In most scientific disciplines, new knowledge is gained by following a formalism known as the Scientific Method. It exists in a variety of different forms but can be distilled into four major parts: (1) Characterisation of a problem and formulation of a research question, (2) construction of a hypothesis and derivation of predictions, (3) experimental testing of these predictions, and (4) re-evaluation of the hypothesis in the light of the experimental results [1].

To illustrate this rather abstract formalism, we can imagine the following scenario: We would like to have some toast for breakfast. However, we notice that our toaster is not working, and we aim to investigate the reason for that. Therefore, we formulate our research question: ‘Why is the toaster not working?’ and construct the hypothesis ‘The toaster is broken.’. From this, we can derive the prediction that the toaster should continue not working when we plug it into another outlet, which can easily be tested experimentally. Operating the toaster from a different outlet can have one of two results, which will decide how we will re-evaluate our hypothesis. If the toaster now works, we will have to change our hypothesis to ‘The first outlet was broken.’, while if the toaster still does not work, we can count this as evidence that our hypothesis is correct [2]. To truly act scientifically, we would then try to replicate our findings, by continuing to test the toaster on a number of other outlets, ask other scientists to reproduce our results, and have experts evaluate every aspect of our reasoning in a process called peer-review.

It is important to note that even if we succeed in all three steps above, we can not prove that the toaster is broken; we can easily imagine scenarios where the toaster is not working on different outlets, but is also not broken, e.g., a power outage. In fact, the Scientific Method never delivers proof for a hypothesis, but instead offers a way to test and potentially disprove it [1].