Hill on Strength:  Hill observed that when the correlation between two sets of observable data is especially strong, it is compelling to consider the possibility of a causal relationship based on the association’s strength alone.  As an example of this, Hill pointed out that mortality rates from heavy tobacco smokers was twenty to thirty times that of nonsmokers.  Before even hypothesizing biological reasons for smoking causing cancer, a researcher could not ignore such a stark difference between heavy smokers and nonsmokers and their lung cancer mortality rates, and would be compelled to investigate further, rightly supposing a meaningful association was present.

Cummings (2010) and consistency and plausibility:  Rather than investigate the possible causes of this well observed correlation, researchers are often incentivized to spend their effort (and funding) uncovering new and novel links between associations that may to be equally strong.  When encountering such studies we should look to other criteria in the Hill arsenal, such as his principle of consistency.  Cummings’ (2010) research on the connection between sunshine and the seasonality of human birth rates drew on data from nine distinct locations around the globe, with a average study period of nearly 17 years.  When Hill suggested that consistency was key to assert causal connections, he noted the value of data, “repeatedly observed by different persons, in different places, circumstances and times”.  The Cummings study did just that, and sought what data was available from locations as disparate as Vietnam, Finland, and South Africa, among others.  The study was looking for indications that exposure to environmental light intensity (ELI) in the 1-2 months preceding conception increased the rates of children born 9-10 months later.  The breadth of Cummings' data and rigorous approach to isolating actual light exposure (and not being confounded by dark, rainy skies on long summer days, for example) allowed Cummings to conclude with some confidence that there are “significant positive correlations between ELI and birth seasonality in six culturally, geographically diverse environments.”  Hill also stressed plausibility as a criteria for studying causal connection, and the Cummings study addressed that also, citing the human body’s increase in the production of reproductive hormones in response to sunlight exposure.  He further rested his theory on biological, not cultural, origins by noting how there is similar evidence of birth seasonality among baboons and chimpanzees.

Verma & Verma (2015) and temporality:  Verma (2015) summarizes a body of scientific literature about links between myopia and high intelligence.  The study acknowledges the myriad challenges espousers of this relationship must overcome—not the least of which is a universally accepted definition of “intelligence”.  Anyone hearing this assertion for the first time quickly tries to imagine the alleged scientific basis for the claim.  Verma summarizes what various studies purport are the medical links between myopia and intelligence, such as the association between ocular and brain development, and the likelihood that myopes read a lot and are thus better prepared for “intelligence tests” that reward those examinees who read a lot.  The latter claim is really behavioral, not biological, and this study tries to isolate stronger medical reasons why myopia develops in intelligent individuals, but struggles to find objective, observable data that may be independent of behavior.

    Verma recounts another hypothesis of the myopia-intelligence correlation, dubbed the “eye-brain gene."  The idea requires considerable mental yoga to accept as plausible and is not worth regurgitating; however, there is a more fundamental Hill criteria that the various studies on this topic fail to satisfy, and that is the principle of temporality.  Hill observes that in strong cases for causation, the data indicates what condition leads to what outcome.  In other words, it should be clear what is the horse, and what is the cart.  The literature summarized by Verma is woefully deficient in this respect, as while those target studies can point to clear associations between the two conditions, researchers cannot articulate whether myopia is the product of high intelligence, or somehow causes it.  If, for example, it is true that myopia and high intelligence are the result of some shared gene, how can one refer to their relationship as “causal”?

Rogan & Gladen (1993) and biological gradient:  Another novel criteria suggested by Hill as useful in determining causation is the principle of biological gradient.  This notes that a characteristic alleged to cause an outcome should do so on a measurable basis along a dose-response curve.  For example, if exposure to some airborne pollutant causes a respiratory disorder, than exposure to a high dose of that pollutant should cause a severe case of that disorder.  It is interesting to consider this criteria in the context of studies asserting links between infant breastfeeding and cognitive development.  The Rogan (1993) study does just that, and focuses particularly on how longer breastfeeding periods may produce relatively higher cognitive development than would shorter periods.  (Fig. 2)