Sunday, October 10, 2010

Mechanistic Research: The Only True Answer to A Question

Image courtesy of Google Images.

Science is simply wonderful. Sometimes it is also wonderfully simple. Some of the most important scientific discoveries ever made have been elegantly simple and, after the fact it is hard to understand how the discovery was missed up to that point. However, science can also be wonderfully complex. Anyone who has studied advanced organic chemistry has probably wondered at its complexity. The same could be said for much of the physiological processes that occur in biological systems, particularly advanced eukaryotic ones such as humans. I think we do not celebrate science enough nor give scientists enough recognition for the goodness with which they have illuminated the world. Sadly, there is also much poor science (or more accurately pseudo-science, for any science that is poor is not technically science) out there as well.

In my opinion, very good science seeks not only to make observations in our world, but to explain the mechanism behind the observations. It is astounding how much money and time is spent on non-mechanistic research. Just think for a moment of modern obesity-related research. Literally billions of tax dollars are spent each year on research that describes the increasing rate of the obesity epidemic in Western societies. Yet much of this type of research is not mechanistic. It makes no attempt to learn why western populations are more obese than their parents’ generations. In the extreme, one could opine that non-mechanistic research is worthless in terms of true scientific value. Think of an example of research into the health effects of smoking. Nowadays we all know that smoking is bad for you and we take that knowledge for granted. But think back 50 or 60 years and imagine designing research to determine the effects of smoking on health. The easiest type of research to conduct would be to examine the health of smokers and non-smokers and make a comparison: a cross-sectional study. Longitudinal research, involving the examination of the health of smokers over a long period of time, might be more valid because it may remove other confounding factors more effectively. Indeed, both of these types of research have been conducted on smokers in the past, and they both had some value in helping us arrive at the conclusion that smoking is bad for your health. Yet these types of research are not mechanistic: they tell us nothing about why smoking is bad for your health. Not until researchers examined the mechanisms behind airway and endothelial functions following exposure to the contents of cigarettes were we able to really understand why smoking may cause cancer, why it may cause heart disease, and why it may cause a host of other long-term health problems.

The main reason that non-mechanistic, or descriptive, research is less valid than mechanistic research is that conclusions drawn from observations alone can very often be faulty. A classic extreme example of this is the relationship between airliner crashes and fatalities by seat selection. Picture a descriptive researcher observing that more people die each year in plane crashes while sitting in window seats than sitting in aisle seats. With this information, the researcher may be tempted to draw a conclusion based on the observation such as: you are safer sitting in an aisle seat than a window seat when you travel by air. The researcher may even be tempted to go further and try to explain a reason for the observation such as: those who sit in the window seats are closer to the edge of the aircraft and may be exposed to greater impact forces during a crash. It is not hard to imagine people changing their behaviour based on this explanation of a true observation, but they would be completely misled if they did so. In this example, there is another factor that actually explains the observation which has been completely overlooked because the research was not mechanistic. The relationship between rates of death in airline crashes and seat selection is actually explained by the fact that more people choose window seats than aisle seats when traveling by air. When an airliner crashes, almost invariably everyone is killed. Put these two facts together and you find two things: 1) more people will die each year in window seats; 2) there is no relationship between your choice of seat and your chances of survival in a crash. So in fact the research, though accurate in describing the observations, has done a disservice to society in trying to explain the relationship without actually examining an appropriate mechanism for the explanation. These types of observational research reports are rampant in the popular news. Almost every day one can watch or read the news and find an example of health-related research that is urging you to take Vitamin C, get more (or less) sunlight exposure, avoid or use caffeine or alcohol, or any other number of behaviours that may help your health. There may be some truth to some of these recommendations. The problem is, without a mechanistic explanation, you don’t actually know whether they are true or merely based on a chance observational relationship.

This brings me to the topic of mechanistic science in relation to religious beliefs. Science is often purported to be in conflict with religion. Some say the two are incompatible. Others insist that the two can co-exist quite comfortably and do not necessarily contradict each other when it comes to explanations of the world around us. I fall firmly in the former group. Science is in direct conflict with religion for a very good reason: religion does not attempt to make mechanistic explanations; rather it is based purely on observations. I do not wish to get into a discussion of the evolution – creation “debate” (as if there were an actual debate on the topic!) at this point, but it is a relevant example. Occasionally, and recently, some creationists have attempted to strengthen their position by introducing the notion that there is some science in their theory. Some even go so far as to call it “creation science”. (That creationists are smart enough to figure out that being scientific lends validity to their point of view, yet continue to criticize and demean true science is at once incredibly humorous and deeply sad). Yet creationists almost never make any attempt to include mechanism in their “observations”. Almost invariably, a creationist, when asked how god created the world in six days comes up with some variation of “by magic” (i.e. no reasonable mechanism). This pattern, of course, extends far beyond creationism and into every realm of religious belief. Ask any believer what their explanation for how their god healed a relative and you are likely to initially get a simplistic answer: “by prayer”. Follow that up by asking for their explanation of the exact mechanism: “Yes, but how did your god actually remove the tumour from your aunt’s brain?”, and you will likely get a very blank stare. No thought is ever given to the mechanism. As I’ve pointed out, when there is no understanding or explanation of the mechanism, then the observation becomes significantly less valid, and the chances of misinterpreting the observations are increased. Mechanistic research, assuming it is done properly, rules out misinterpretation of observations. There is, of course, a very good reason that mechanistic research is not conducted by those attempting to show that god created the world, how he works in people’s lives, or even that god(s) exist: there is no mechanism to be uncovered. There are only observations with faulty conclusions. The mechanistic work done by physicists on our universe’s beginnings are infinitely more acceptable explanations for how we came to exist.