Why does Hume say what he says about Induction?
The analytic system of induction is that by which one can predict occurrences in the universe around us. It is the method through which a great deal of our knowledge is found, yet is the topic of controversy in philosophical circles, due to work done by David Hume. The 'received view' is that Hume was sceptical about induction - that he believed that conclusion through induction was unreasonable. Recently, however, a different thesis has been put forward claiming that Hume was not sceptical, he was simply pointing out the weaknesses of induction. In order to examine this question, this 'received view' shall be looked at first, followed by an explanation of the new view. After these two opposing views on Hume have been compared, inductive method shall been studied, to see if these views on induction can be better clarified or improved upon.
The 'received view' of Hume's opinions on induction, put most simply, is that Hume was a sceptic about induction; that he believed that to base knowledge on inductive method was unreasonable. Proponents of this view see Hume's argument as follows: All factual belief is based solely on instinct, therefore factual belief is irrational. Since inductively derived belief is a subset of factual belief, then inductive conclusions are also irrational. Put another way, the argument for Hume's scepticism is this: The entire institution of inductive reasoning can not be justified, so no inductive conclusion can be justified. This argument is arrived at by interpreting Hume's passages on induction as a critique, rather than simply an investigation. Hume does indeed say that induction has no logical necessity, due to the fact that the relation between cause and effect seems to be one of 'imagination'. Supporters of this 'received view' claim that this means that Hume believes that a conclusion reached by induction is not a rational one - that Hume is an irrationalist.
More recently there has been a move towards reconciling Hume and inductive method, mainly to save a great deal of Hume's work with inductive method. Opponents of the 'received view' of Hume on induction ask why Hume would undermine inductive method, when it is inductive method that he develops and uses all through his philosophy. In order to save induction, opponents of the 'received view' attack the premises that the 'received view' imposes on Hume. These are such that Hume tries to find 'external justification' for induction - a non-circular argument for the rationality of induction - and fails, thus confining induction to the category of irrational method of conclusion. The opponents of these premises point out that Hume is not trying to justify induction outright, and therefore justifying the everyday use of induction. They argue that Hume simply shows that 'demonstrative reason' does not prove matters of fact, and that induction has no logical necessity. What these statements rest upon is basically Hume's idea of causation, upon which his ideas of induction lie. Induction is used to prove matters of fact, yet induction is not a 'demonstrative proof'. Neither can it be shown that the process of induction can be justified logically. This is the idea that the opponents of the received view wish to be understood as Hume's view. Beauchamp and Mappes use many quotes to very convincingly argue that this, not the 'received view', is the case.
At this point, we shall see what can be gleaned from an inquiry into induction itself. The process of induction is one in which a conclusion is based on consistency in previous experience. For example, if every time I have let go of my pen, it has dropped, I come to the inductive conclusion that the next time I let go of my pen, it will drop. This method of conclusion is quite different from deduction, which is a far stronger form of reason. To come to the same conclusion by deduction, I need to have the premise 'Pen's always drop when released', and if and only if this premise is valid can the conclusion be deductive, rather than inductive. Another way of looking at the difference between induction and deduction is that deduction deals with validity and invalidity, while induction deals with belief. This point makes the problem of induction clearer - is it possible to somehow show inductive conclusions to be logically valid by deductive standards.
In order to attempt to do this - justify induction in deductive terms - two paths could be followed. The first follows on from the point made before by the fact that an extra premise is needed to make induced conclusion valid in terms of deduction. Also, this premise must be seen to be valid in order for the justification to hold. In order, therefore, to justify induction in general, a 'Supreme Premise' is required that would make inductive conclusions valid. One form this premise could take is that of 'for all x and all y, whenever n cases of x and y occur, and there are no cases of x and ¬y, then all cases of x are cases of y'. This would certainly make inductive conclusions deductively valid, yet this is not an attractive solution. The reason for this is that the number n will be arbitrary - no matter what number we choose for n, there is no logical reason for having that number, and there is no specific number that guarantees that the perceived inductive relation between two objects is true. A different form of the 'Supreme Premise' could be 'Nature is Uniform'. This form of premise is far too vague for the purposes of guaranteeing induced conclusions are valid in everyday existence. Indeed, it is the use of such premises that is conducive to long-term errors in science - the assumption that nature is uniform manifests itself in the assumption that our perception of nature is uniform, which leads to the persistence of scientific errors, and the persecution of heresy. Although this type of 'Supreme Premise' is more promising in allowing induced conclusions to have status of being deductively valid, it is still insufficient.
Another way in which an attempt to raise induction to the level of deduction could be made is through mathematics and probabilities. For example, if you have a bag of marbles, and the marbles are of to colours, black and white, and the only way to find out the proportion of marbles is to remove them one by one, then as more marble are removed, probability dictates that the sample you hold outside of the bag is likely to reflect the proportion of black to white inside the bag. This way, as more instances of connection between two objects or events are witnessed, then it is logically more likely that the inductive conclusion from those repeated connections is valid. This model, however, can never be sufficient to provide logical proof for induced conclusions. In the mathematical example, exact knowledge of the contents of the bag can not be known until every marble has been removed. In the real world, however, there are infinite chances for the connection to occur, and so the model must have bag of an infinite number of marbles. If the bag has an infinite number of marbles, then the model becomes a less accurate representation of inductive method. If one unexpected marble is pulled from the bag of infinite marbles, then it does not necessarily disprove a proportion, is simply makes it more accurate. In induction, if one unsuspected event occurs, then the initial conclusion is rendered invalid. Again, this has difficulties in fully representing induction, and it has prove inadequate in providing validity to induced conclusions.
One question that is raised by this attempt to justify induction in terms of deduction, is why is this necessary? Induction by definition is not deduction - it is the method we employ in everyday existence to find explanations of the objects and events around us. If these objects and events could be explained through deduction alone, then induction would not be required. Due, however, to the unknowable nature of the objects around us, induction proves necessary. This is why is was stated earlier that induction deals only with beliefs, while deduction deals with validity. If this distinction is made clear in the use language, then the question of justification of induction becomes unnecessary. An important point, however, is how do we determine which induced conclusions are justified and which are not. Due to the nature of induction, the best way this is done is to say that if the criteria by which we predict an event have been totally accurate up to that point, then inductively, they are justified.
This last point about how to justify induced conclusions is at the root of Hume's views on induction. It would seem that the 'received view' of Hume on induction is that he is sceptical about induction because it is impossible to justify through deduction the use of induction. This has been seen to be a pointless mission, and the opponents to the 'received view' believe Hume to be innocent of such frivolity. The opinions of Hume laid out by the opponents of this 'received view' are more consistent with the concept of induction as it is seen today - that as it only deals with belief, then it is impossible to logically justify it, yet it is the only way we can attempt to arrive at matters of fact. This seems to be the most sensible way of dealing with the problem of induction, and the opponents of the 'received view' of Hume's opinions on induction have made it possible to reconcile Hume and induction.