Induction Discussions Across the Centuries:

Induction

Discussions Across the Centuries: David Hum, Paul Edwards, and the Ongoing problem of Induction

Though the pages of a philosophy textbook might seem full of dusty long-dead men, the logical progressions and conclusions outlined by these men do not occur in isolation, and philosophy as a body of knowledge is far from a series of isolated and independent conjectures. Instead, philosophy operates as a dialogue spanning the millennia of human investigation into problems of knowledge, reality, and purpose, with the arguments and conclusions of philosophers long past still serving as points of departure for more contemporary thinkers. Some problems continue to evolve as this dialogue progresses, with agreements found and foundational arguments generally accepted and built upon, yet for certain other issues the discussion remains at early stages for centuries at a time.

The problem of induction is one of the latter type of philosophical problems in its fundamental lack of progress despite centuries of discussion. This is not to say that arguments in this area have not led to more sophisticated and detailed examinations of the problem as some of philosophy's best and brightest have turned their attention to the issue, but ultimately no real progress has been made in developing a solution to the induction dilemma that has been generally accepted or not strongly refuted or limited by the same fundamental problems originally identified. This could lead to the obvious conclusion, of course, that the problem might not have a solution, however this is not something that can be concluded lightly.

Two arguments spanning much of the period during which the problem of induction has been a focus of certain philosophers will be detailed and examined below. David Hume's identification of the problem, which is considered by many to be the first true codification and definition of the problem of induction (though Hume never used that specific term), occurs in his an Enquiry on Human Understanding, first appearing in the eighteenth century. Paul Edwards, a twentieth century philosopher, dealt with the problem explicitly by addressing Bertrand Russell's own definition of the problem, however his argument finds a dialogue with Hume's description of the problem just as directly. Through an examination and comparison of Hume and Edwards' arguments on the problem of induction, it can be seen that although the problem can be sidestepped it still cannot be solved.

Hume's Definition of the Problem of Induction

Hume distinguishes between two kinds of true knowledge: relations of ideas and matters of fact. Relations of ideas are logically coherent and self-evident in and of themselves, such as the statement "all apples are pieces of fruit;" apples are by their very definition pieces of fruit, and therefore any contradiction of this statement can be demonstrated to be logically inconsistent without appeal to any further observations, relations, or facts. Matters of fact are essentially things that are directly observable, such as "the shower is on." The truth of this statement can be determined through simple and direct observation; either the shower is on or it isn't, so the statement can be verified or falsified not through logic but through the senses.

If these are the forms of knowledge, then, there are many things we typically assume and might say we "know" that are in fact mere guesses that are not truly backed by fully rigorous logical or observational evidence. Hume insists that we cannot actually know things about future events no matter how many matters of fact we have established with previous observations. A statement such as "the sun will rise tomorrow" cannot be said to be a true statement of knowledge as it is not a logically self-evident relation of ideas -- the sun is not by definition "that which rises tomorrow," no matter how many times it has risen on previous tomorrows -- and nothing that happens on today's tomorrow could ever be a matter of fact, as the future is not directly observable. Anything that is thought to be "known" about future or hypothetical events is actually, according to Hume, and assumption made through inductive reasoning that cannot actually be justified in a logically certain manner.

This is the problem of induction in a nutshell, and it is something that has alternatively been seen as one of the most severe limitations on true knowledge about the world or as a non-issue in any practical terms. If inductive reasoning cannot be trusted, then all past experience and even experimental data is essentially meaningless in predicting the future and there is no logical reason to assume things should occur one way simply because they have occurred that way before. Many have pointed out how silly it would be to go through the world without inductive reasoning -- not being sure if the door would open when the handle is turned, etc. -- but this does not actually address the logical problem of induction.

Edwards Attempted Answer

There have been attempts to address the problem of induction at the fundamental logical level, some of them seeming to come closer to success than others, but ultimately no one has been able to definitively demonstrate in a logically consistent manner (i.e., without a reliance on inductive reasoning) that inductive reasoning is a valid source of knowledge. One attempt, and a very good attempt, is made by Paul Edwards in his answer to Bertrand Russell's restatement of the problem of induction, and this answer seems to combine both practical and logical arguments against the supposed problem of induction. In response to the basic question, "if a has always led to b in the past, is there a reason to believe that this will be the case in the future?" Edwards attempts to frame this as an issue of semantics, asserting that such a question is to misuse the word "reason," ascribing a much higher-order and restrictive meaning to it than common sense and usage would imply.

Edwards essentially acknowledges that there is no way to supply a reason for believing that a will lead to b as long as "reason" has the definition of an observable fact or something on the level of a relation of ideas -- such a reason inherently cannot exist in a situation that cannot be directly and immediately observed. The context in which the question is posed, according to Edwards, inherently defines "reason" as something that cannot exist for belief in future events, and therefore makes it necessarily true that the answer to the question posed above is "no." The problem of induction, then, is disingenuous.