The Deductive-Nomological Model

The Deductive-Nomological (D-N) Model gives an account of explanation through its basic form, the Covering Law Model. The D-N Model asks the basic question “What is a scientific explanation?” The aim of this paper is to answer that question and further develop the definition of an explanation by problematizing the D-N Model’s account of explanation, providing a solution to one of those problems, and then further problematizing that solution. By examining the details of an example that the D-N Model explains well, we can see why this model was popular in the first place before describing two of its major problems. Then, by looking at Wesley Salmon’s account of scientific explanation, we can see just how problematic the flaws in the D-N Model

are, and we can scope out potential solutions to the problems. Finally, we will delve into the teachings of van Fraassen to explore his criticism of Salmon.
The D-N account of explanation, also called the Covering Law Model, states that an explanation simply points out an instance of a general law.

Scientific explanations are broken down into explanans and explanendums. Here, explanans are things that do the explaining, and are further divided into laws and conditions; explanendums are the observed phenomenon. The D-N model states, more specifically, that scientific explanation is a demonstration of how the explanendum is an instance of a true general law. Hempel and Oppenheim’s Bent Oar Example is a terrific illustration of a time when the D-N Model gave an accurate account of explanation for the event that occurred. In this example the explanans are the general law that light bends between mediums and the condition that refraction of index and density of water are different than those of air. The explanendum is that the oar appears bent in water. This explanation says that, due to the conditions, the observed phenomena occurred and is in accordance with the law. While there are situations in which the D-N Model does a decent job of capturing the explanation, there are many instances where it does not provide a fully compelling account of

explanation.
The first of these instances is due to the Relevance Problem. In these cases, the condition used to explain the observation is not relevant to the explanation. Meaning the same outcome would be observed had the condition been removed from the equation. In class we talked about the example of a man on birth control, which is a perfect representation of the relevance problem. The law in this example is that all men who take birth control pills are not pregnant; the condition is that a man, Toby, takes birth control pills. The observation is that Toby is not pregnant. According to the D-N model, these explanans fully explain the observation that Toby is not pregnant. Here, the explanation is that Toby is not pregnant due to his birth control pill regimen; that makes perfect sense under the Covering Law Model. The problem is that condition is not relevant to the situation. What it means to be relevant will be discussed in detail later, but for now it is enough to say that since human men do not have the ability to gestate, the observation will remain the same whether or not Toby takes the birth control pills.
Second, there is a problem with the symmetry of the D-N Model. This Asymmetry Problem is caused by a reversal of the condition that does the explaining and the observation being explained. The barometer/ storm example is a great representation of this problem. The law is that when barometers fall, storms appear; the condition is that the barometer is falling. The observed phenomena is that there is a storm front rolling in. What makes this explanation problematic is that the barometer falling did not cause the storm to appear. This problem is called asymmetrical because when the condition and the observation are reversed, the explanation makes sense. If we apply this to the barometer example, the law remains that when barometers fall, storms appear, but now the condition is that there is a storm rolling in and we observe the barometer falling. In simpler terms, the new explanation is that the barometer falls because there is a storm. Salmon eventually explains that this is also inaccurate. The two do not have a direct correlation, rather the lowering of air pressure is an indirect link (a common cause) between the two. This, however, will not be discussed any further in this paper, as there are more pressing issues to potentially solve.
Now that we have discussed some of the major issues with the D-N Model account of explanation, it would be appropriate to propose a potential solution to one of the problems discussed.

Salmon’s writings state that scientific explanation requires a description of causal processes, of which there are two kinds: causal connection and common cause. The focus of this section will be on solving the relevance problem and, according to Salmon, what it means to be statistically relevant is that there is a causal connection. That is, there is a direct connection between A and B, where A directly causes B or vice versa. In the example of the man on birth control, there is a one hundred percent correlation between a man taking birth control pills and not getting pregnant, but there is also a one hundred percent correlation between a man not using birth control pills and not getting pregnant. Salmon’s solution is to say that, due to Toby’s lack of a uterus and other functioning female reproductive organs, if Toby stops taking birth control he will never get pregnant. Therefore, there is no causal connection, making the pills irrelevant to this

situation.
That being said, Salmon is not entirely correct in his assessment of what is required for explanation. For him, a complete set of statistically relevant factors, the corresponding probability values, and the causal structure behind relevant factors (all of which are context-independent) are needed to give a full explanation. In contrast, Van Fraassen states that it is the context-dependent salient factors that are the most important or most relevant factors in a situation’s outcome. He describes what Salmon is doing as more of an interest in causation, which focuses on “a net of causal relations” and science’s description of said causal net (van Fraassen, p. 70-71). Whereas an interest in explanation requires a more intensive focus on the why questions and the salient factors in the causal net that answer those questions (Van Fraassen, p.70-71).
Van Fraassen uses the Irish Elk Extinction example to show the importance of salient causes. There are many factors that contributed to the extinction of the Irish Elk and van Fraassen recognizes that, but Salmon would say that all of these factors are the most relevant and are necessary to explain why the Irish Elk went extinct. This is where van Fraassen would criticize him, because while there are many causal factors that contributed to the extinction, the salient factor was antler size. This means that when asked why that species of elk went extinct, one would say that antler size is the only important factor, because if all the other contributing factors disappeared, the elk still would have gone extinct. But if we kept all the other factors and eliminated the antler problem, there is a good chance the elk would have survived. To further explain the importance of context and simultaneously criticize Salmon, Van Fraassen used Hanson’s Car Accident as an example. As we discussed in class, this example contains many salient causes, so there are many equally good explanations for the outcome. This is because the salient cause changed depending on which person was asked what question. When asked about the cause of death after the accident, the doctor would say something about a multiple hemorrhage, the mechanic would blame the faulty break system, and the city planner would blame the location of the tree. Context is incredibly important and Salmon does not include it in his description of an explanation, making his solutions invalid.
Defining explanation is a difficult task. When we first encountered the D-N Model, it seemed to be a straightforward account of explanation, but upon further inspection, we found several problems associated with this account. The problems of relevance and asymmetry were best outlined using the example of a man on birth control and the barometer/storm example, respectively. After discussing Salmon’s account of explanation--the causal account—where explanations reveal the causal structure, it seemed that we had a full, believable justification for an events outcome. Once again, we were wrong, because van Fraassen found problems in Salmon’s account of explanation. Through the Irish Elk example and Hanson’s Car Accident, we came to understand the weight salient factors hold in an explanation. The basic question of the Covering Law Model, ‘What is a scientific explanation?’, is best answered by saying that an explanation should consist of properly aligned and relevant explanans and explanendums, where the cause of occurrences of the past, and how and why a phenomena took place are detailed in their entirety.