Throughout this chapter, I have only considered the relatively simple cases where the specification of an appropriate target system is successful. Yet part of the reason why formulating an explicit account of target systems is important, is that it can be very difficult to specify appropriate targets for a particular scientific purpose. For example, scientists sometimes erroneously classify relevant factors as irrelevant, or partition the system in a way which cannot be easily modeled. These target systems, while still real, are not successful given the scientists’ purposes. A full account of target system evaluation is
beyond the scope of this chapter, yet I will give the basic elements of such an account in the remainder of this section.
I begin my account of target system evaluation by considering three kinds of errors that theorists try to avoid. The first kind of error is when a target system is specified yet the abstractions and/or partitions are not optimal. The second is when a target system is specified, yet there is some mistake in the identification of its parts and/or properties. The third is when a theorist completely fails to specify a target system. I hope that the examination of these cases will provide additional support to my view that target systems are concrete real parts of the world, as it will show that many cases where targets seem to be immaterial are actually cases of mistakes in target system specification.
5.1 Inapt Target Systems
The way to think about the relative success of target systems is to determine whether or not they are apt. A subject or object is apt when it is suitable or appropriate given a particular purpose or set of circumstances. An apt description of a situation is suitable given the circumstances and gets the relevant information across to the audience. Whether a particular piece of information is relevant or not depends on the context of the situation. In the case of target systems, the context is the phenomenon we are trying to explain and the characteristics of the model or experiment that will explain it. The relevant information is the set of parts and properties that target selects from the actual system which will go into the model or experiment.
The relationship between target systems and our descriptions of them is similar to the relationship between things in the world and the terms we use to refer to them in natural language. For example, we might ask someone to bring “the glass on the table”. By doing so we are picking out a particular object in the world and saying something about its location. We leave out a lot of information, because we don’t consider it relevant. However, there might well be more than one glasses on the table, in which case our failure to specify which glass can lead to confusion. It would have been more apt to include some more information about the particular glass we were referring to, for example “the glass at the far end of the table”. Yet our failure to include this relevant information does not mean that we are failing to refer to the particular glass. We are simply using a description of it which is not particularly useful, given this context.
The same is true of target systems. Consider again the case of plant-soil feedback. Earlier target systems were rather inapt, as they did not include soil microbes. As it turns out, soil microbes are relevant for explaining plant competition, hence later targets, such as those specified in the work of Klironomos (2002) are more apt. It is possible to view these target systems as better than previous ones, but only in the sense that they help scientists explain a particular set of phenomena. They are neither better overall, nor more realistic in themselves. Target systems cannot be judged in terms of their aptness in isolation of their context, as they are just parts of the world. Whether or not a target system is apt depends on its relationship to the model and the phenomenon of interest. Of course, a description or representation of the target system which includes soil microbes and feedback is more realistic than another which does not include them. Still, the descriptions are descriptions of things in the world, which are parts of the world and hence real.
5.2 False Beliefs about Target Systems
Mistakes of relevance are not the only kind of errors that can occur in the specification of target systems. A different kind of mistake occurs when we manage to specify a target system, yet this specification induces or relies on false beliefs about the target. In the context of plant-soil feedback, this would occur if a scientist identified positive and negative feedback between the plants and soil biota, yet mistook mycorrhizal fungi for bacteria.8 In this case the scientists would be specifying an actual target
system, with particular plants and microbial biota, yet they would also have false beliefs about those biota (that they were fungi instead of bacteria).
Cases like the above suggest that we should maintain a conceptual distinction between target systems and our beliefs about them. Target systems are real parts of the world, but we can have mistaken beliefs about them. Going back to the analogy with language and reference, cases like these are similar to Keith Donnellan’s water-in-the-martini-glass example (Donnellan, 1966). If someone looks at an “interesting-looking person holding a martini glass” and asks “Who is the man holding a martini?”, they have succeeded in asking a question about the actual person in the room, even if the martini glass contains only water. In the case of target systems, the scientists are succeeding in specifying a target and asking questions about it, yet they have some false beliefs about it. These false beliefs might turn out to be
36
8 This hypothetical situation might seem absurd, but it is in fact quite easy to mistake one for the other if the partitioning of the target
problematic and require the re-specification of the target system, yet they do not change the nature of the target itself. The importance of the problems depend on what the target system will be used for and the manner in which it will be used.
5.3 Failure to specify a Target System
A more drastic type of problem occurs when scientists fail to specify a target system altogether. This might be the way to interpret the invocation of phlogiston to explain the processes of oxidation and reduction in the 18th century. The idea was that metals were composed of calx and phlogiston and would
become dephlogisticated when burned (Weisberg, Needham, & Hendry, 2011). We now know that there is no thing with the properties of phlogiston. The question is whether scientists in the 18th century were
specifying a target. If they were specifying a target, then this is a problem for my view, because they would specifying something imaginary.
There are two ways in which we can interpret cases like these. The first is to show that they are, in fact cases like the ones in 5.2, where scientists are succeeding in specifying a target system, yet have false beliefs about it. Indeed, one way to interpret this 18th century science is to identify phlogiston as hydrogen,
dephlogisticated air as oxygen and phlogiston-saturated air as nitrogen (Weisberg et al., 2011). The important point is that phlogiston was not part of the 18th century target system. Instead the scientists were
specifying a target system including hydrogen, but mistakenly attributed properties to it that it did not have. In other words, the scientists succeeded in specifying a target but had false beliefs about the target.
Yet even if we think that 18th century scientists were simply failing to specify any actual target
system, this does not jeopardize the nature of target systems in general. Failing to specify a target system is not the same as specifying a target system that does not exist. Sometimes scientists make big mistakes and are not talking about any real part of the world. Yet, as they continue to experiment and refine their views, they often do end up specifying an actual target, albeit with a number of mistakes and false beliefs. All this shows is that the natural world is complex and it sometimes takes centuries to even begin understanding what it is like and how it works.