Decision making under humility

Different epistemic attitudes and how they influence our decision-making

Photo by Burst on Unsplash

This year, I have written a few articles that have focused on the practical role of knowledge in our lives. The dominant theme was that we regularly need reliable information that we do not worry about to make decisions and to act - and we call that reliable information knowledge. However, this might seem to cut against the main theme of this Substack. Remaining humble about what we know presumably makes it harder to act. If we're not sure our information is reliable, then decisions will take more effort - and undermine the practical benefits of knowledge.

While this is a valid point, it isn’t yet a clear argument against epistemic humility. We need to explore the practical consequences of all the different epistemic attitudes, rather than just highlight a potential issue with one of them.

In this post we will do just that - working through the four core epistemic attitudes I have written about before. Before we get started, it is important to note that these aren't focused on any particular fact or claim. The point isn’t to analyse our confidence about specific claims but understand our core attitudes towards knowledge in general - what do we think knowledge is like and when does something count as knowledge?

With that in mind, we can revisit definitions for the four basic attitudes. The first is epistemic confidence, namely that humans can acquire knowledge, and wide ranging knowledge, with a high degree of certainty about its truth. At its most confident, this can drift into epistemic certainty, the conviction that humans have achieved genuine and reliable knowledge and no-one can reasonably doubt it. In contrast is the attitude of epistemic humility, that genuine knowledge is difficult to acquire for everyone, even with the best of care and attention. A bolder, but related, conviction is epistemic skepticism: the belief that knowledge is not achievable and we genuinely know nothing (or very little).

Each of these attitudes generates criteria for when something has sufficient reliability for us to count it as knowledge. At one end, when we adopt epistemic certainty, if our reasons for accepting something fall short of generating certainty, we cannot really count it as knowledge. At the other end, epistemic skeptics don't need to worry about when something counts as knowledge, as there is none, but need to find some other way of creating enough reliability to be able to make decisions and act.

The intersection between knowledge and decisions lies in our common need for reliable information - knowledge - to be able to make a decision. The different epistemic attitudes interplay with this decision making requirement in interesting ways.

Skepticism

At first glance, a genuine epistemic skepticism completely undermines our ability to make these decisions. It assumes that we can never have knowledge and therefore can never achieve real reliability. Yet people still have to make decisions, especially as not making a decision is itself often a decision. And if the facts don't matter - as there are no facts - we have to find some other criteria to use.

What other criteria gets adopted can vary wildly. It might be pure personal preference - there are no facts so I can choose whatever I want. Or it might be whatever is useful to achieve group or personal gain - we'll choose what suits our interests as there are no facts to stand in the way. More commonly, people will adopt a moral or ethical framework for decision making to replace the role that reliable information or knowledge might play. Some strands of social justice approaches or some religious philosophies are examples of this - we know what the right thing to do is regardless of the facts.

However, while epistemic skepticism is often attractive intellectually and can work in some situations, it falls over in most practical contexts. To pick one example, I cannot maintain there is no genuine knowledge about which foods are healthy and which are poisonous and expect to survive very long.

Certainty

Let's jump to the other extreme and consider what happens with decision making if we accept an attitude of epistemic certainty. In some situations, it makes decisions very easy - whenever we are certain about the relevant knowledge. There is no room for doubt and the right decision is almost always clear. However, we face many decisions where we don't have certainty about the information involved and then things start coming apart.

As noted, if we believe that humans have achieved reliable knowledge that no-one can reasonably doubt - if we have an attitude of epistemic certainty - we cannot count information that we are uncertain about as knowledge. But if our expected standard is certainty, how do we make decisions when we aren't certain?

We generally have two options. One is to respond by taking extra time and effort to chase the data or information we need to achieve certainty. But this takes time, often a lot of time, and certainty often isn't possible. So maintaining certainty as our standard easily leads to slow decisions that are too late, or paralysis and no decision at all. The second option is to manufacture some kind of unjustified certainty - through selective reading of the evidence, confirmation biases, or even twisting data. Creating a false certainty allows us to make a decision and act, but it means that the information we are acting on isn't reliable like we think it is. This, in turn, points to a further problem that arises if we assume epistemic certainty is the standard.

Sometimes people make bad decisions, and having flawed information or knowledge as an input is a common cause. However, if certainty is my guiding standard, it is often then really difficult to admit later that I got the decision wrong and I'm more likely to stick with the decision despite things going badly. It's hard to learn from a mistake if you felt like you had to be certain about the knowledge and decision in the first place.

One interesting point is that these practical flaws are common in government and business decision making - and you can likely think of examples. This analysis points to one reason: we often have a misplaced expectation that epistemic certainty is both possible and the standard.

Confidence

Adopting an attitude of epistemic confidence avoids the extremes of epistemic certainty. It still maintains that we can find knowledge that we can be certain about, but that isn't always our situation: there are fields and issues for which certainty isn't achievable. Nevertheless, this attitude maintains a high confidence in our human abilities and processes as reliable pathways to uncovering knowledge.

Again, an attitude of epistemic confidence makes decision making easy when we are highly certain about the relevant knowledge. We don't need to second guess or worry about the reliability of the knowledge. But, also again, it can lead to problems when we don't have confidence in our knowledge. As in the case of epistemic certainty, the common responses are the same - we delay to find confidence, or manufacture unjustified confidence. And again it's hard to change your mind later if the confidence was unfounded.

More generally, this high confidence in our human abilities often leads to us readily accept the claims of anyone we consider to be a competent epistemic authority. We are less likely to double check, compare with multiple sources or think it through for ourselves. This can mean we become more susceptible to the various cognitive biases.

Humility

Adopting an attitude of epistemic humility is not to say that we don't really know anything or know very little. Instead it holds that we often do know things with sufficient reliability, but getting there isn't easy and we sometimes get these things wrong. Moreover, getting it wrong doesn't mean that is some inherent flaw in any of us, just that knowledge is hard.

On this attitude, when we are confident in any particular piece of knowledge, then making a decision isn't much different to if we assume confidence or certainty. However, that happens less often and it changes our approach after the decision. If our confidence turns out to be misplaced, then that won't be a big surprise. This means it is much easier to acknowledge the mistake and change course.

Under conditions of uncertainty, however, epistemic humility is highly practical and beneficial. If we accept that knowledge is hard, then it will not surprise us that the information we have for a decision isn't clear. And because we don't think that certainty, or even confidence, is the norm, then we will be more comfortable making a decision based on partial and uncertain information. A practical focus means that we have to make the best decision that we can now, but we are humble enough to know that we might need to change our minds or our direction if or when new information comes to light. We can be freed up to act now if we don't expect to have confidence or certainty in our information in order to make a good decision.

We hear common expressions of this practical humility in business and governance jargon or slogans. Words like agile, adaptable, or responsive and phrases like 'fail fast', 'alpha / beta testing', 'build and iterate' all encourage us to act now based on what we know and learn as we go. The point that is often unacknowledged is that these practices only make sense if we are humble about we know. When we accept that the world might not work like we think it does, all of these approaches help us iteratively calibrate our knowledge to reality - the same dynamic that good scientific practice has adopted for centuries.

When these approaches fail in government or large companies, and they often do, it is often because incentives and power structures demand epistemic certainty. A Prime Minister is expected to exhibit certainty about what they know and what they will do, and will get roundly criticised if they change their mind. We expect epistemic certainty, but don't realise how this changes our incentives and decision-making structures.

Humility for decision making

This very quick exploration of the topic has shown how each of the epistemic attitudes gives rise to distinct decision-making approaches, many of which you will probably recognise from experience. Each makes decisions in some contexts easier and harder in others.

What is notable about the attitude of epistemic humility is that while it somewhat complicates the easier decisions where we are confident about what we know, it also frees us up to act more decisively in the messier situations. Even if we ignore all the other arguments for epistemic humility, this is a trade-off that has a lot of practical benefits for decision-making. It is in the ambiguous or uncertain situations that we normally need help making decisions, and epistemic humility makes that easier.