Andrew Easterbrook

The flaw in my previous post is my assumption that in order for us to make a legitimate judgement on irrationality, we have to be able to say all irrational decisions are bad. This is wrong. In order for us to generalise in such a way, it is just necessary only to assert that because most irrational decisions are bad, the irrational decision [x] is bad. More loosely, we are justified in saying “it’s highly likely that your irrational decision was bad.”

Either way, we are using a form of inductive reasoning. Hume thinks inductive reasoning is invalid, since it rests on the principle of uniformity:

Since that principle cannot justify inductive reasoning deductively (it’s not logically necessary, the antecedent is logically consistent with the denial of the consequent) nor inductively (that would be circular) we are not justified in using inductive reasoning. In other words – inductive reasoning proceeds on the assumption of uniformity of nature, but uniformity of nature is not a necessity, so we need to justify it by some other means. But if we justify the uniformity of nature on inductive principles we are being circular! And therefore the statements:

1. All x’s observed so far are y’s
2. w is an x

Do not justify the conclusion that w is a y.

There is an added layer of complexity with the irrationality issue, since we know that not all irrational decisions ought to have been different (and therefore have less reason to suppose that the next irrational decision will be bad – but we do still have reason to suppose). For the sake of simplicity, lets assume that all irrational decisions are bad in the meantime, and worry about the probability aspect of the question at the end.

The problem is not solved by simply saying that we really mean “it’s probable that w is a y” – Hume’s reply is that probabilistic connections, no less than simple causal connections, depend upon habits of the mind and are not to be found in our experience of the world. That is, saying it’s probable that w is a y still runs into the same problems, since that conclusion (whatever the probability arrived at) is also based on the principle of uniformity.

There’s a huge amount of literature on the problem of induction and I can’t hope to even simplify the main schools of thought in one blog post. I’ll simply state my view, which in some ways is similar to Karl Popper’s ideas on falsification. Induction is justified in giving probabilities not certainties (even in seemingly certain cases like the sun rising each morning, it’s more appropriate to say there’s a 99.9999999% chance it will rise tomorrow), and any evidence that suggests that our inductive conclusion is wrong should be taken to invalidate it or at the very least cause us to lower our estimated probability of the conclusion.

Applying this to the rationality problem, I think we have to conclude that there are probably certain areas in which an inductive conclusion that irrationality is bad is a good inductive inference (for example, when talking about non-emotional, non-controversial topics). But when one starts to delve into areas that involve high degrees of emotion, or other areas of thought where we know permissible irrationality could occur, I think we need to revise our inductive conclusion that the irrational decision is bad.

Conclusion: go ahead and say “your decision was irrational, so you should revise it” but only if you know you’re on solid ground. Solid ground includes, but is not limited to, ground that’s not clouded by emotion, instinct or gut feelings.