I am becoming increasingly convinced that the folk saying “I’ll believe it when I see it” is wrong, at least some of the time. Or at least, people have a remarkably predictable ability to ignore seemingly convincing evidence in certain situations, when that evidence appears to contradict closely held beliefs. I’ve started calling this the “I’ll see it when I believe it” phenomenon.

Two of the major areas where I see this pattern have also been identified by Paul Graham as areas where people tend to fight pointlessly¹, and I think our observations are consistent, although I do not agree with Graham’s conclusion on this issue.

To rephrase Graham’s claim, Graham believes that so-called ‘religious wars’ are the result of conflicting identities. Because an argument about one’s identity may potentially place that identity at risk, people have a tendency to attack anything which has the appearance of disagreeing with their identities. Graham claims that politics and religion are areas which tend to be incorporated into people’s identities, making it difficult or impossible to have an intelligent discussion which questions anyone’s political or religious assumptions. Furthermore, Graham theorizes that having an identity which includes a large number of things actually decreases one’s intelligence by reducing the number of topics around which one can have an intelligent discussion. Therefore, in order to increase one’s intelligence, one should strive to minimize one’s own identity.

Graham’s claim is problematic because if true, it would essentially make it impossible have an intelligent discussion on this topic. Obviously I have an identity, and since an argument which suggests minimizing my identity is in fact a threat toward said identity, I cannot possibly allow myself to comprehend the full implications of Graham’s statement. Or can I? There must certainly be plenty of counter-examples where intelligent people have held rational discussions on topics which have been critical parts of their identities. While this may not be the norm online, it is a sufficiently important case that failing to explain it would be a fatal flaw.

I suggest that rephrasing Graham’s statement in terms of ‘world views’ rather than ‘identities’ may help enlighten the debate. I also propose that rather than asking about the case where discussions devolve into irrational arguments, we should consider the subtle case where rational arguments are presented on both sides, and each is internally consistent, but they fail to be consistent with each other.

For the purposes of this discussion, I define a world view as being the combination of all internal facts which a rational agent knows about the world. The world view is a function of perception, as it is derived by analyzing observations about the world, constructing internal models to predict the world, and then comparing the results of those models against future observations. When a world view closely reflects one’s physical reality, the correlation between predictions and observations is high. Otherwise, the predictions and observations may fail to agree, sometimes miserably.

It is important to realize that the essence of what we call ‘understanding’ is in the ability to ‘predict now’. In other words, if one’s predictions about what should be happening right now line up with the inputs one is receiving, then one can be said to understand the situation. Otherwise, the internal model is obviously faulty, and one cannot be said to have an understanding of the situation. Note that under such a definition, understanding is not literally the act of precisely simulating the world, because these methods are usually approximations or heuristics. Humans are, after all, not machines, and have little or no affinity for those sorts of computations. Computer Science has done a good job of showing exactly how hard computing those sorts of properties can be in a formal system. Accepting that understanding is a heuristic process based on recursively refined predictions derived from an internal data set, we move on to differences between our world view approach and Graham’s identities.

The critical difference between Graham’s identities and the world views proposal is that world views are inescapable, and smaller is not better. Indeed, having a smaller world view limits what one can predict, and therefore understand. Furthermore, having a wide world view will be important to being able to understand the actions and arguments of others. Having a limited world view will reduce the percentage of intelligent discussions in which one can engage.

Let us turn for a moment to the proposed case study for evaluating this hypothesis. It is truly a remarkable happening when two highly intelligent people have a rational, reasoned discussion, where novel information is presented on either side, and yet neither side appears to recognize the implications of the evidence presented. And yet, I claim that this happens all the time, especially on the topics of politics and religion.

Let me stress how strange this should seem. We are all confronted, on a nearly continuous basis, with novel information. To an extent this has been accelerated by the internet, but even before the information age, there have been printed publications, and of course, face-to-face discussions. A great deal of this information gets absorbed quickly into our world view (at least among people who consider themselves to be intelligent). And this allows us to execute quickly on this new information, and make rational arguments about recent events. (Again, let me stress that this is not an easy process. This information can frequently be game changing in completely new ways, which can fundamentally change our lives. But intelligent people usually seem to have little difficulty absorbing this sort of new information.)

There are also times when information is ignored quickly. Usually this is the case when frivolous or nonsensical arguments are made. These can be quickly dismissed, because they present little of interest to our insatiably curious minds. And there is nothing surprising about this rapid dismissal. But it should still seem odd when a rational and internally consistent argument is discarded without any apparent consideration.

I hypothesize that these cases are the result of a miscategorization of information. Remember that information is filtered by the listener, and information which seems irrelevant or uninteresting or which simply doesn’t fit well into the listener’s world view may be discarded without further consideration. How quickly this information is discarded is dependent on how far outside the listener’s world view the information lies. So information which is in fact rational and internally consistent may be discarded before receiving full consideration if it is unpleasant in one or more of the ways above. This could be the case, for example, if the speaker argued a political view with which the listener had had a previous bad experience. Further attempts to explain that political view might be discarded more quickly than the original. If so, then the listener could form a cycle where negative internal feedback on that political issue would cause the listener to continue rejecting arguments from that political view with increasing confidence (this is called a bias). This negative feedback cycle explains how a person, having decided that an argument is invalid at one point in time, may continue to assume that new arguments remain invalid even if the content of those arguments changes.

But how then does an individual accumulate the initial bias which causes this feedback cycle? As I stated previously, world views are inescapable. Every one of us is imparted with social knowledge which we accumulate as we grow up. This knowledge comes from our parents, our teachers, our peers, and everyone around us, and is frequently referred to as ‘cultural knowledge’, or in post-modern terms, the ‘master narrative’. It is important to remember that this knowledge is still individual, but for the most part, the initial knowledge sets will agree among individuals who grew up in similar conditions. It is this initial set of facts which we believe to be true which sets us up to reject the ideas of others who have grown up differently. By this mechanism, the cycle is set in motion and we are prepared to participate in political flame wars when we come of age.

Let us finally return to the original premise. The ‘believe what you see’ saying comes from the recognition that our internal world view, and thus our belief system, is derived from our observations of the world around us. However, the saying fails to explain that this is a cyclic process, where our world view influences how we interpret future observations, and may ignore some observations entirely if they don’t fit well into our existing world view. The ‘see what you believe’ hypothesis attempts to explain this other side of the coin.

What is there to do for those people who are interested in having discussions about historically contentious topics? I agree with Graham here that one may not be able to intelligently discuss these topics with everyone one meets. However, having framed this as an understanding problem and not an identity problem, I believe that the critical piece is to find people with an open mind. And keeping an open mind is fundamentally an issue of how well one can incorporate unexpected observations into one’s world view where doing so would normally be difficult. This sort of mental juggling is not easy, and can be easily as harmful as it can be helpful, as in the case of the gullible person who can be manipulated easily. However, maintaining a balance is essential to one’s ability to absorb new ideas. Therefore, as a heuristic, I suggest that intelligence may be a good indicator for open mindedness. ‘Wait,’ you may say, ‘I know intelligent people who have trouble absorbing new information.’ If so, I would say to you that those people would be even more intelligent if they were better able to absorb new information.