Over two hundred years ago, David Hume spotted a major problem with the whole method of reasoning by induction. Deductive reasoning begins from some general rule that we trust totally. We then apply the rule to a problem in front of us. Inductive reasoning begins by looking at a whole lot of examples of a problem and trying to find a common pattern among them. Then we formulate a rule about how these kinds of situations turn out, a rule based on the pattern we think we have spotted. Then we test the rule in new situations, over and over, until we really sharpen and deepen our understanding of it. Inductive reasoning is the reasoning method that empiricism and science depend on.³

However, Hume said that if we draw generalizations from our masses of experience of the real world, no matter how careful we are in how we observe reality, formulate our generalizations, or conduct research to further test, refine, and bolster these generalizations, we still can’t say with certainty that any of them is true. To do so would be to posit that the events of the future will be like the events of the past. We can’t make that larger claim because we haven’t been to the future.

Bayesianism slips out of the problem of induction. It simply says we are always gambling, checking the generalizations that inform our gambling—even our most basic ones, the ones we need in order to see reality and form generalizations at all—against real-world data constantly. By choosing to live in this state of permanent tentativeness, we are gambling on alert, rational gambling as being our best gamble.

But we aren’t putting all our eggs in the single basket of any one model of any part of reality. Rationalists end in doing that, as they attempt to reason their way from sets of concepts they say they just know to premises they won’t question to policies they won’t analyze, no matter how ineffective or destructive the consequences of those policies may appear from the empirical evidence to be.

With Bayesianism, we also don’t get stuck like the empiricists, stopping in a stymied funk in our progress toward a kinder, wiser world, which is what happens if we keep staring at the problem of induction and refuse to get on with life until that problem is solved. It isn’t going to be solved. Bayesianism gives us a viable way out.

Thus, we can get on with it—the task of formulating a moral code based on our best current models of the real world. In the coming chapters, the Bayesian view of the human mind, combined with two of the most basic ideas in physics and a model of cultural evolution, will enable us to build a modern moral system. And then, finally, we may be able to make a case for theism, a belief in the existence of God.

Notes

1. Douglas R. Hofstadter, I Am a Strange Loop, (New York, NY: Basic Books, 2007).

2. Plato, The Phaedrus, Perseus Digital Library. Accessed April 17, 2015. 

http://www.perseus.tufts.edu/hopper/text?doc=plat.+phaedrus+265e.

3. John Vickers, “The Problem of Induction,” Stanford Encyclopedia of Philosophy, March 14, 2014. http://plato.stanford.edu/entries/induction-problem/.