IS ECONOMICS SCIENTIFIC? IS SCIENCE SCIENTIFIC? – Part 2

Part 2

In The Dappled World, Cartwright (1999, 31) upholds “metaphysical nomological pluralism,” which is “the doctrine that nature is governed in different domains by different systems of laws not necessarily related to each other in a systematic or uniform way; by a patchwork of laws.” In this view, “covering laws,” such as those of gravity, which hold over several unrelated domains, are distorted and lack realism when we disregard the necessity to apply them differently to the different domains.

“Perhaps laws in physics are not deployed in explanations as if they are true—but I claim something stronger: if the evidence is taken seriously, they must be judged false. . . . Laws we use are designed only to tell truly what happens in each domain separately” (Cartwright 1983, 12). Ronald N. Giere (I999, 79) argues for a similar approach to model based laws, but without making Cartwright’s claim for essential disunity. He instead argues that we must only stipulate that we “imagine the universe as having a deﬁnite structure, but exceedingly complex, so complex that no models humans can devise could ever capture more than limited aspects of the total complexity.” Thus, no fundamental law, and no model based on cause—and—effect relationships between phenomena, would be universally true, or hold in all circumstances—just as Cartwright claims. Like a subway map or a geological survey report, different models of the same area can be useful, depending on our goals. If we were traveling, we would want a very different model than if we were predicting earthquakes. But in both cases, we are examining the same terrain, and our model might distort the irrelevant parts in order to make crystal clear the relevant parts.

Milton Friedman ([1953] I968, 525) took this View farther in “The Methodology of Positive Economics,” where he claimed that “the goal of a positive science is the development of a ‘theory’ or ‘hypothesis’ that yields valid and meaningful (i.e., not truistic) predictions about phenomena not yet observed.” He claimed that accurate prediction is the sole goal of a model and that therefore, the assumptions of a model might well be unrealistic—win order to weed out any aspects of the world that do not lend themselves to prediction. But Cartwright points out that sometimes, the goal of a model is not just predictive power.

With laws of phenomena prediction is the goal, but with fundamental (one might say ontological) laws, other considerations must be called into play. For example, “many abstract concepts in physics play merely an organizing role” (Cartwright I983, 19). There is value in having fundamental laws that cover broad ranges of phenomena, but the price paid for range is loss of predictive and descriptive accuracy.

Cartwright proposes a model—based account of regularities in nature, based on the way a scientist would actually arrive at a fundamental law when thinking about a regularity or a set of observed phenomena. The first piece of Cartwright’s model of models consists of simple descriptions of phenomena that we encounter in everyday life. For Cartwright, these come in two varieties. The ﬁrst is raw observation. We walk around the Earth surrounded by facts. Perhaps we even search for such facts as regularities in the movement of planets or in the corrosion of seashores.

The other type of facts with which we deal are phenomena that we organize and create in laboratories and under other controlled circumstances. These we use to test, isolate, and analyze a certain aspect of reality. We try to “shield” it from other effects so that we can determine its cause. “Experiments are made to isolate true causes and to eliminate false starts”(Cartwright 1983, 7). In the controlled world of simplified and isolated effects, facts and their causes will interact in a very different way than in messy everyday life. Yet this is how we establish causation.

At this level lies a submerged but necessary component of Cartwright’s model: purpose. There is an exceedingly large number of phenomena that interact with each other in various ways. An observer is surrounded by myriad events, but organizes and pays attention only to some of them. Organizing these so as to highlight regularities cannot be done arbitrarily. Economists, for example, pay special attention to behavior that will make someone a proﬁt or a utility gain. Purposeful attention of this sort is an important part of the ﬁrst element in Cartwright’s paradigm.

Second: from the regularities found in chosen phenomena, we are able to generalize laws of phenomena and rules of thumb. These laws and rules hold a very special place in Cartwright’s model of models. They are what truly describe reality. Their function is to describe the past patterns of a phenomenon in order to be able to predict its behavior when it recurs.

It is important for Cartwright that this applies to very specific events. Laws of phenomena might, for example, be used to describe the shape of rain droplets. But these will be raw descriptions; no causal necessities are postulated. Such laws are often tested in laboratory conditions. What happens when two chemicals are mixed, or when a car crashes? What will be the behavior of a laser beam? The laws and rules of thumb that apply to real—life phenomena such as these are usually the things to turn to when you want to know what will actually happen in any given situation. How accurately will a particular kind of clock tell time? The best way to answer is not to contemplate the durability of metal gears under uniform stress, but to run a bunch of clocks against a dependable benchmark.

Laws of phenomena are sensitive to change if phenomenon patterns warrant it. These “laws” make no claims beyond usefully describing past regularities and perhaps narrowly extrapolating those descriptions to future phenomena that fit the initial phenomenon descriptions. The measure of a good law of this sort is its descriptive accuracy. (Though the occasional deviation may not be accounted for, the general pattern ought to be.) Prediction is not demanded, however, since one does not pretend to know when the phenomenon being described will occur again. One says only that if it does recur, it will be have in a certain way.