Ab absurdo, ad libertatem
Previous   |   Next       

Pessimism, Realism, and The Success of Science

Jun 17, 2009


The domain of science is the natural world. Consequently, an initial estimate of the purpose of scientific theory might be to say that its role is to interpret and describe nature. Science is defined as knowledge of facts and laws, gained and verified by exact observation and organized experiment, while theory is defined as a reasoned set of propositions, developed from and supported by established evidence, and serving as an explanation for natural phenomena; thus, scientific theory substantiates and explains scientific knowledge. Intuitively this description seems clear, but such a summarization requires justification; are we justified in the production and belief of theories?

A theory is produced by translating experimental observations into linguistic statements and mental abstractions. The question then becomes: can theoretical statements of human ideas act as accurate representations of the natural world? To evaluate the precision of science is to evaluate the products of our physical apparatus, and the problem may be restated thus: how should we regard scientific knowledge?

The investigation of scientific knowledge is the focus of the philosophical debate between realism and anti-realism. Both contend that science aims to provide a literally true story of what the world is like. The realist however states that the proper epistemic attitude towards our scientific theories is that of belief; we should believe that theories are true or approximately true representations of a mind-independent world for all observable and non-observable phenomena. The anti-realist purports that the proper epistemic attitude towards theory is one of acceptance; science does not extend beyond the ordering and correlation of phenomena, and theories cannot be true or even approximately true.

To better understand this difference, consider the scientific theories of Isaac Newton and Albert Einstein. For hundreds of years Newtonian mechanics was accepted as true (or approximately true), only to be shown false and replaced by Einstein's Theory of Relativity. This particular case displays a general characteristic of scientific theory that is relevant to the problem of evaluating science: since the advent of the scientific method, many theoretical explanations and entities have been proposed and later disconfirmed. Accepting the fallibility of science, how can we know that current scientific theories will not be proven false in the future? Why should we believe our current theories are true or even close to the truth? In short, why should we be scientific realists?

This paper will examine why we might plausibly approach science as realists, beginning with a description of the "no miracle argument" (NMA) for realism, followed by a discussion of the "pessimistic meta-induction" (PMI) as an anti-realist response to the NMA. The cogency of the PMI will be examined, and a potential realist solution to the PMI will be presented.

Realism by Default : The No Miracle Argument

The ideas surrounding the NMA were encapsulated in the realist defense of scientific truth presented by Hilary Putnam in 1975; "The positive argument for realism is that it is the only philosophy that does not make the success of science a miracle." In evaluating the truth or truth-likeness of science, the NMA states that the success of science itself demonstrates and confirms the veracity of science and scientific theory. This is an application of inference to the best explanation, where the best explanation for science being successful is that science is true or approximately true.

NMA uses the success of particular theories as premises in a meta-abductive argument regarding science as a whole:

  • Science is the collection of all scientific theories
  • The particular scientific theory, theory 1, is true or approximately true (that is, the particular scientific theory being discussed possesses explanatory ability and enables us to make predictions)
  • ...
  • The particular scientific theory, theory N, is true or approximately true

  • Therefore, all of science, being the collection of all scientific theories, is true or approximately true

Arguing from specific cases of success, the NMA makes an inference to the truth of a generalization covering them, and so the NMA promotes a realist justification for scientific truth by approaching the pursuit of science on the whole as an experiment, and appealing to the observable result of that experiment. Hence, the NMA makes an empirical claim about realism; that the observational success of science as an experiment supports the position of the scientific realist.

Again, this position seems intuitively clear, but again: is this argument justified?

An Anti-Realist Response to NMA : The Pessimistic Meta-Induction

The NMA concludes that the success of science may be inferred from the past success of theories, and so the NMA uses an abductive format in an attempt to show that abductive inference is credible. But is an abductive argument that appeals to particular cases of abduction acceptable? Setting aside the question of circularity, a more direct objection can be leveled against the NMA: many past theories have been proven false, so how can we believe that current theories will not be proven wrong in the future, and thus how can we believe that our current scientific knowledge is true or even close to true? Unlike the charge of circularity, this question is a direct response to the NMA, in that it is expressed in a format that mirrors the format of the NMA. The NMA is an optimistic meta-abduction on the success of past theories, and the question now asked is a pessimistic meta-induction on the failure of past theories. This pessimistic meta-induction is commonly known as the PMI, and was introduced by Larry Laudan in 1981 as an anti-realist reply to the realist NMA.

The goal of the PMI is to deny that the tentative empirical success of some scientific theories implies the truth of science as a whole, and runs thus: if a past theory provided empirically successful observations then it was described as true; however, if the past theory has been replaced, then the past theory must have been false, and the empirical success of the theory could not have conferred truth onto the theory.

The PMI generalizes the failure of science thus:

  • Science is the collection of all scientific theories
  • The particular scientific theory, theory 1, was held to be true, but was later proven false
  • ...
  • The particular scientific theory, theory M, was held to be true, but was later proven false

  • Therefore, it is not possible to know which scientific theories that are currently held to be true will later be proven false, and it is not possible to believe that all of science, being the collection of all scientific theories, is true or approximately true

Arguing from specific cases of failure, the PMI makes an inference to a generalization covering the instances, and advances the anti-realist case against the NMA by showing that the truth of a theory cannot be the best explanation for empirical success, because many theories that were empirically successful are now believed to be false.

The PMI has explicated the fact that it is not possible for the realist to support the claim that every component of science is true or approximately true, but does this provide a decisive refutation of realism?

The Realist Response to the PMI : Blocking Pessimism

The PMI has shown that it is unreasonable to believe that every scientific theory currently held to be true today will not be proven false in the future. What then would it be reasonable for the realist to believe? In building toward its conclusion, the PMI analyzes all of science, and thus all scientific theories. But is it reasonable to give all scientific theories equal weight when evaluating science as a whole?

Some theories are buttressed by a larger amount of support than others, and many theories belong to domains of inquiry that are immature or even suspect. Consider the evidence that exists to support atomic theory, and the number of successful theories in which atomism performs a crucial explanatory function, then contrast the viability of atomic theory with the unsubstantiated theory of water memory as put forth in the field of homeopathy. Both hypotheses claim to be scientific, but it is not logical to regard water memory as equal in worth or stature to atomism, however in examining science the PMI does precisely this, and undue regard is given to objectionable theories.

Still, the PMI may rightly be considered as having shown that it is unreasonable to believe that all of science is true or nearly true. However, in developing this conclusion the PMI has provided the realist with the basis for a counter-argument. Acting on suggestions found within the PMI, the realist response may be built upon a restatement of the PMI. Realism can move forward by replacing the unconditional requirement of absolute truth with the conditional requirement of approximate truth. In other words, realism can succeed if:

  • it does not position itself as an all or nothing proposition
  • it allows only genuinely scientific theories into the sphere of science
  • it can demonstrate that previously accepted theories which have proven to be false nonetheless contain true or approximately true elements that were essential to the empirical success of the theory, and that these elements were preserved in later theoretical formations, and thus that these elements provided a reasonable approximation of reality and are true or approximately true

Condition i is a restatement of the PMI; that it is not reasonable to believe that every statement in every scientific theory is true or approximately true. Condition ii states that all theories and disciplines, past or present, should not count equally in the evaluation of science and theory; when determining whether or not to permit a theory into the collection of theories designated as "science," each theory must be assessed individually, taking into consideration factors such as the credibility of the scientist(s) involved and the maturity of the domain of inquiry. Condition iii states that the job of realists is to distinguish the elements of past theories that were necessary in achieving empirical success, and to show that these elements continue to play an indispensable theoretical role.

To illustrate these conditions, let us examine a specific case in which a theory previously held to be true was eventually proven to be false, by considering the replacement of Newtonian mechanics with Einstein's Theory of Relativity. Speaking strictly, the scientific community at large accepts that Newtonian mechanics are false, and that Relativity is the best available approximation at truth. However, many scientists who apply Relativity are working to move beyond it. The scientific community supports attempts to expand and even replace the Theory of Relativity, precisely because Einstein's theory is not regarded as immutable. Thus, condition i is satisfied. Newton's work is widely recognized as one of the greatest scientific achievements of all time, thus condition ii is satisfied. Now, let us consider condition iii. When dealing with the interactions of objects at low speeds (relative to the speed of light) and in conditions of weak gravity, both theories provide acceptable explanations. However, Newtonian mechanics is confirmable only until we encounter instances in which the curvature of space-time becomes so large that this explanation fails. Here we find that Einstein's explanation remains confirmable in every instance, and we see how it is that Relativity came to subsume Newtonian mechanics; Einstein developed Relativity by combining successful elements from Newtonian mechanics, such as the gravitational constant, with successful elements from other theories, and was thus able to construct an inference about nature that was more extensive than Newton's. When integrating predicates from several scientific models, Einstein's theory eliminated insecure elements, and preserved the key features of the original models that offered enduring representations of nature. Hence, condition iii is satisfied.


How should we regard science? The anti-realist observes that many past theories have been acknowledged as acceptable explanations of reality and later disconfirmed, and argues that because there is no procedure for determining which theories will be proven false in the future it cannot be reasonable to believe that science is true or almost true. When faced with the fallibility of scientific theories, why should we be scientific realists?

In considering a scientific theory that has been proven false, the job of the realist is to demonstrate that though the theory was discredited it did in fact contain true or approximately true elements, and the theoretical elements and generalizations that contributed to the empirical success of the theory were preserved in consequential theories. Consequential theories provide a mechanism for refining earlier theories, enabling the scientist to filter out inconsequential elements. The theoretical elements and generalizations that did not contribute to the empirical success of a theory will be omitted from later theories, thus improving the overall truth or approximate truth of science as a whole. Acting on this recognition, the scientific realist can block the PMI.

Part of the series: UWO