The biological sciences are experiencing rapid change, including omics-scale experimental; observational and computational tools; and the migration from a reductionists to a holistic approach (systems biology). However, one thing that remains constant is that we still seek simple explanations of complex systems.
We continue to operate under the assumption that the final product of our work – the understanding of the system behavior – is something our brains can grasp, and we can convey in a paper or discussion. Our inability to make predictions for real systems like lake ecology or human health lays bare the limitation of this approach.
Other people, from car mechanics to climate scientists, have learned to delegate understanding to computers. If the goal is to make predictions of complex system behavior, biologists will have to follow suite.
Why we love simplicity
Simple solutions are elegant and sexy. Just about everyone subscribes to Occam’s Razor and Einstein’s famous notion that explanations or models “should be as simple as possible”. Ecologists are fond of Liebig’s Law of the Minimum, which says that the nutrient in limited supply will control the productivity.
We love questions directed towards simple answers, preferable of the yes/no type.
We love questions directed towards simple answers, preferable of the yes/no type. Is heart disease caused by bad cholesterol or the ratio of bad to good cholesterol? Is this lake nitrogen or phosphorus limited? High-impact papers summarize the main conclusion right in the title, like “Even modest trans fat consumption leads to obesity in teenagers.”
Ironically, even in the field of complex systems science, much of the focus seems to be on simple rules underlying the behavior of systems.
Not so simple
What if there is no simple solution to biological systems? Maybe they are too complex to be understood by a human. A couple of years ago, Toledo had to shut down its drinking water plant because of a massive toxic cyanobacteria bloom in Lake Erie. This is puzzling, because we spent billions on reducing phosphorus input to the lake, which according to the simple model of lake eutrophication, is supposed to make things better.
The common paradigm is that caloric intake relates to obesity, which leads many people to opt for horribly-tasting diet soda. Unfortunately observations now show that this has the opposite effect – it actually leads to weight gain. Many alternative simple explanations are being debated (it’s the nitrogen, the artificial sweeteners make you hungry), but we are not considering the possibility that these system may be too complex for us to understand.
It is possible that there are simple solutions to biological systems, and that the scientific process of debating and testing different hypotheses will eventually lead us to find them. But if the solution is actually complex and all of our hypotheses are simple, then we will never arrive at the answer by this process. For the sake of basic biological science, keeping it simple may not be a problem. We can still generate good insights and educate the next generation of scientists.
Often we are faced with the need to make decisions, like how much phosphorus treatment to mandate in the Lake Erie watershed or which soda to drink. Here the obsession with simple will prevent us from developing the comprehensive understanding required to make predictions.
Delegating understanding to computers
How can we develop a predictive understanding of complex systems? The solution is to delegate the task of understanding the system behavior to computers – we do the same for other complex systems.
We have to accept that not all science is simple.
If my (wishful thinking) modern, high-end Mercedes car has a problem, does a mechanic still try to understand what is happening by doing a test drive and listening to the motor hum? Will he realize that there may be a problem with air-gasoline ratio and adjust the carburetor? No, he hooks it up to a computer and accepts and follows the instructions it spits out.
Climate scientists seem to accept that they cannot understand Earth’s climate and rely on model predictions, without the ability or attempt to provide simple solutions. All the various components of the system sort of interact to produce the system-level emergent behaviour.
Why do we refuse to do the same in the biological sciences? Where is my computer model that tells me what is happening with Lake Erie?
Getting to this point will require fundamental cultural changes. We have to accept that not all science is simple. Presently, complex solutions – even if correct – are utterly useless to academics. Papers with titles like “Model of Lake Erie can predict temporal and spatial patterns of cyanobacteria growth and toxin production – but reason remains unclear” would not be of interest to other scientists and could never be published in a respectable journal. Presently, that’s considered blasphemy. But such science would have a tremendous impact on lake ecology and human health, so we have to support it somehow.