Thought-provoking essay on cause and correlation in modern science
[Via Boing Boing]
Science is the best method we have for understanding the world. That doesn’t mean that everything scientists ever think they’ve figured out is correct. And it doesn’t mean that we’re doing science in the best way possible right now.
For a great illustration of this, I recommend reading Jonah Lehrer’s new piece in WIRED, about the problems we run into as we learn more about individual parts of complex systems and then assume that we understand the big picture of how those parts work together. A lot of scientific research, particularly in medicine, operates off assumptions like this and it can lead to big mistakes. Case in point: Back pain. In this excerpt, Lehrer explains how MRI technology that allowed doctors to get a better look at the spines of people with back pain led them to make inaccurate conclusions about what was causing the back pain.
Science has been very successful using reductionist approaches in biology. Isolate a single protein understand it and you know a lot.
But the belief that understanding each part by itself and then putting them all back together would also allow you to understand the whole – well, that is just plain wrong.
We pretty much have solved all the major medical problems due to a single cause, one where understanding a single component was valuable. Now we are left with a chaotic mixture of disease, syndromes and oddities that have no single cause but arise from a more generalized problem with the system.
Science is basically about creating models that usefully describe the world around us. Simplified models can be very useful, if they accurately describe the world.
However, complex systems often are not amenable to much simplification. Attempts to create simplified models only results in errors, in inaccurate descriptions of Nature.
But few of our institutions have been developed to deal with a system. Doctors and researchers specialize. They use an analytical approach and break things down.
We now need scientists and institutions that synthesize and pull things together. Until we accomplish this on a large scale, we will continue to confuse causes and correlations, to continue to let confirmation bias lead us astray, and continue to make very expensive mistake.