In a previous blog post, I talked about the definition of “fact” in a scientific context, and discussed how facts differ from hypotheses and theories. The latter two terms also are well worth looking at in more detail because they are used differently by scientists and the general public, which can cause confusion when scientists talk about their work.
In common parlance, theory is often used to refer to something that is rather speculative. Because of this, it sometimes takes on a negative tone (for example, when creationists refer to evolution as “just a theory”). This definition strongly contrasts with the definition of theory as it is used in science: a theory is a carefully thought-out explanation for observations of the natural world that has been constructed using the scientific method, and which brings together many facts and hypotheses. The term hypothesis is good to define in this discussion as well: a hypothesis is an idea that we can test with further scientific observations.
With these definitions in mind, a simplified version of the scientific process would be as follows. A scientist makes an observation of a natural phenomenon. She then devises a hypothesis about the explanation of the phenomenon, and she designs an experiment and/or collects additional data to test the hypothesis. If the test falsifies the hypothesis (i.e., shows that it is incorrect), she will have to develop a new hypothesis and test that. If the hypothesis is corroborated (i.e., not falsified) by the test, the scientist will retain it. If it survives additional scrutiny, she may eventually try to incorporate it into a larger theory that helps to explain her observed phenomenon and relate it to other phenomena.
That's all fairly abstract, so let's look at a concrete example involving some recent research I undertook with a group of collaborators. The theory of evolution states that the process of natural selection should work to optimize the function of an organism's parts if the changes increase the chances of the organism successfully producing offspring and the changes are heritable (i.e., can be passed down from generation to generation).
Consider a turtle's shell. Turtles with stronger shells will be more likely to survive encounters with predators, and thus will be more likely to successfully produce offspring. Over time, natural selection will weed out turtles with weaker shells (i.e., those individuals will produce fewer offspring), resulting in a species that has relatively strong shells.
But what happens when there are multiple selective pressures at work? We might hypothesize that turtles that spend most of their time in water face a trade-off between having a strong shell and one that is streamlined (making them more efficient swimmers), whereas streamlining would be less important to turtles on land, allowing them to evolve stronger shells even if they aren’t very streamlined.
My collaborators and I tested this hypothesis in the following way. First, we digitized the shell shapes of a number of turtle species, some of which are aquatic and others of which spend most of their time on land. We then used an engineering technique called finite element analysis (or FEA) to examine the strength of the differently shaped shells when they were subjected to a crushing force, similar to a predator's bite. To measure how streamlined the shells are, we measured their cross-sectional areas, with the idea that a domed shell with a tall cross-section is less streamlined than a flattened shell with a low cross-section. Finally, we used a mathematical model of natural selection to estimate how much of a trade-off between strength and streamlining each species was forced to make, given the observed shape and strength of its shell.
Our results corroborated our hypothesis that aquatic turtles are forced to make more of a trade-off between strength and streamlining than turtles that live on land. In general, the shell shapes of our aquatic turtles were more streamlined but weaker than those of our land turtles, and our mathematical model of natural selection indicated that selection for streamlining was acting more strongly on the aquatic species.
As with any idea in science, our results are open to further testing. For example, other researchers might develop a better model of natural selection that shows that our model was overly simplistic. Or they might collect data from more turtle species that shows that our results were based on a false pattern stemming from sampling too few species (we considered 47 species in our dataset, about 14% of living turtle species). For now, though, our results can be added as a piece of evidence that is consistent with the predictions of the large explanatory theory of evolution.
If you would like to learn more about this research, the scientific paper describing the work can be found in the Journal of Vertebrate Paleontology. You can see some of the turtle specimens that we used in this research in The Field Museum's exhibition Specimens: Unlocking the Secrets of Life, open through January 7, 2018.