Applying knowledge through explanation

Explanation is one of the manifest aims of science. An explanation allows one to understand the connection between items of knowledge. It is to be noted that explanations involve more than mere sound, law-involving arguments. To this end, one should draw the connection between explanations and causation. A cause is a factor that ensures that an effect takes place. Accordingly, causes and explanations work together to link different items of knowledge to each other – it is a science that should not be overlooked. Hempel attempted to use this type of science to formulate his deductive-nomological (‘DN’) account: an explanation resulting from logical derivations from laws and other information. Hempel’s theory has been criticised as not sufficiently relating to causation. This article will consider both views to comment on how explanations aim to achieve scientific understanding.

Hempel’s deductive-nomological (‘DN’) Account

The importance of the pursuit of explanation in scientific theory is not without its criticism. Duhem (1954) argued that explanatory knowledge must relate to the essential nature of things that is neither possible nor desirable. However, Hempel shifted this perception by theorising that explanatory knowledge can be recognised by the statements put forth by scientists when they claim to have an explanation. This can be so, Hempel claimed, without inquiring whether a scientist’s explanation is in fact a true account of the state of things. Hempel’s DN account represents a sociological approach. It is not concerned with the falsity or truism of the explanation. Rather, it is concerned with the goal at which scientists are seeking to aim, whether or not the scientist has taken all the necessary steps along the way of realising the end goal.

The DN account is designed to capsulate the form of deterministic scientific explanations (such as the outbreak of the Syrian War). Such explanations are deduced from events to be explained from a set of true propositions (including a minimum of one statement of a scientific law). The event to be explained is called the explanandum. The set of explaining statements (or the set of true propositions) is called the explanans. Strictly speaking, A DN account results in a deterministic event explanation, by covering a particular pattern of behaviour that occurred. Such accounts (or covering law accounts) will always derive from a deductive argument.

As Hempel puts it himself:

[A DN explanation] shows that, given the particular circumstances and the laws in question, the occurrence of the phenomenon was to be expected; and it is in the sense that the explanation enables use to understand why the phenomenon occurred.

Explanation addresses what one should have expected in the past, rather than what one should expect in the future (which has to do with prediction). There are three objections to the DN account that are worth mentioning.

The first objection to the DN account: irrelevancy

Hempel’s DN account requires: the premises to be true and the argument both valid and law-involving. An objection to this theory has been developed by Kyburg, Salmon and others. This objection is based on the idea that judgments of explanatory relevance are unable to be accounted for by Hempel’s theory. This is because the conclusion of an explanation may include something that is entirely irrelevant to the end result. For example, an explanation under the DN account could provide that:

Max felt the heat when he walked outside on Tuesday, thus: it was sunny on Tuesday.

This explanation appears to link the conclusion about the weather with Max’s whereabouts. In reality, whether or not Max was inside or outside on Tuesday, does not change the fact that it was sunny on that day. Max’s location is irrelevant to the end explanation. The citing of the event that Max walked outside neither physically nor relevantly relates to the explanandum: that it was sunny on Tuesday. Salmon contends that this issue, that the relation between the factors that make up an explanation and the explanandum itself lack physical relevance, is a deficiency in Hempel’s DN account.

The second objection to the DN account: lack of causation

Salmon makes a second objection: that the DN account does not sufficiently address the explanatory role of causal relations. The first objection illustrated that the DN account has the tendency to legitimise explanations that involve irrelevant factors leading to a conclusion. This creates an issue of irrelevancy in itself. However, it gives rise to a further issue: there is no causal factor resulting in the end effect. A cause is a factor that ensures that an effect takes place. Hempel’s theory allows for irrelevant factors rather than causal factors – and therefore the conclusion, effect or result cannot always be ensured by a pre-determinative causal fact. Inevitably, this calls into question the legitimacy of the conclusion or proposition put forth by the explanation. Salmon argues that the DN account needs a causal relation. This view has become particularly prevalent upon the rise of empiricist accounts of causation.

The third objection to the DN account: stringent legal requirements

The DN account is limited in the sense that it only recognises explanations if they cite a law, and it requires that the law encompass the explanandum (except for in probabilistic explanations). Thus, the DN account excludes those every day explanations that sufficiently possess a cause and effect relationship, but which lack any mention of the law. For example, the explanation that the bait on the rod was the cause of a fish that was caught by the rod – would not fall within the requirements of the DN account.

The IS account

Hempel’s probabilistic explanation of events (based on induction) is otherwise known was the IS account. It reflects the DN account, which gives explanations for events in a determinative manner. Both types of explanations comprise of law-involving arguments, which give reason to presume the occurrence of an explanandum event. Though, while the DN account is based on deductive reasoning strictly entailing the explanandum, an IS account is based on inductive reasoning which only confers probability on the occurrence of the explanandum.

Inductive reasoning includes an additional requirement, adopted by Hempel, that is not required by deductive reasoning and thus, this requirement marks a difference between the IS and DN accounts. Induction requires that all background knowledge be considered. Every element must be mentioned and form part of the inductive argument. Hempel incorporated this requirement into his IS explanation, which he named maximal specificity. The limitation to this requirement is that as soon as a new premise is added to the explanation, the level of probability by which the explanation would confer on its conclusion, will decrease.

Objections to the IS account

First, the IS account only gives partial explanations because it is concerned with probability rather than conclusiveness. Further, it is difficult to state that explanations create a high probability to confer truth on their explananda (the set of propositions preceding the conclusion). A second objection to the IS account revolves around maximal specificity, which requires that all background knowledge to be included in probabilistic event explanations. Certain background factors of the explanation may be relevant to an explanation but nevertheless unknown. Such unknown factors are not required to be considered in justifying the legitimacy of an explanation under the IS account. Coffa argues that this hinders the nature of expectation inherent in explanations – and that not the inquiry should not stop at acquiring background knowledge, but the search should continue for other relevant facts initially unknown.

The Statistical Relevance (‘SR’) account

Salmon suggests that the factors cited in explanations must have statistical relevance to the explanandum. He argues that the purpose of an explanation is not to prove that there was an expectation for the explanandum; but rather, to describe that factors are statistically relevant to the event of the explanandum. This requires, not a mere improvement, but a complete transformation of Hempel’s theory. Statistical relevance is offered as a broad, objective relation and does not require an account of the background knowledge of the explanation. All relevant factors should be mentioned – whether they are positive or negative.

Objections to the SR account

First, by the SR account requiring all relevant factors to be mentioned, regardless of their degree of relevance, they comprise an overflow of too much information. Second, statistical relevance does not necessarily result in a causal relation between the factors cited and the end conclusion. Third, sometimes statistically relevant factors are not relevant to the explanation. For example, if someone was texting on their mobile phone while riding a bike there is a probability that they will crash, however they ended up biking in a puddle and crashing because of the tension between the water and the bike wheel. The fact that the rider was on the phone may have been relevant statistically, but not to the end explanation of why the bike crashed. Finally, statistical relevance is mainly only applicable to explaining events.

The Unification account

Under the unification account, Friedman describes that a phenomenon is explained, and unified with all other phenomena, as a pattern of subsuming laws. There is a vast amount of phenomena that can be subsumed under the theory, which is proportional to its unifying power. The unification account is simple in that it precludes partial explanations, incomprehensibilities and irrelevancies. For example, unificationism would accept the explanation it was sunny on Tuesday rather than Max was outside and therefore it was sunny on Tuesday.

Objections to the Unification account

First, the unification account cannot explain why explanation follows the direction of causation. Second, due to the unification account unifying all phenomena in order to explain one phenomenon, it depends on what all other phenomena are. It makes explanation overly global.

Conclusion

Overall, there are many different types, levels of and approaches to explanation. Further issues as regards explanation persist in terms of whether there must be one standard for evaluating explanations, the spread of idealization in science but absence of it in explanatory accounts and lastly, the pragmatics in explanation. However, what can be certain is that there is an express causal approach to explanation. Absence of a proper causal relation was one of the main objections to all accounts and therefore the nature of a causal relation between facts cited and the conclusion in an explanation is an essential feature of explanatory work.