Contents
John Stuart Mill, in A system of logic, formulates five canons (rules or laws) as «regulative principles» for these experimental methods. These canons prescribe methods for discovering and demonstrative substantiation of causal laws and causal relationships.
1. Similarity method (consent method)
If two or more cases of the phenomenon under study converge in only one circumstance, then this circumstance is the cause (or part of the cause) of the phenomenon under study.
An example of a deduction according to this first method is taken from J. Dollard and N. Miller: “It is interesting to note that one of the common symptoms of the extreme manifestations of fear in battle is difficulty in speech, which can vary from complete dumbness to stammering and stuttering. . Similarly, a person who has an acute fear of public speaking loses the gift of speech. Many animals stop making any sound when frightened, and it seems clear that this tendency is adaptive, preventing them from attracting the attention of their enemies. In light of these facts, it is possible to speculate that the fear drive has an innate tendency to elicit a voice-blocking response.
2. Method of differences
If the case in which the phenomenon under investigation occurs and the case in which it does not occur are exactly alike in every detail, with the exception of the one being investigated, then the circumstance that occurs in the first case and is absent in the second is the cause or part of the cause of the phenomenon under study.
The illustration of the difference method is taken from the work of R. Bushbaum. “The primitive brain, as we have seen in planarians, serves mainly as a sensory relay, a center for receiving stimuli from the sense organ and then transmitting impulses down the nerve trunk. This is also true for the Nereis (annelid worm), which, after the removal of the brain, does not lose the ability to coordinate its movements — in fact, its motor activity even increases. If he encounters an obstacle, he does not step back and does not change direction, but persists in his unsuccessful attempts to continue moving forward. This highly maladaptive type of behavior shows that in a normal annelids (in our case, Nereis), the brain ganglion performs an important function that it does not have in flatworms (planarians) — the function of inhibiting movement in response to certain stimuli.
3. The combined method of similarity and difference
The combined method of similarity and difference is often regarded as a simple joint application of the method of similarity and the method of difference, in other words, it can be applied only in those cases where the first two methods are applicable separately. However, the formulation of this method by Mill himself does not agree with such an interpretation, namely, his formulation is as follows:
If two or more cases of occurrence of the phenomenon under study are similar in that they contain the same general circumstance, and two or more cases of non-occurrence of the phenomenon are similar in that they lack the same circumstance, then we can conclude that this circumstance , which differs in both series of cases, is the effect or cause (or an integral part of the cause) of this phenomenon.
It would seem that in such a formulation such a method would simply be a double use of the consent method. A more general interpretation of the combined method explicitly provides for the separate use of the first two methods — similarities and differences.
There is a third interpretation of the combined method, which makes it a much more powerful tool of induction in situations where it is impossible to directly apply either the method of similarity or the method of difference. In an appropriate classification, all inductive inferences falling under any of the three models described are generally regarded as examples of the combined method of similarity and difference.
4. Method of residuals
If in the phenomenon under study a part of the circumstances can be explained by certain reasons, then the remaining part of the phenomenon is explained from the remaining previous facts.
Version of the wording: If we subtract from the phenomenon under study that part of it, which, as established by the previous inductive reasoning, is a consequence of certain antecedents, then the remainder of this phenomenon must be a consequence of the remaining antecedents.
The residual method is sometimes referred to as a strictly deductive rather than an inductive model of inference. The application of any of the previous methods requires consideration of at least two cases, while the residual method may be limited to the analysis of a single case. None of the other methods formulated by Mill require recourse to any a priori established causal laws, while the use of the residual method relies directly on such laws.
5. Appropriate change method
If a change in one phenomenon is followed by a change in another, then we can conclude a causal relationship between them.
An illustration from Mill himself regarding the causes of the tides: “However, we still have one possibility. Although we cannot completely exclude a certain presupposition, we may be able to produce, or nature may be able to produce for us, some modification in it. The modification implied here is not tantamount to its complete elimination… when we find that all changes in the position of the Moon are followed by corresponding changes in the time and place of the rise of the water, this place always turns out to be either the part of the Earth closest to the Moon, or that part of the Earth, farthest from it, we have sufficient evidence that the moon, in whole or in part, is the cause that causes the tides.
Criticism
Mill’s canons have been criticized on a number of different grounds. They say nothing about an analysis of the circumstances themselves. If the circumstances considered are not adequately analyzed or classified, these methods will not work. For example, if the circumstances are bourbon and water, scotch and water, brandy and water, and vodka and water, the similarity method will lead us to conclude that intoxication is caused by water. However, the analysis of the use of alcoholic beverages and other ingredients that cause alcohol intoxication requires prior knowledge of causal relationships.
Another requirement for canons, which is a condition for drawing useful conclusions from them, involves the elimination of irrelevant circumstances. However, here, too, the relevance of the circumstances can only be discovered in the course of preliminary research.
Any productive use of Mill’s Canons requires preliminary hypotheses as to what circumstances may be causally related to the phenomenon being studied. In the presence of such hypotheses, the canons are useful in excluding possible causal circumstances.