Sunday, November 22, 2009


Will the Sun Rise Tomorrow?

How do we know that the sun will rise tomorrow?

Bertrand Russell, in Chapter VI of his 1912 book The Problems of Philosophy, asks this question of us, as a way of introducing what philosophy has called “the problem of induction”.

No one has yet seen the sun rise tomorrow. However, we suppose that it will, based on what we have always seen in the past. Predicting what will happen with things we haven’t witnessed on the basis of things we have is called “induction”.

As likely as it seems that the sun will rise in the morning, what we know of astrophysics tells us that a comet may hit the earth tonight and knock it off its orbit before our first cup of coffee on the morrow. We cannot know the future as concerns the sun, nor indeed can we conclusively know anything about anything else for which we have no already existing empirical evidence.

It is extremely valuable to always be aware of this unfortunate reality, for it reminds us that even the most rigorous science is fallible. In effect, induction is a kind of informed guesswork, and we are wise to approach all claims based on induction with a measure of skepticism.

Is guesswork a sufficient proof of anything? Can we use it as a guide to our actions in the future? This is the problem of induction.

Saying that we can never be absolutely certain is a useful observation; to say that we can never be reasonably sure is not.

Russell does not explicitly state that one can never use past experience as a guide to future experience; only that to do so one must use what Russell calls, rather grandly, the “principle of induction” This principle is best summarized by Russell himself. We cannot expect certainty about that which we have not yet directly observed; therefore “…probability is all we ought to seek.” (Russell, p. 66)

Human beings are active, agentic creatures, pushed and pulled about by all manner of internal and external motivations, who seek knowledge not for its own sake, but because knowledge serves as a guide to action. Knowledge is a tool a person uses to accomplish a task. Knowledge must be useful and usable to qualify as knowledge. Knowledge which has no use is, by definition, useless; we often describe it as “useless information”.

This formulation can help us distinguish between knowledge and information. Knowledge has an immediately discernable use-value. Information that has no recognizable use-value is not regarded as genuine knowledge, and our brains don’t for the most part retain such information.

Human knowledge therefore relies on induction, for if knowledge is something to be used, it must be reasoned through induction that it can be used in some unknown and unreliable future. When we walk from point A to point B by putting one foot in front of the other, we induce that since our last step found solid ground, so our next step will fall on solid ground. Of course in our millions of steps there come times when this innate knowledge proves fallible.

In order to determine the usefulness of anything, we must use our powers of induction. We have had experience of this cause and that effect; we predict that a similar cause in similar circumstances will produce a similar effect.

Thus the concepts of inference, prediction, and induction inhere in the concept of knowledge. Knowledge is a hypothesis about the unexamined on the basis of the examined. The previously examined is no more than information, unless one can use it to predict the future. The sun rose yesterday. So what? A conclusion from induction – say, that the sun will also rise tomorrow – is a hypothesis on which one can base useful action, however certain we are that such knowledge will ultimately prove unreliable. Anyone who claimed to know no more about the sun than that it rose yesterday cannot claim to truly have knowledge of the sun.

This important point needs to be restated. Knowledge is inseparable from the intention of the knower. Knowledge is not a dry thing that lives on a shelf in a library. Knowledge is inseparable from the act.

Induction differs from deduction simply in that the conclusions of an inductive argument are never certain. At best, induction can point to a reasonable probability, such the probability that the sun will rise in the morning. Deduction can point to certainty, but only through a sort of sleight of hand.

Here is an example of a deductive argument:

1. All mammals suckle their young.

2. Foxes are mammals

3. Therefore, foxes suckle their young.

On the face of it, this seems a bulletproof argument, until we remember that the whole thing hinges on our definition of “mammal”. Part of what defines an animal as a mammal is that it suckles its young. Thus a deductive argument brings nothing new to the table – it merely allows us to make something more obvious.

It is only induction that allows us to make the great leaps in knowledge of which humanity is so rightly proud. Our intelligence as a species is really our powers of induction. Intelligence can be regarded as the ability to predict the future from the past – and Jeff Hawkins does so in his book, On Intelligence.

Knowledge is a guide to action based on probability. All projections for the future are of course based on an analysis of the past. All knowledge is therefore empirical. Probability is in fact an empirical principle. It has been repeatedly proven on the basis of empirical evidence that given a sufficient sampling it is possible to predict, with a reasonable degree of certainty, that the future will resemble the past. Of course, one needs a large enough sampling, and one needs to allow for a greater or lesser margin of error. Saying that probability theory is based on empirical research does not mean its predictions are deductively conclusive or constitute certain proof.

The best probability theory can offer is a prediction that given certain circumstances a given course of events will occur within a range of frequency – most of the time. And in this topsy-turvy world, perhaps, in the words of Russell, “…probability is all we ought to seek.”

No comments: