Blog

 

I am not afflicted by a belief in a god. I am unable to simultaneously champion the scientific method and an untestable hypothesis about the existence of a deity. I came to this realisation as a young man, and at that point I stopped bet-hedging on an eternal life in paradise and abandoned beliefs in things for which there is no evidence. I do not fully understand how I made this decision. It was an emergent property of my brain, a complex network of neurons, which had been wired, in part, by my experiences up to the point I made the decision. I’d love science to be able to explain how I made the decision I did, but currently it can’t. Perhaps the work I am just starting to do on understanding complex ecosystems may one day help us scientists explain the emergent thoughts and decisions of our brains. In this short essay I want to explain the challenge I am addressing in a new grant that is just starting. Although this work is on ecosystems, it could one day prove useful in understanding other sorts of complex systems, including brains.

 

The universe has a set of rules, or laws, that determine its dynamics. These rules include the ways that gravity, electromagnetism, and the strong and weak nuclear forces determine how matter interacts, and how evolution modifies form, function and the distribution of life. Without rules, the universe would be completely unpredictable. In a ruleless universe planets would not orbit around stars, or moons around planets; we could never predict what molecules would form when hydrogen and oxygen atoms interacted; and we could not predict how chains of amino acids would fold to form proteins, the workhorses of life, as each time they folded a different outcome would be observed. No sense could be made of a ruleless universe, and any sense of order would be short-lived and entirely down to luck.  

 

Science is all about trying to identify and describe, ideally with equations, the rules of our universe. Given this, a key of aim of science would be to explain how we came to exist. Science has come a long way in answering this question, and there are various amazing breakthroughs that have contributed to our understanding of the universe. The standard model is a fabulous example of an amazing scientific insight. It is an equation that describes how fundamental particles – quarks, electrons, bosons and neutrinos – interact via the weak and strong nuclear forces and electromagnetism. The standard model was constructed from numerous observations and experiments on the behaviour of particles, and it predicted the existence of a particle called the Higgs boson. Physicists and engineers designed and built the Large Hadron Collider to search for the previously unseen fundamental particle, and they found it, just as the standard model predicted. 

 

Despite such astonishing successes, science does not have all the answers, particularly when it comes to complex systems with multiple interacting parts. Science is at its most powerful when used to investigate how a small number of things influence one another, rather than more complex cases where hundreds of things interact. There are two reasons for this. 

 

First, imagine you wanted to know how the growth of a species of plant is influenced by temperature and watering regime. You might have three temperature treatments – say 15C, 20C and 25C – and two watering regimes – drenching with a fixed volume of water every day, and every three days. Each day you measure the height of each plant and count the number of leaves. It takes two minutes to record data from each plant. There are six combinations of treatment. Regular and infrequent drenching at each of the three different temperatures. Measuring six plants will take 12 minutes a day.  Not very long you might think. However, all experiments require replication. This is because different plant genotypes may grow at different rates, and chance events may mean the odd plant might become infected with a disease or be attacked by herbivorous insects. You might decide it is sensible to have 10 plants in each temperature-watering combination, making a total of 60 plants. Two hours of your day are now taken up just measuring plants.

 

In the natural world there are very many factors that can influence plant growth beyond temperature and water availability: neighbouring plants, soil type, genetics, light, herbivores, and disease, to name but a few. You might want to expand your experiment to include more treatments, but each time you do this, you will increase the number plants you need. In addition, you might decide that two levels of a treatment is not sufficient and you should really have three, or even four. Your experimental design very soon becomes unmanageable. You will have thousands of plants, and you’ll need a team of helpers to carry out all the measurements. 

 

The second reason that the scientific method faces challenges with complex systems is that models of them can be hard to understand. Models that contain many factors are often impossible to analyse to gain analytical understanding. The more variables a model contains, the harder it is to understand why it produces the predictions it does. We are often able to build simulations of complex systems, and these models can make accurate predictions and do astonishing things (think ChatGPT), but we often do not know why. In some cases, we end up with a system and a model, neither of which we understand.

 

In recent years it has become increasingly easier to collect large amounts of data from complex systems. In my field, ecology and evolution, we can fit high-tech tags to animals to track where they are and what they are doing, and this can generate significant amounts of data. Coupled with environmental data from satellites and sensors, and individual genetic data, we can build data sets containing tera- or even peta- bytes of data. The data can then be analysed with statistical models, which can contain very many terms. These statistical analyses are a useful way of identifying which factors appear to influence one another, but they tell us little about the dynamics. They provide but a small step to real scientific understanding.

 

Along with a host of collaborators, and three post-docs, I am just starting up a new project that will use statistical analyses of complex ecological systems to build complicated models of the systems. These models will hopefully reasonably accurately generate the emergent dynamics of the systems that interest us. We will then use a new theoretical approach I have been developing to simplify these complex models. At each simplification, we will examine whether the emergent dynamic we are interested in is still predicted. If it is, then we will consider the model simplification as acceptable. If the emergent dynamic disappears following a simplification, we will have identified a key driver of it. We will put that driver back into our model and continue to simplify the model by removing other variables. Eventually we hope to generate simple models that provide new, general, insight. 

 

As with any approach, the one I am developing builds on existing methods, but our hope is will bring a new tool to the toolbox of the scientific method. If it doesn’t, we will have at least discovered an approach that doesn’t work, despite its promise. Although I would be disappointed our approach had failed, we would still have learned something useful. 

 

The complex systems we will study are the ecosystems of Yellowstone National Park, the streams of Northern Trinidad, and ecological communities on mainland Australia and her offshore islands. Although our approach will be trialled on complex ecosystems, each of us involved in the project is ourselves a complex system. Our thoughts and beliefs arise in our brains, shaped by our experiences and the ways neurons in our brains form networks.

 

Who knows, perhaps in time as scientists start to make in-roads into better understanding complex systems that generate fascinating emergent properties, we may start to understand why some people believe in a god and others do not. Belief in a deity is an emergent property of our brains, so it should be as amenable to study as adaptive evolution, an emergent property of the natural ecosystems. Time will tell.