The cognitive errors outlined so far (in the blog posts Thinking Errors and Placebos) are common, they are particularly hard to notice, and they are hard to correct. The positive effects mentioned in the comments are undoubtedly present, but I will show in today’s blog that many effects are nevertheless mostly negative in nature.

This is particularly evident in science, which strives for neutrality, independence of results from individuals, and intersubjective verification, to name but a few. But if it is the case that cognitive errors occur systematically, then these ideals are, in principle, difficult to fulfil. That and how cognitive strengths and weaknesses influence scientific activity is a hitherto neglected area in the philosophy of science. Historical or social factors of influence are much better studied, but in my view need to be complemented by the cognitive dimension.

To demonstrate this thesis, I will use the framing effect in medicine as a case study for the sake of brevity (warning, a bit of self-promotion: Those interested in further case studies are referred to my dissertation “The Blind Spot” or its popular version “Pitfalls” by Beck.).

This somewhat old effect (the classic study is by Tversky and Kahneman 1981, Science 211, p. 453) states that the formulation (the frame) plays an important role in logically equivalent decision tasks. People tend towards safety in gain formulations and towards risk in loss formulations. This effect initially only describes behaviour in win/loss formulations, but it affects all risk considerations, e.g. in medicine.

McNeil et al. 1982 (NEJM, 306, p. 1259) studied 238 patients, 491 medical students and 424 radiologists. Two treatments for lung cancer (surgery and radiation) are presented. In each case, a subject receives only the positive formulation (90% survive) or the negative formulation (10% die), as well as the subsequent life expectancy. Both treatments are presented in both formulations. Variations and controls result in 24 groups, all of which show a large frame effect. The astonishing result:

On average, radiotherapy was preferred to surgery 42% of the time in the mortality frame and 25% of the time in the survival frame. (McNeil et al. 1982, p. 1261)

In other words, generalised: Who tells your doctor what, and in what form, about therapies and drugs, has a decisive influence on what he or she prescribes or recommends to you, partly independently of what is de facto the case.
De facto would mean that the treating doctor knows the original trials, has compared them and treats only on the basis of these figures. Sadly, this is not the case. Although so-called “evidence-based medicine” has caused a stir recently, it is far from being a standard. But it should be, because human fallibility, with all its flaws, makes such an approach all the more important.