Schools of Inductivism and Deductivism

¶ … Schools of Inductivism and Deductivism

Argumentation in general has lain the way, ever since the philosophers of Antiquity, to two opposing schools of inductivism and deductivism. As opposed to inductivism, deductivism starts with the formulation of a hypothesis, but leaving open the possibility that this could be proven wrong by testing existing data in the hypothesis. Although generally regarded as an interpretative approach, there are also opinions that grant deductivism its evaluative perspective. According to this perspective, "anything less than a relation of entailment between premises and conclusion is unsatisfactory."

This means that, in working with deductivism for scientific explanations, there are two distinct components to be referred to and discussed. The first is the premises on which the deduction works, while the second is the way that the premises lead to the conclusions and whether this is logical or not. This is a reasonable assumption: as previously mentioned, deductivism starts with the hypotheses, which is based on a certain set of premises. The validity of the final conclusion should be well interrelated to the validity of the initial set of premises. It is logical to understand that if these are not all correct, the final conclusion will also not be reflecting the actual truth.

On the other hand, the path from those premises to the final conclusion is also essential in its validity. Again, similarly to the validity of the premises, if the path by which the conclusion is reached is not correct, then it is also likely that the final conclusion will be not necessarily incorrect, but different from the valid one which would have followed the valid path.

The first and potentially most important argument in favor of deductivism in scientific explanations is by following on its opposite, inductivism, and on Karl Popper's immediate arguments against inductivism as a reasonable base for scientific explanations. Inductivism is based on a number of finite observations from which a general conclusion can be drawn. As such, if we were to prove, for example, that all ravens are black, we would have a set of observations of ravens, numbered Raven 1, Raven 2 etc., note that all are back and derive from this that Raven k will also be black.

Popper's argument against this is simple: one cannot state that Raven k or all ravens are black, according to the methodology here described, until all ravens have been observed and noted as black. However, this is close to impossible, first of all because of the feasibility problem (observing all ravens in existence) and second because the hypothesis is so easily made invalid with the discovery of only one exception (one raven that is not black). According to him, a set of experiences is not enough to infer an overall conclusion.

If we tie Popper's argumentation to the nature of scientific method and we consider the fact that the role of scientific method is to draw valid theories and conclusions from existing data and observations, deductivism seems to be closer to validity than inductivism. One of the arguments in this sense is quite simple and is strongly tied to the issue of validity.

In inductivism, as previously mentioned, a scientific theory can lose its validity on the basis of one simple observation that is not congruent with the other observations. As shown, one raven that is not black will throw away the validity of the theory that all ravens are black. Deductivism, at least from this perspective, seems to be closer to the rationality of scientific method. Inductivism tends, from this perspective, to give out an inaccurate view of scientific methodology.

However, the most important argument that Popper seems to bring relates to the fact that observations in induction only bring about punctual conclusions and singular statements and is not sufficient to generate reliable and viable final conclusions. On the other hand, deductivism is able to generate a universal theory, which is obviously fundamental when it comes to scientific methods.

We will only briefly mention Hume's skeptical arguments against inductivism because, in our opinion, they are circular arguments themselves rather than an appropriate philosophical approach to deny inductivism and propose inductivism as a proper scientific method.

According to Hume, an inductive hypothesis can be justified either by a deductive or inductive rule. If it were justified by a deductive rule, then it would not be an inductive hypothesis. If it were justified by an inductive hypothesis, it would be circular and, again, not functional. According to Hume, this is enough to show that no inductive rule can be justified.

Besides the fact that this seems to be the sort of philosophical syllogism that can be applied just as well for a deductive hypothesis when applied as a scientific method, it is all the more a circular argumentation in its and cannot form the basis for further discussion.

Before moving forward on deciding how deductivism relates to theories of scientific explanation, we should have a short closer look to the nature of science itself, as well as of scientific methodology. According to different sources, the basic components of science can be data, theories and shaping principles. These three components will help understand why deductivism is better fit for scientific methodology than inductivism.

In my opinion, one of the most important arguments in favor of deductivism as a proper and applicable scientific method comes from the fact that a scientific method will only admit as evidence facts, "statements that everyone can verify by direct observation." I would go further than this though and point out that the strength and value of deductivism comes from taken observation to the state of experimentation and continuous testing of a certain hypothesis.

Indeed, rather than the simple observation in the case of inductivism, we are dealing here with experiments, that is, not only trying to find an observation that denies the validity of the original hypothesis, but turning this hypothesis into a theory and testing its validity against different constraints and a validation mechanism. I think that this experimentation in itself will tend to increase the value of the theory in the end.

In fact, Popper himself points out that the two main problems with inductivism and the induction mechanism come from the fact that induction is not a logical truth and that it is not derived from experience. We have referred to the latter, what about the first? Is induction a logical truth? In my opinion, it is not and the reason for this is that it is not a mechanism that brings the original theory/hypothesis from one phase to the next, through a set of logical steps along the way.

Deductivism, through Popper and, later, through the likes of Peter Godfrey-Smith (although obviously going back as far as the Greek Antiquity), proposes a 4-steps model for the deductive algorithm that includes (1) gathering data and observations on the subject, (2) constructing a hypothesis from this set of observations and data, (3) deducing the consequence and experiment to see its validity and (4) decide on the level of corroboration.

In this sequence, the 3rd and 4th phases seem most important from a deductivism perspective. The theory or hypothesis itself is not important, because it will be matched against the empirical evidence in existence in the real world and, only upon the corroboration of the conclusion, will it be confirmed in its validity. From this point-of-view, the level of irrationality that might accompany a human perspective is at least partially removed.

In deductivism, as related to scientific method, according to Popper, the essential part is that of experimentation, which can reasonably check the validity of the theory or hypothesis by measuring its capacity to predict future events. In other words, a hypothesis will be tested according to the validity of the results that the experiments provide, as compared to the hypothetical statement originally made when this was first issued.

This argument seems a better approach to scientific method than the inductivism, as in a comparative approach experimentation vs. observation. With observation, one simply records data and measures it against the hypothesis. With experimentation, one actually tests the hypothesis in practice and records not simple data, but outcome data from those experiments, as measured against the original hypothetical approach.

There is however a common flaw in both deductivism and inductivism, especially in the context of scientific method: in both cases, observation or experimentation should be non-finite in order to conclusively be able to record the validity of a theory. How many experiments should a scientist perform in order to falsify his scientific theory and where can a scientist actually successfully support the idea that the experimentation phase is over and the proposed hypothesis is valid? There is no conclusiveness here and scientific rigor is also lacking in the case of deductivism, at least from this perspective.

As some have pointed out, there are two approaches in determining the validness of deductivism in scientific methodology: descriptive and normative. From a descriptive perspective, deductivism is wrong because it presents only a limited view of things and reasoning obviously goes beyond that. However, would deductivism be true from a normative perspective?

This is a question that relates with the fundamental question that a principle would be justifiable from a deductivist perspective: if H - hypothesis is the best explanation for the fact or facts that are to be explained, is it reasonable to accept H. As being valid? We will not go into a discussion on what reasonable might mean and what can actually be considered being reasonable. However, in my opinion the answer to such a question could be 'no'. We can accept H. As being reasonable, but I don't think we can accept H. As being valid. On the other hand, deductivism in scientific methodology does not necessarily propose 100% valid statements and hypothesis, but hypothesis that can be worked with.

Is this enough for a scientific hypothesis? Again, this is a difficult question to answer. On a highly likely scale, it also greatly depends on how valid the hypothesis needs to be in the scientific framework in which it is being used. Quite often, one can use an operational hypothesis rather than a valid one and this can be enough to generate a valid hypothesis later on.

Many, including Musgrane, have argued that conclusiveness is not necessarily directly impacting the validity of the premise. In other words, the simple fact that there are not conclusive elements to support the hypothesis does not necessarily make it invalid or not a sufficient reason not to take into consideration the initial premise.

This is partially true, the problem appears that we seemed to have used the same argumentation when denying inductivism its capacity to act as a scientific methodology. Indeed, inductivism was denied and deductivism preferred exactly because it was vulnerable to new observations and not valid because of that. Similarly, deductivism is not conclusive either.

The problem I see in Musgrave's interpretation and support for deductivism is that he argues that even if there is no evidence to probabilistically support the hypothesis, it does not necessarily mean that it is not true. In other words, he argues that rationality can also accept evidence-transcendent beliefs. In my opinion, this is not something that can go hand in hand with the scientific method, which needs to rely on facts to support theories and hypothesis. Additionally, I don't think it is something rational and can lead, in fact, to irrational explanations of facts.

On the other hand, the lack of conclusiveness or finiteness should most likely not disavow the usefulness of deductivism as an introspective towards scientific method. In science, one has to agree that there are theories that do not encompass the entire scientific spectrum. The best example in this sense related Einstein's theory of relativity with Planck's research into the microcosm and his study of quantum mechanics.

In Einstein's case, his theories best present the macrospace and work when applied to the movement of large objects in space, to planets etc. Planck's theories best work when applied to microspace. However, the two theories, both deductively created, do not deny one another, but, in fact, complete each other by covering two perspectives of things. Their validity, in this case, is not limited by their applicability and I think this is also a very important thing to take into consideration when discussing deductivism and its applicability and relationship with the scientific method.

Here we have two different theories, both valid in their own framework, but invalid when applied to the others' framework and set of premises. There validity however has been proven given the distinct premises they have worked with. Can we support, in this sense, the idea that deductivism works as a scientific method applied within a limited applicability range, but without contradicting the statements we have previously made on this subject?

Science in general and physics in particular has not yet discovered a hypothesis or theory that can successfully justify events in different of its areas. Even more so, this is probably a direct consequence of the fact that premises sometimes differ in validity, depending on the different segments of the materialistic approach. The framework can make a certain premise valid in one system, but invalid in another system. This is true in the previously mentioned theories laid out by Planck and Einstein, but also in thermodynamics or electricity theories, for example.

This can go even to the same branches of a particular science, for example geometry. Euclid has stated, based on his postulates, that the sum of the angles in a triangle is exactly 180 degrees. On the other hand, non-Euclidians such as Riemann stated that it is either less or more than that. The reasoning behind these different deductive approaches comes from the initial premises: Euclid uses his fifth postulate, while Riemann does not include its validity in his own research, which means that his final deductive process leads to a different conclusion.

The reason I have mentioned this is because it is easier to understand lack of conclusiveness that deductivism brings to scientific method and still see why deductivism is still a valid approach. Rationality can be seen as a limited concept itself and something rational in one system can become irrational when the system is changed without this affecting the initial rationality in any way. I think that, from that perspective, we can also extrapolate deductivism and argue that it can be successful as a scientific method even if the number of experiments will be limited.

One of the problem that can be found in deductivism as a successful scientific method is the nature of the premises on which the entire deductive process is based and the obvious question of how we can actually determine the premises with which we start the deductive process, given the fact that the premises themselves cannot be actually deducted from anything else, but represent the initial phase of the deductive process.

In my opinion, this is not an argument that can make us deviate from the support of deductivism as a scientific method. The facts that a deductive mechanism starts with are evidentiary facts, facts on which one can agree on in an uncontroversial manner and facts that can constitute reasonable and valid premises of our deductive mechanism.