Humans have always strived to understand the world they live in. The need for explanations for why things work or why they are here is widely discussed on a daily basis (Hempel, 1966). Some turn to God, others look at forces of nature and some seek other ways of explaining things (Hempel, 1966). Whilst this may be helpful on an individual level, it does not help science (Hempel, 1966), therefore something else must help us explain things. Scientific Explanation has been around since Pre-Socratic times (Woodward, 2017). The idea is that there are ways to explain the worlds we live in (Woodward, 2017). These can be done through scientific and non scientific examples (Stanford). The end goal is to try and explain WHY things happen (these can be events or physical things) (Woodward, 2017). This essay will look at the background to scientific explanation, Hempel’s DN/IS model or more commonly known as the Covering-Law Model. The essay will discuss in depth the Causal Theory before concluding that causal information is necessary to explain an event.
As mentioned above, the goal of science is to try and explain the WHY of things and events. The need for explanation is rooted deeply within all of us as we ask why questions everyday. Science tries to explain the broader why questions. Explanations have “two major “constituents”: an explanandum, a sentence “describing the phenomenon to be explained” and an explanans, “the class of those sentences which are adduced to account for the phenomenon”” (Hempel 1966″,Woodward, 2017). In simple terms an explanandum is the thing or event that needs to be explained and and explanans is the thing/event/laws that gives reason as to why that event happened. There are three conditions to explanations (Hempel, 1966, Woodward, 2017). Firstly the event needs to have happened (Hempel, 1966, Woodward, 2017). It would be hard to explain an event that has not happened, (this will be looked at later). Secondly, there needs to be Explanatory Relevance Relation i.e there needs to be a relationship between explanandum and explanans (Hempel, 1966, Stanford). And thirdly, there must be truth (Hempel, 1966, Stanford). An explanation cannot be based on false information. Hempel (1966) also said that there was a requirement of testability. Basically an explanation needed to be empirically tested (Hempel, 1966). These requirements are there to satisfy the empiricists. These conditions are applied to his Deductive-Nomological Model also known as the Covering-Law Model (Hempel, 1966, Woodward, 2017). The Nomological means that it looks at laws of nature or things that look like laws but are not logically necessary (Woodward, 2017). Laws are used in order to explain the explandum phenomenon. According to Hempel (1966) “laws are regularities that play a key part in explanations”. They “meet certain further conditions” (Woodward, 2017) which help connect an event to a reason to why it has happened. They can be thought of as a universal form (Hempel 1966). The Covering-Law model (or DN) uses general laws in order to determine the explanation by using it along with facts to create the explanans sentences which will then give an explanandum. The L are the laws whilst the C stands for facts. Laws will only be considered if they are true.
L1, L2″,.”,.”,.”,.”,.”,Lr = Explanans Sentence
C1, C2″,.”,.”,.”,.”,Ck = Explanans Sentence
E = Explanandum Sentence
The use of this can be shown through the example of the future position of Planets (Woodward, 2017). Take use of Newton’s laws of motion and inverse square law governing gravity which explains the Law aspect of the equation and couple that with more information on the mass, velocity and position of both the Sun and Mars, which is fact (Woodward, 2017). A person will deductively arrive at the explanandum of the future position of Mars (Woodward, 2017). The laws and facts satisfy all the conditions that was set out; the explanans and explanandum have a relationship, they are true due to the laws and facts and therefore even though the event itself has not happened, we can come to the conclusion that it will as the facts and laws are true (Woodward, 2017). However, sometimes sentences do not explicitly list a law but presupposes that there is one (Hempel 1966, Woodward, 2017). For example, ice will melt when salt is put on it (Hempel, 1966). There is no mention of a law but it is supposed that the freezing point of water is lowered when salt is put on it (Hempel, 1966). It does not give a lot of information, but the idea is that if “Event of kind F occurs it is always followed by event G” (Hempel, 1966). Basically salt being put on the ice (F) caused it to melt (G). The idea is that some events are explained by other events and that you did not need cause to explain things as long as there are laws and facts that are true and connected (Hempel, 1966).
However, there is some confusion over what is classed as a law as there needs to be a distinction between law and accidental truth. For example, a generalised sentence such as all students eat pot noodles, can only be true accidentally and is not a law (Hempel 1966, Woodward, 2017) . Even if true, it cannot explain why all students eat pot noodles (Hempel 1966″,Woodward, 2017).In contrast the sentence that ice will melt if salt is put on it is a law as it can be used with information to explain why the salt makes the ice melt as demonstrated above (Hempel 1966, Woodward, 2017). Laws should be used liberally as sometimes it does not apply. For example “All bodies of pure gold have a mass of less that 100, 000 kg” (Hempel 1966″,). Whilst this might be true based on observation and data collected to date”,there is no basic law of nature that shows that there will never be a body of pure gold that has a mass greater than 100″,000, therefore it is an accidental truth (Hempel, 1966). Law can support counterfactual conditions, a generalisation cannot. For example; “If A were the case then B would be the case” (Hempel, 1966). Laws can also support subjunctive conditionals; “if A should come to pass then so would B” but to become a finite conjunction there needs to be additional information (Hempel 1966, Woodward, 2017). There is also the concern of probabilistic laws (Hempel, 1966). Probabilistic laws give a high chance of something being true but is not a law or truth. For example, if someone were to contract the measles, the explanation that they were around someone who had the measles would link the explanans and the explanandum. However, because not all forms of measles is contagious, we could not say that that it is law and therefore can only say with high chance that the first person contracted measles from the second person, therefore it is inductive, not deductive (Hempel, 1966) . This can be viewed at through the IS model:
The DN, IS models are useful in helping understand the connection between events and allows us to conceptualize them as nomic expectability (basically based on laws). However, Hempel fails to accept causation in its basic form i.e X = Y only if X causes Y as he believes that there needs to be empirical data for the explanation (Woodward, 2017). Further to that, there is an issue with laws within his theory as the conditions are not always necessary, for example, with low probabilities or good examples. Also, explanations have the same structure for predictions but some predictions are not explanations. There is no mention of historical explanations which do not involve laws. With this in mind, there needs to be a better example for explanations. This leads us on to Causal Theory.
Causal explanations is essentially the cause of something happening for example, X causes Y (Woodward, 2017). Causation is usually the most focused on as it enables us to establish a connection between why things happen in a simple form without concerning ourselves with scientific law (Woodward, 2017). All explanandum events have their causes, as can be seen with the demonstrations above (Lewis, 1986). Causal history is complicated as there are multiple layers that lead to the explanandum event. For example, X causes Y, but R causes X which in turn causes Y, but R is caused by M, which in turn causes X and causes Y and so on (Lewis, 1986). The issues with causal history is that you may never finish it, it breaks off into different parts and if someone said that R is the reason for Y then they are correct but someone is equally correct if they say M caused Y (Lewis, 1986). All explanandum is causal dependent (Lewis, 1986). Lewis (1986), believed that all explanations had to provide some information about its causal history in order for the event to be explained properly. Causal history can be very specific or can be abstract, which shows that there are many ways in which an explanation for an event can happen without “naming one or more of its causes” (Lewis, 1986). For example, imagine there was a Triangle of glass that had various refractive index and a beam of light hit at point A and left at point B while travelling on point C between A and B (Lewis, 1986). Point C is the least time to travel between A and B nd according to Fermat’s principle, that will be the path that all light travels (Lewis, 1986). There seems to be no causal explanation for this, however there is if the person can bring in additional information around the event (Lewis, 1986). Fermat’s principle will use some law to explain the reason for the light travelling on that path and from there the person will be able to see a pattern which in turn allows them to gain an understanding of the causal history of the explanandum (Lewis”,1986). On another note, negative information can still explain causal history (Lewis”,1986). The example Lewis (1986) used was a star collapsing then just stopping. It stops because it has gone as far as it can, but that does not provide information as to why it cannot go further (Lewis”,1986). However, there is a causal history as to why it stops at a certain point (Lewis”,1986). Dispositions can also provide information on causal history (Lewis”,1986). Dispositions “is to have something or other that occupies a certain causal role”, so while dispositions can be explained without causal history, there is still a causal history behind the reason for possessing that disposition in the first place (Lewis”,1986). For example, if a person is immune from a disease because they got a vaccination that gave them antibodies to fight the disease then there is an element of causal history as something is caused by something else, there is an explanation as to why they are immune (Lewis”,1986). Existential facts do not count towards causal histories, rather they provide truth making instances that do count towards causal histories (Lewis”,1986). Lewis’ (1986) idea is that there is some kind of “general explanatory information’ which means that similar kind of events might hold similar features, but that is not bound by universal laws unlike Hempel’s (1966) Covering-Law theory (Lewis”,1986).
Salmon (1998), on the other hand, believed that causation involved a physical process, for example a bullet flying through the air (Salmon, 1998, Galavotti, 2018). Salmon (1998) used statistical relevance and probabilistic causality in order to determine a causal explanation for the mechanisms behind an event occuring (Galavotti, 2018). Salmon (1998) believed that an explanation required a causal relation between events. For example, the barometer and the storm. The barometer will tell you when a storm is coming by there being a sudden drop (Galavotti, 2018). The barometer explains that a storm happened but not why the storm happened (Galavotti, 2018). Salmon (1998) showed that statistical correlations sometimes invoked explanations as statistical properties did not always provide causal information (Galavotti, 2018). Salmon (1998) developed the idea that a distinction can be made between statistical and causal relevance for probabilistic causal explanation. He used a “screening off” relation in order to connect the relationship between statistical and causal relevance (Galavotti, 2018). The idea is simple. Taking the barometer and storm example; “B = Barometer, S = Strom and P = drop in atmospheric pressure” that would cause the barometer to drop and a storm to happen (Fraassen 1977, Galavotti, 2018). The idea is that there is a relationship between B and S so “p(S|B) > p(S)” as B is “statistically relevant to S” (Fraassen”,1977, Galavotti, 2018). So if P is looked at, “p(S|P&B)=P(S|P)” (Fraassen, 1977, Galavotti, 2018), B becomes irrelevant in the presence of P which means that B gets removed from the equation but B does not remove P as the “screening off relation is asymmetrical” (Galavotti, 2018). This helps break down causation in order to see what really affects something, which in this example is the storm. There is criticism to the Causal theories. One is that there seems to be a lot of causes that are not all explanatory. For example, The Big Bang Theory does not explain why I had breakfast this morning, although one could argue that if I looked back far enough in the causal histories, I would find the reason.
However, we could use Pragmatic, Syntax or Semantics in order to determine specific context of the causal explanation that we are interested in (Scriven, 1975). For example, if there was a car crash we would want to know what caused the crash, however there are multiple reasons that the crash occurred, (Lewis 1986). There was another car which caused the car to swerve, which caused the drunk driver to break, which spun the car out of control because the tyres were bald (Lewis 1986). So how would a person determine the cause of that crash? They could look at the pragmatics of the situation, for example the language and context that the example was used in (Fraassen, 1977). They could look at the syntax, which is the logical form of the event (Parker and Brander, 2008). Or they could look at the semantics which is looking at relevance relation (Parker and Brander, 2008). By breaking the causation line down using one of these three tools, it allows a person to give an explanation that is appropriate to the context in which it occurred (Fraassen, 1977). The quality of the answer depends on how it fits with the context (Fraassen, 1977). There are three different areas that need to be looked at. They are the topic, the contrast class and the relevance relation (Parker and Brander, 2008). The topic is the event or fact or law that needs to be explained (Parker and Brander, 2008. The contrast class refers to the different alternatives. These can be ambiguous, for example why did Adam eat the apple? This could mean, Why did Adam eat the apple?, Why did Adam eat the apple?, Why did Adam eat the apple?. “A good explanation will cite what actually happened over the alternatives” The Relevance Relation does not always ask for the causes of events, instead it just asks for actors which “favour the explanandum event in other ways”, for example, “Why is the Porch light on?” answer, because someone turned it on, is a relevance relation (Parker and Brander, 2008). So a ‘Why’ question can be answered in two ways (Hendrey, 2019). They can be answered by Direct answers which provide an explanation or through a Corrective answer which “denies a presupposition of the question” (Hendrey, 2019). Direct answers can be thought of through this equation; Pk in contrast to (the rest of) X because A. Corrective answers can be thought of through this equation 1. Pk is true, 2. No other member of X is true, 3. There is at least one true A such that R(A, ). Different kinds of corrective answers can show how a preposition is different (Hendrey, 2019). It can show how an explanandum event does not occur for example, ‘Why did you finish the Peanut Butter?”, “I did not.” It can also “reject the request for explanation” which means that there is no answer to the ‘why’ question but it can answer related questions(Hendrey, 2019). There is some criticism to this. Salmon (1998), believed that Van Fraassen’s (1977) theory “fails to put any significant restriction on the nature of relevance relation as relevance relation can be defined by extension”. Essentially, this means that anything can be deemed an explanation.
To conclude, I believe based on the theories and models discussed in this essay that explanation of an event must give casual information. No matter which system is used, it talks about causation in some way or form. With Hempel (1966) it is through laws and facts, with Lewis (1986) it is with causal histories and with Salmon (1998) it is through statistical correlations. Either way, there is some form of ‘X is caused by Y’ understanding that shows that casual information is necessary to explain the why questions. Whilst each theory does have its downfalls, the overall aim is the same and that is to answer the why questions which can be done through casual information.