses cause criminal behavior. Given what we know about stress as an interactive system (without even addressing crime), I find this quite objectionable.
And in the paper itself, the conclusion goes like this: “In summary, this study demonstrated that cortisol and cardiovascular responses to an ecologically valid psychosocial stressor were reduced in adolescents with both early-onset and adolescence-onset CD. These findings were not explained by differences in subjective responses to the stressor, suggesting a discrepancy between mood changes and physiologic reactivity in CD.”
Scientists are very careful about making claims of causality in their papers – that’s an easy peer-review criticism. So it is always better to couch assumed causes in less direct language. The way I read that above statement is: (1) CD kids have reduced physiological stress responses; (2) subjective responses don’t explain that, so it’s the reduced stress reactivity that explains CD.
So the press read it right, it seems to me. If the subjective responses don’t explain criminal behavior, it must be the biology.
Great blog, I look forward to reading more. Thanks!
The Neurocritic, who runs such a good site, added this addendum to the Bad Boys post over there, which I think is quite useful as it recognizes me going thrashingly overboard…
ADDENDUM: For a lengthy and thorough thrashing of this article (and of The Neurocritic’s light treatment thereof), see Bad Boys or Bad Science by Daniel Lende of Neuroanthropology. Although my sarcastic COPS reference was an admittedly lazy critique of the press coverage and only an oblique criticism of the diagnosis of conduct disorder, I believe Lende goes overboard in saying the authors are complete biological determinists, because they did acknowledge the alternative [i.e., see the Fairchild et al. quote above, “…rather, both CD subgroups may have experienced increased social adversity during development (e.g., maltreatment)…”].
A colleague of mine also emailed me with this comment about the post and the research:
The low CORT – “antisocial” behavior link is an interesting and curious one, and pretty robust. [It has been] found in Nepal [and elsewhere]. I agree with you, though, that a big problem with these “standardized” stressors is their essential non-comparability in terms of meaningful impact across individuals from different backgrounds. One work-around is to ask individuals to narrate events of a specific nature (offensive, emabarassing, scary, etc.) that happened to them personally, but of course that has its problems as well.
It is a tough thing, when your science gets out into the media. They like to twist things around immediately… You nailed it though – Fairchild’s quote is incriminating – “These findings basically indicate that antisocial behavior is probably more biologically based than many people recognize and is similar to conditions like depression and anxiety” – that is definitely old-school dualistic thinking. What happened to context?! I find it doubtful that low cortisol could ever “cause” aggressive behavior; it is more a symptom of chronic dysregulation perhaps, or chronic under-arousal (or lack of connection to social context), but I don’t know of a pathway through which it could directly “cause” antisocial behavior. I agree that a degenerate incapable of producing any CORT is not a good model here!!
So here is to acknowledging alternatives, contexts, robust links, criticisms, developmental pathways, social adversity, and all the rest.
Thanks for your response, Dlende.
Although it sounds appealing on the surface, there are at least two problems with your colleague’s suggestion to use the retrospective narration of meaningful adverse life events as a stressor. The first is that such a procedure would not be standardised and there would be considerable inter-individual differences in the ability to vividly recall or imagine the adverse events that were being described (regardless of whether they concerned an embarrassing or a frightening situation). Cortisol responses are highly variable both between and within individuals even when standardised procedures are used. This suggestion would be likely to introduce even more variability and problems with interpretation of results.
The second, and more important, problem is that such a procedure would probably not cause physiological stress responses (and particularly increases in cortisol release) in most individuals. Most neuroendocrine studies involving children and adolescents have failed to elicit stress responses in the majority of the participants (see Gunnar & Vasquez, 2006, for a review), and it actually requires a fairly intense stressor to elicit meaningful cortisol increases in > 50% individuals. Even the most widely used social stressor (the Trier Social Stress Test (TSST)), which involves public speaking in front of an audience, often does not cause statistically significant cortisol increases in studies of children or adolescents. A recent meta-analysis showed that in order for stressors to be effective, they had to include elements of social evaluation and induction of feelings of loss of control (Dickerson & Kemeny, 2004). It may also interest you to hear that some adult studies using the TSST have shown larger stress responses in low SES relative to high SES participants, so it doesn’t necessarily follow that being of low SES means that you are relatively unresponsive to stress. Also, the literature on the impact of early adverse life events on subsequent stress responses in adulthood is pretty mixed. Some studies find exaggerated cortisol or cardiovascular responses to stress in adults who have been abused as children, whereas others find that stress responses are blunted in such individuals.
I think we all agree (at least, the reasonable among us) that social and biological (and genetic) factors are critical. The interactions between them are key. Differing points of view are just a matter of emphasis. Again, that holds for logical and scientific thinkers, not for abusive anti-psychiatry trolls who launch into tirades of personal insults and profanity at the mere mention that any form of mental illness might have a neurobiological basis (as one among other influences).
I happen to have my very own pathetic and deluded Australian anti-psych troll, who left this [unpublished] comment on my Bad Boys post. An edited excerpt:
Pushing unreplicated piece of sh*t studies in an effort to undermine the concept of free will, and brand boys with your misandric f*cking pseudoscience. Of course you eugenic social darwinist c*nts are nothing but biopsychiatric fanatics, that much is clear.
Neurocritic, that is just awful! The troll obviously hasn’t read much of your blog.
Haha, I am a student and i had to present this study (i choose it after the seemingly bais news report in my local). If only i had seen this blog!!, i did however (much to the nods of my Prof) state the clear flaw that this study doesn’t make sense of the fact many of the CD participants could be habituated to stress due to thier increased social antagonist encounters due to thier lifestyle.
Loved the thrashing though!
Easy to dismiss the research when you have never had a single original discovery of your own
Elaine, rather than trolling, it would be useful if you discuss substantive points or indicate research that you have published that supports or adds to what’s under discussion.