In his essay “Arguments Concerning Scientific Realism”, philosopher Bas van Fraassen makes an analogy of the success of scientific theory to Darwinism: the most successful scientific theory that can capture the regularities of nature always wins the competition in the jungle of scientific theories. This analogy is made in response to the demand for an explanation of why our scientific theories can make successful predictions from the scientific realist’s point of view. Scientific realists posit that science is aimed at truth, and truth is the only explanation for the success of science. The successful predictions of scientific theories would be a miracle if science does not provide a true story of the world.

 I will first articulate my view that the aim of science is to provide the best model for positing the explanatory power of observable (p.s. by observable, what I mean is detectable by any form of instruments as grounded by the current knowledge of physics. The limits of observation are not due to technological limits, but rather they are based on the limits of obtaining information. Our understanding of these limits is provided by our current knowledge of physics- which could later be proven wrong in principle.) phenomena. This suggestion is a modification of Van Frasseen’s constructive empiricism that science aims at empirical adequacy. Since aiming at empirical adequacy implies the acceptance of a theory involves a belief that it is empirically adequate, which would easily be weaponized by scientific realists. I will then incorporate van Fraassen’s Darwinism analogy from an anti-realist’s perspective to explain the success of science. Finally, I will provide a realist’s objection to a Darwinism explanation as well as a possible solution from a compatible perspective.
Van Fraassen provides a minimum statement of what scientific realists posit: (1) Scientific realism is the position that scientific theory construction aims to give us a true story of what the world is like, and (2) that acceptance of a scientific theory involves the belief that it is true. 
To provide a convincing argument that denies both (1) and (2), I will start by examining how most scientific hypotheses and theories are made in relationship to the device of inductive reasoning. Inductive reasoning begins with us extracting information from our observations, for example, our experience of the sunrise. We have seen the sunrise and have seen the sun rise again and again. Based on that observation, we make predictions that “the sun will rise again tomorrow”. When the sun does rise the next day, we are assured that this prediction is reliable and reasonable and conclude that “it will rise every day henceforth”. We take this connection into our system of knowledge as a piece of fact that describes our reality. Therefore, on any given day, we can reliably conclude that the sun will rise tomorrow without actually looking at the sunrise. This is how we can use observed facts to conclude unobserved facts. 
However, since humans have only existed for a relatively short period of time, within the time of our past existence, even within the entire lifespan of the human race, we have only experienced sunrises every day. Our belief that the sun will rise tomorrow is just a relatively convincing prediction concluded by a relatively small number of samples, considering that humans have only existed for a fraction of the time our universe has existed. Although we have high confidence that the sun will rise again the next day and consider it to be a fact, we do not know for sure.
The source of our high confidence stems from that we take the predictions made by inductive reasoning as known information into our consideration. However, this method has its limits. It cannot provide credence for something that is in principle unprovable by our future observations. The virtue of inductive reasoning is not intrinsic in the method itself. It has certain virtues that were granted by the stable relationship between our observations and predictions. We can think about it as a contract to actualize the foreseeing value: we could take the predicted outcome as a piece of fact (with high confidence) and prove it later based on future observations. Therefore, for the proposition that is in principle unprovable, or has no causal relationship to us, the virtue of this method loses its legitimacy accordingly. 
In the realm of scientific methodology, we construct the model that can best explain the observed facts, and confirm or disprove them by experimentation. The minimum requirement to accept a scientific model is to match the model predictions with the observed facts from the past (at least most of them since there might be errors). We conducted experiments on these models, and those which had not yet been disproved, survived. We also conclude from our experience that certain virtues are in favor of the theories that hold the most explanatory power so these virtues become one of the indications for choosing the best theory. 
As Van Frassen points out: ”if a scientist accepts a theory, he thereby involves himself in a certain sort of research programme.” By far, I use “we” to refer to all human beings who share the communal scientific knowledge, but the way how a theory is actually accepted is based on certain standards from the scientific community. The acceptance of such theories is based on 1. Matching with the empirical evidence from the past. 2. Matching with the experimental data. 3. Indication by certain virtues (Including aesthetic virtues such as simplicity, unification and other virtues such as inconsistent with the past theory. The virtue of testability should be a prerequisite for our acceptance of the theory instead of a preference. History of science shows that the obsession of mathematical aesthetic virtue often leads to great discovery, but without the limitation of observability and testability, it always leads to the conclusion that is not in alignment with our physical reality.) which is not required. This means, based on how the theory is accepted, the acceptance of scientific theory does not guarantee it will continually make successful predictions. So the only belief involved in this process is the belief that a successful theory could potentially make accurate predictions in the future.
This is how we accept certain scientific theories, which do not involve the belief that they are the truth-  a true story that describes both the observable and unobservable reality. From the mechanism of how scientific theory is accepted, we conclude that the acceptance of a theory involves the belief that such theory can continually make successful predictions, instead of the belief that it is true or empirically adequate. (NB: it is worth pointing out that if the goal of science is not the truth, but rather the best model that can provide the most accurate predictions, it would automatically allow the acceptance of unobservable entities and structures such as string theory into the scientific model. Our acceptance is based on our confidence that these models that involve the unobservable can help us to make a more precise prediction about the observable world.) 
Van Fraassen’s analogy of Darwinism, therefore, seems convincing in explaining why our best scientific theory can make such successful predictions- the goal of making the successful prediction is embedded into the very method of how it is being chosen. Since the scientific theory that does not make successful predictions has already been replaced, it should not be a surprise that our best scientific theories are those which make accurate predictions. 
A scientific realist, however, would not be satisfied with this kind of explanation and would argue that it is not a real explanation of why the best theory is making successful predictions. The scientific realist might argue that this explanation is merely descriptive of why “we” choose such a theory as the best theory, and not an accurate reason for why the theory is successful. If we take one of the interpretations from Leplin about the “No Miracle Argument”, one can argue that this Darwinism view does not provide any explanation for why such theory can make “novel predictions”, such as why the theory can predict certain cosmic regularities with such precision. As Leplin stated, “The explanation must appeal to some property of the theory, something distinctive in its content that enables it reliably to forecast the unfamiliar and unexpected. Unless this content is interpreted realistically, the theory’s novel success appears purely accidental.” From a scientific realist’s point of view, there must be an explanation of why there is such a stable connection between the predicted success and the best scientific theories, and the explanation can only be that the science is capturing some property that leads to the truth (It is also worth pointing out the two uses of “truth” in the realist’s argument. In most arguments, the use of truth simply means the sentence is correct or that it entails the correct picture of the world. But specifically in the context of Leplin’s explanation, the use of truth means “the law of nature” that is governing how the world is functioning.) As a result, the explanation demanded here is why there is a stable connection in which our future observation can be predicted, and why our best scientific theory can capture that. 
The denial of this demand, as provided by van Fraassen, that  “There cannot be a requirement upon science to provide a theoretical elimination of coincidences, or accidental correlations in general, for that does not even make sense.” seems like does not provide a satisfying explanation from the realist's perspective.
From my point of view, our intuition of “the fundamental law of nature” comes from our sense of mathematical knowledge and the strong correlation between our best scientific theories and mathematical principles. Our best theoretical model that posits the explanatory and predictive power of observable phenomena implies that there are some stable relationships between our observation of reality and mathematical principles. From a realist’s perspective, the observable datas suggests such stable relationships do in fact exist, and that the purpose of scientific theories is to discover the relationship between our world and these mathematical principles (in parts based on the virtue of inductive reasoning). In other words, the realist would consider the product of science, i.e. “theories using mathematical knowledge in describing reality”, as “the fundamental law of nature” that govern our reality.
However, if we compare this aim and the aim that I mentioned before:
(1)  Science aims to discover (p.s. the action of ascertaining the stable relationships of the world could be seen as discovery even though our scientific methods are human invention) the stable relationships that manifest in our world with respect to mathematical principles.
(2) Science aims to provide the best model (p.s. It is important to distinguish between a “model” and a “law.” The distinction between the two terms raises issues concerning the legitimacy of how scientific facts should be acknowledged in our society.) positing the explanatory power of the observable reality.
We can see that these two aims are not incompatible with each other- scientists can believe science is aimed at both (1) and (2), but it has to be based on the reasonable acknowledgment of the limitation of the scientific methods. In other words, what the realist consider as "truth", is simply saying reality can be predicted by theories, supported by the past observations. And that what the anti-realist saying about non-truth, is simply denying the perfection of the scientific devices, pointing out its limitations, that we can not assign a reasonable level of confidence on our knowledge about what is in principle unobservable, and that our successful theories may in principle to turn out to be wrong any day in the future. 
Fraassen, B. van. (n.d.). Arguments Concerning Scientific Realism.
Bird, A. (1998). In Realism (pp. 92–94). essay.
 Leplin, J. (n.d.). A Theory's Predictive Success can Warrant Belief in the Unobservables Entities it Postulates (p. 127).
Back to Top