Graphical abstract from the Karpinski paper
There is a new study, High Intelligence: A risk factor for psychological and physiological overexcitabilities, published in the journal Intelligence, that has already garnered some attention from science journalists:
Hyper Brain, Hyper Body: the Trouble with High IQ in Neuroscience News
Having a High IQ May Lead to Increased Risk of Mental Illness in Thriveworks
First of all, I agree that the study is interesting and validates some of my personal experience. Also, I do think the concept of overexcitabilities can be a useful way to conceptualize gifted experience, and I am very encouraged to see any research on these topics. Research on giftedness usually takes a low priority, especially research on the kinds of difficulties gifted people face. However, when I read this paper, I was really disappointed by several major weaknesses in both their experimental design and their conceptual model. The best thing I can say of this study is that it encourages further study. The downside is the spawning of weak and credulous science reporting that contributes to negative stereotypes of gifted people and plays into the public’s willingness to define the problems of giftedness as defects located solely within gifted people: “hyper body” and “hyper brain”.
Issues with the study design
The authors claim to have demonstrated a correlation between IQ and psychological and physiological overexcitabilities. However, they did not test IQ (which presumably varies among Mensans) or use any other psychometric testing for overexcitabilities or psychological conditions, nor did they test specific physiological measures. Instead, they took a non-random sample of American Mensa members and gave them a questionnaire asking them to self-report whether they had specific physical and psychological conditions. The results of the questionnaire were then compared to the prevalences for those conditions found by the 2015 National Health Interview Survey and other national health surveys.
The Mensa questionnaire and the national health surveys did not ask the same types of questions, and they were not administered to groups with similar compositions in terms of age, race, socioeconomic status, education, or gender distribution.
“Participants included 2213 that identified as male (60%), 1472 that identified as female (40%), 22 who identified as non-cisgender, and 8 who declined to state. Regarding race, 87.9% of participants reported being European American, 2.7% as Asian American, 2.6% as Hispanic or Latino, 1.6% as African American, and 5.2% as other, mixed race, or who did not respond. The age range of participants was from 18 to 91 years (M = 53, SD = 15.18). Thus, the majority of participants were older, male, and European American.”
For comparison, the US Census Bureau reports that 51% of the population is female, 77% European American, 13% Black or African American, and 6% Asian. Essentially, the national survey data and the Mensa questionnaire data were apples and oranges. No attempts were made to control for other factors that could contribute to differing rates of various health conditions. Socioeconomic status is potentially a large contributor here. As Cory Doctorow pointed out on Boing Boing in his blog, Some thoughts on whether intelligence is linked to anxiety and depression,
“…in the US, access to medical care and thus formal mental health diagnosis is strongly correlated with income. That means that you’re more likely to have a professional affirm that your distress is caused by a clinical disorder if you’re wealthy than if you’re poor. High IQ correlates with wealth, and the wealth distribution in the study draws from disproportionately high-income subjects (“41.7% earned over $100,000; 16.9% earned between $76,000 and $100,000; 20.1% earned between $51,000 and $75,000; 14.9% earned between $26,000 and $50,000; and 3.9% earned less than $25,000″) so maybe they’ve found a correlation between wealth and mental health diagnoses.”
There is also very little evidence to support the assumption that American Mensa members are even representative of the gifted population. To get an idea of how restricted their pool of volunteers was, consider that the population of the United States is roughly 320 million people. Approximately 2% of those people could be expected to have IQ scores of 130 or higher (Mensa membership is open to those in the top 2% of intellectual ability). That would be 6.4 million people. American Mensa claims over 50,000 members, which represents less than 0.8% of intellectually gifted Americans.
From this study, we can conclude only that people who join Mensa (and who volunteer for studies) are more likely to report having health conditions than the general population of the United States. Indeed, the fact that the respondents reported markedly higher rates for every condition the questionnaire targeted as “related to overexcitabilities” should be a red flag. A stronger study would rely not on self-report, but on actual medical records or physiological and psychometric measures.
Issues with the conceptual model
The authors offer a conceptual model for the mechanism associating overexcitabilities with health conditions, but they fail to give any a priori operational definition for determining which health conditions are related to overexcitabilities. Indeed, they did not administer any instrument to determine the degree of overexcitability in those they surveyed. An example of this confusion is their classification of Autism Spectrum Disorder (ASD):
“Because ASD is listed in the DSM-IV, we have listed it as a psychological overexcitability. However, it is also included herein as a physiological overexcitability given the immunological processes at play that have been demonstrated in the scientific literature.”
The fact is that nobody yet understands the etiology of autism, nor has anyone established a physiological or psychological basis for it. The authors are making a huge assumption by classifying ASD as any kind of overexcitability. Is any condition found in the DSM-IV (why DSM-IV?) a “psychological overexcitability”? Lacking any clear operational definition of overexcitability from the authors, we must turn to Dabrowski’s meaning of the term. According to the authors, “[Dabrowski] found these overexcitabilities to be associated with personality development, and observed symptoms of slight neuroses among them as well, such as depression, mild anxiety, and tics.” It does not sound like Dabrowski was describing the kinds of pathology described in the DSM. In fact, he was describing something he saw as a driver of personal growth for gifted individuals. It is worrisome to see Dabrowski’s positive concept broadened and focused on pathology.
Is any condition that involves immunological processes a “physiological overexcitability”, and why only immunological processes? The authors note that chronic inflammation is associated with many medical conditions, but there is by no means a consensus regarding a causal relationship between inflammation and any of the conditions they considered to be related to overexcitability. Instead, the authors engage in some faulty reasoning: stress, autoimmune disorders, allergies, mood disorders, ASD, and so on are all correlated with inflammation. Stress is correlated to intellectual overexcitability. Intellectual overexcitability is correlated with rumination. Rumination is correlated with anxiety and depression. All of these things may be true, but correlation is not transitive. This means that “X correlates with Y” and “Y correlates with Z” does not imply that “X correlates with Z”.
The model the authors propose is based upon a flimsy chain of unjustified reasoning. Beyond that, its most glaring deficiency is its lack of consideration of the impact of a person’s social environment on their stress level and mental health. The authors include environmental triggers in their model, but according to the diagram, they oddly do not consider the environment to be a contributor to worry, rumination, or psychological overexcitabilities. We don’t need to postulate a “hyper body” model to explain the stress that gifted people experience throughout their lifetimes as a result of their intellectual mismatch with the people around them. The simplest explanation is that gifted people have the same basic emotional needs as everyone else, but that their intellectual differences make it more difficult for them to get those needs met.
Ask yourself whether you would expect a study to find such a strong correlation between high IQ and every condition the questionnaire asked about. Doesn’t that raise some red flags? If true, how do we explain the studies which did not find a correlation between high IQ and the health problems? Or the studies that found a link between intelligence and greater life expectancy? Why would we expect to find no link between increasing IQ and increasing health problems and then suddenly see a large effect at the cutoff point for Mensa membership? The reported rates from their questionnaire are so much higher than expected that we should look very closely at these results. In particular, we need data on the rates reported by this group for conditions that are not supposedly associated with overexcitabilities. If this result is not simply an artifact of the study methods, we should observe that Mensans do not report higher levels of diagnosis for conditions not associated with overexcitabilities, and we should be very careful to define which conditions are associated with overexcitabilities to avoid simply reclassifying everything that Mensans report as being associated with overexcitabilities! We should look very closely at any study which proposes a complex and loosely defined conceptual model which appears to explain everything.
It is important to find out if people with high IQ are more susceptible to certain health conditions or have physiological or psychological differences that could affect treatment outcomes. But we need studies that actually define overexcitabilities in measurable ways. We need studies that are well-controlled. We need studies with better samples. We need studies that include actual physiological and psychometric measures. This study did not do any of those things.