What is realism and anti-realism? - Chapter 4

Realism holds that the physical world exists independently of human thought and perception. An opposing side, idealism, thinks that the physical world is in some way dependent on conscious human activity. It is a long-term debate in philosophy, belonging to the area of metaphysics. A similar debate, that is really about science, and what is of concern in this text, is about the debate between scientific realism and anti-realism/instrumentalism. 

What is scientific realism and anti-realism?

Scientific realists think that science aims to have a true description of the world, and that it often succeeds in this. Anti-realists think that the aim of science is to find empirically adequate theories, meaning that it correctly predicts the results of experimenting and observation. For them it's truly about empirical adequacy and not truth. Physicists may talk about unobservable entities, but they are basically convenient fictions, introduced to help predict observable things. Whereas realists see the theorizing about unobservable things as true attempts to describe the world. Anti-realists are often called instrumentalists, because they see theories as instruments to help predict observable things, instead of as attempts to understand reality. And anti-realists often come from the viewpoints of empiricism. They also think that the models often used in science support their view, as they contain assumptions known to be false (but otherwise it would be too complex), and so they could never truly explain the world, but they could succeed in empirical adequacy. 

Realists do not agree that scientific knowledge is limited by what we can experience. They feel there is every reason to believe our best scientific theories, and they talk about unobservable entities. Realists also think that science goes for approximate truth, when exact truth is difficult, and models can help greatly with this.

What is the "no miracles" argument?

The "no miracles" argument is basically about the fact that there are theories which focus on unobservable entities, that have great empirical success. It says it would be a crazy coincidence if a theory makes accurate predictions about things that do not even exist. And if these entities are just convenient fictions like the anti-realists maintain, why do these successful theories have successful practical implications too? Is it a miracle? So this is an argument for scientific realists. It is a plausability argument: an inference to the best explanation. This means it might not prove that realism is right, but the best explanation for the fact that these theories have empirical success, might be that the theories are true, which supports the realism's view.

One anti-realist response to the "no miracles" argument, is that there are many theories that seemed empirically successful but later turned out false. So the inference from empirical success to theoretical truth is not always accurate, say the anti-realists. 

Realists have responded to this again, by modifying the argument in two ways. The first adjustment is that they claim that a theory's empirical success may not show complete and full truth, but at least it shows approximate truth. And the second adjustment refines the notion of empirical success. When saying empirical success is not just about fitting known data, but also about predicting new observations, it's less easy to find examples of empirically successful theories that turned out to be false. But still, there are some, so the "no miracles" argument is still kind of challenged.

What is the distinction between observable and unobservable?

It seems easy, but it actually is philosophically problematic. One of the main arguments for realism is that it's not possible to make the distinction in a principled way. This is against anti-realism because since they claim that science cannot give us knowledge of unobservable things, they presume there is a distinction. 

One of the issues is the relation between observation and detection. Some entities are not really observable, but can be detected using apparatus. But this is not always clear to distinguish. So when is something observed and when is it detected? It seems to be a smooth continuum. Realists use this in their advance because it speaks against anti-realism, just like the fact that scientists often talk about 'observing' when using detectors.

Anti-realists response with saying okay, maybe it is a vague term, but a vague term can still be used well. There may be clear-cut cases and less clear-cut cases, depending on the kind of entity, but according to anti-realists, like Bas van Fraassen who contradicts realist's Maxwell's claims, observable or not observable is still something that can be distinguished. 

What is the underdetermination argument?

Anti-realists also need an argument for why knowledge of unobservable things is impossible. One argument focuses on the relation between empirical data and theories. The underdeterimination argument, by anti-realists, says that scientific theories which are about unobservable entities are underdetermined by the empirical data, because there can be other theories or explanations for the outcomes, as only observable phenomena can be tested and the unobservable things can not. The idea of underdetermination leads to the anti-realist conclusion that agnosticism is the attitude to take towards theories about unobservable things.

Realists respond by saying that there always may be more possible explanations, but this does not mean the explanations are equally good. So one theory can always come up as the best one. And it seems underdetermination is not a true problem in actual scientific practice, as it's often difficult to find just one fitting theory, let alone multiple.

However, anti-realists feel that these philosophical worries are still genuine and important even if the practical implications are not really there. They also feel like there is never a good way of finding out which theory is the best one when there are multiple. 

Realists have another comeback though, as they say anti-realists apply the underdetermination argument selectively. They think that if it's applied accurately, even knowledge of much of the observable world can not remain, because: what is observable, is not always actually truly observed, and so there can be other explanations. And so, if we apply this argument consistently, there's almost no scientific knowledge anymore, as much science is about unobserved things. It seems underdetermination can not really be a barrier to knowledge. Some even say the underdetermination argument is just Hume's problem of induction put another way.