Biases That Contribute to the Neglect of the Long Term and of S-Risks

Original text Luciano Carlos Cunha · Volume X, p. 37 of the PDF edition · English translation

3.1. Introduction

In this text we will address biases that may lead us to give little importance to ethical issues related to the long term and also to the long-term effects of our decisions, which also affects the planning of activist strategies, whatever the cause.

We will also look at the biases that may lead even people who care about issues related to the long term to neglect, specifically, the risks that in the future suffering will be multiplied to astronomical levels (known as s-risks), making the suffering that exists today "a grain of sand" by comparison.

The presence of these biases may explain why even people who care about the long-term future tend to focus on x-risks (the risks of human extinction) and give little or no importance to s-risks.

3.2. Biases that contribute to the neglect of issues related to the future

The following biases incline us to neglect issues related to the future.

Temporal bias. This consists in giving less weight to an event the further in the future it lies. Underlying temporal bias may be egoism, leading us to give less weight to what is distant from us (it is as if we thought the present era is more important because we are in it).

Availability heuristic. This consists in thinking that the cases we remember most easily are necessarily the most representative. This bias also probably plays a role in the neglect of issues related to the future, since the concrete cases we remember obviously do not concern the future. Thus, for example, when people think about which situations contain the largest number of individuals their decisions could affect, what usually comes to mind are situations in the present moment.

Bandwagon effect. Since most people care only about (or give greater weight to) the short term, we tend to do the same. This is facilitated by conservatism bias, which inclines us to resist abandoning our prior beliefs.

Duration neglect. This bias consists in the tendency to neglect the duration of an event when assessing how good or bad it is⁴². The long-term future may be vastly long (and therefore contain an astronomically greater number of sentient beings). Duration neglect inclines us not to take this into account when we assess the importance of the future.

Present bias. The tendency to choose more immediate returns over returns that will come later in the future, even when the immediate returns are smaller⁴³.

Evaluability bias. This leads us to think that what is easier to evaluate is also more important⁴⁴. Since short-term consequences are easier to estimate, this bias may lead us to think they are more important, or that we will necessarily have a more positive impact if we prioritize the short term. If we do this, we will not be properly calculating the expected value/disvalue of the decision. This will be explained further below.

3.3. Biases that contribute to our failing to take expected value/disvalue into account

The expected value/disvalue of a scenario is obtained by multiplying how good/bad the scenario is by its probability of occurring⁴⁵. The following biases tend to lead us not to consider the expected value/disvalue of long-term events when deciding on priorities and strategies⁴⁶. Some of these biases lead us to neglect probability, while others lead us to neglect how good or bad the scenario is.

Neglect of probability. The tendency to disregard probability when making a decision under uncertainty.

Ambiguity effect. The tendency to avoid options for which the probability of a favorable outcome is unknown (it occurs because of aversion to uncertainty)⁴⁷.

Zero-risk bias. A preference for reducing a small risk to zero over a greater reduction in a larger risk⁴⁸. This bias is also related to a preference for absolute certainty, and inclines us to opt for situations in which we can completely eliminate risk, at the expense of alternatives that might actually offer more benefits.

Scope neglect or scope insensitivity. The tendency not to take the size of a problem into account when assessing it. For example, being willing to pay the same amount to save 2,000 individuals or 20,000 individuals (or even paying more to save fewer).

3.4. Reliance on intuitions

One of the main reasons why biases exert so much force over us is that they are highly intuitive: at first glance they look like evidently correct reasoning, even though they are not. This is why reliance on intuitions is among the main factors that lead us to follow biases.

What we are calling intuitions here are the automatic responses we have to a question before evaluating it critically. In some cases, an intuition may concern describing the facts. For example, people centuries ago had the intuition that the sun revolved around the Earth and that the Earth was flat. In other cases, an intuition may concern evaluating what we ought to do. For example, most people have the intuition that humans deserve greater consideration than nonhuman animals, that harmful acts are more wrong than equally harmful omissions (or even than more harmful omissions), and that we need not be concerned about how our decisions affect the sentient beings who will exist in the distant future.

This does not mean that every intuition is necessarily wrong. It is possible that, after careful analysis, we conclude that we have good reasons to keep a given initial belief.

However, an intuition may also simply be a prejudice that appears correct merely because we have believed it for a long time, because the rest of society accepts it, or because our brain tends to believe it. In short, the fact that a thought arouses in us a strong feeling that it is correct is not, in itself, a reason to think it is correct. Quite the contrary: it is a reason to test it twice as hard, since the very strong feeling that a thought is correct may be one of the reasons why we fail to notice that it is wrong.

The word intuition is sometimes used in a different sense, as a synonym for beliefs that are not derived from other beliefs⁴⁹. For example, imagine that someone asks me why I avoided bumping my elbow on the table, and I answer that it is because I will feel pain if I bump it. Imagine that the person then asks me why I want to avoid feeling pain, and I answer that it is because it is a negative experience. Imagine that the person then asks me why I want to avoid a negative experience, and I answer that it is because of the very fact that it is negative, period. Some people would say that the statement "negative events generate reasons to be avoided" is an intuition, in the sense that it is where justification comes to an end, that is, a statement that is not derived from another.

However, this is a different sense of the word. A belief can be an intuition in this second sense without being one in the first. For example, someone may have tested the belief "negative events generate reasons to be avoided" and continue to hold it because they have found no good reason to reject it (in which case it would no longer be an intuition in the sense of an automatic response that has not been subjected to scrutiny). In what follows, we will always use the word intuition as a synonym for automatic response.

One of the reasons why intuitions are so closely linked to biases is that they usually generate corresponding strong feelings (whether because we have believed something for a long time, because most people also hold it, or because of a biological tendency). On the other hand, conclusions that result from careful examination may not, at first, generate a corresponding strong feeling.

However, with time and practice, it is also possible for conclusions that struck us as counterintuitive to begin generating emotional responses as well (and then our intuitions begin to change). For example, suppose someone concludes that it is unjustifiable to give less weight to nonhuman animals and to the sentient beings who will exist in the future. Imagine that, at first, this conclusion does not generate a feeling of outrage at the neglect of this issue. It is possible that, over years of applying this principle, the emotional response will grow, and what comes to seem strange and to generate outrage is failing to give equal weight to nonhuman animals and to future sentient beings.

3.5. Biases that contribute to the neglect of s-risks

S-risks are risks that the suffering that exists today will be multiplied in the future by many orders of magnitude. All the biases we have seen above can contribute to neglecting concern about s-risks, since they are biases that incline us to neglect issues related to the future.

Curiously, however, discussion of s-risks is relatively rare even among people whose goal is concern for the long-term future. These people tend to focus almost exclusively on x-risks (the risks of human extinction or, at least, of a collapse of civilization). Below are some biases that may specifically contribute to a neglect of s-risks⁵⁰.

3.5.1. Speciesism

Since the overwhelming majority of the individuals who might be harmed if an s-risk were to materialize do not belong to the human species, the tendency to give less weight to the good of individuals who do not belong to the human species (or even to disregard them completely) is probably one of the tendencies that contribute most to the neglect of s-risks.

3.5.2. Optimism bias and pro-innovation bias

Optimism bias is the tendency to be overly optimistic. Someone with this tendency tends to vastly underestimate the probability of undesirable outcomes.

Pro-innovation bias, in turn, is the tendency to be excessively optimistic about the usefulness of an invention or innovation, which prevents one from identifying its limitations and weaknesses.

These two biases together may make many people overconfident about the future, especially regarding the impact of new technologies. For example, they may lead people to strongly neglect the possibility that future technologies will multiply suffering to gigantic levels.

However, we must also take into account that there are biases that can push in the opposite direction, inclining us to overestimate the scale and probability of s-risks. One example is pessimism bias. As noted by Baumann (2022, p. 45,6), most people find it more interesting to think about extreme scenarios than about more ordinary ones (which is why science fiction about dystopian scenarios is so popular). For most people, however, the stronger tendency seems to be to disregard s-risks rather than to be excessively concerned about them.

3.5.3. Substratism

So far, as far as we know, it seems that every sentient being belongs to the animal kingdom. However, this may change in the future. If sentience appears whenever matter is organized in certain configurations, regardless of the material that makes up that matter⁵¹, then perhaps in the future non-organic sentient beings will come to exist (in digital media, for example)⁵² – if in the future such media are configured so as to perform exactly the same functions as a brain, for example. All of this is uncertain, but it is a possibility.

If non-organic sentience is possible, then these beings could come to be very numerous. For example, today it is easy to copy files on computers and on the internet. It may be that in the future it will become easy to produce, for example, simulations of

entire universes in digital media, full of real sentient beings⁵³. These beings, although sentient, may behave very differently from the sentient beings we know today. They may have no faces, and may not scream⁵⁴. For this reason, it may be very difficult to feel empathy for them, with the result that these beings could constantly suffer atrocities like those nonhuman animals suffer today (or perhaps something even worse). For example, nonhuman animals are already much more similar to humans in terms of behavior and expressions, and yet most people do not give them even the slightest consideration.

Thus, one tendency that could significantly increase s-risks in the future is carbonism. Just as anthropocentrism is a form of speciesism against those who do not belong to the human species, carbonism is a form of substratism against those who are not organic.

For this reason, it is extremely important to emphasize that we should consider nonhuman animals because they are sentient, and not because they are animals, because they are natural organisms, because they are living beings, or for any other reason. This will increase the chances that people will come to consider other kinds of sentient beings in the future, should they come to exist.

Another way to understand how unjustifiable carbonism is, is to imagine the following thought experiment: suppose we discovered that we live in a simulation (and that everything around us, including our bodies, is not organic but made of a digital substrate)⁵⁵. Would that mean our own well-being should not matter? Would that mean it should matter less? Clearly not. It should matter to exactly the same extent; we would remain equally capable of being harmed or benefited. If so, then there is no justification for giving less weight to (much less for disregarding) non-organic sentient beings, should they come to exist in the future.

3.5.2. Wishful thinking

Wishful thinking: the tendency to think that what we would like to be true is in fact true.

We would like the future to be wonderful. So thinking that it might turn out to be terrible can be disturbing. For this reason, seeking to prevent s-risks may not be as inspiring as thinking about a wonderful future or as helping individuals alive now. So, as noted by Baumann (2022, p. 42), there may be a tendency, because of this, for people not to care about the risks of future suffering, especially if they themselves are in safe situations.

3.5.3. The collapse of compassion

The collapse of compassion bias means that the more victims a situation contains, the less importance we give it. We may recognize that a large-scale catastrophe is much worse, but in terms of feelings, we tend not to grasp how bad it is in proportion to its size and, consequently, we tend not to give the problem in question the importance it deserves.

Since the issue of s-risks is, of all issues, the one that would potentially affect the largest number of victims, considerations about s-risks usually do not generate the strong feeling that a single identifiable victim does (all the more so because these are victims who do not yet exist).

3.5.4. Proportion bias

Proportion bias inclines us to measure the effectiveness of an intervention in terms of the percentage of the total problem it addresses that its outcome represents, instead of measuring it based on the benefit produced. For example, this bias may incline someone to think that it is better to use a certain resource to help 10 victims of a problem with 100 victims (since that is 10%) than to use it to help 100 victims of a problem with 2,000 victims (since that is 5%). In this case, the person would help 10 times fewer victims while thinking they were helping twice as many (or, worse, knowing that they are helping 10 times fewer victims).

As in the case of the neglect of wild animal suffering, which we addressed in Chapter 2, proportion bias may incline us to neglect (or even to oppose) a focus on the long term and on s-risks. For example, it could incline us to think that, even if we could benefit an enormously larger number of individuals by focusing on the long term, since the long-term future is so vast, this would still represent a tiny percentage of the total number of individuals who will exist over the course of the future. This could incline us to think, mistakenly, that it is more worthwhile to help a vastly smaller number of individuals affected by some present or short-term problem, merely because they would represent a larger percentage of the victims of the problem they face.

3.5.5. Belief digitization

Belief digitization is the tendency to think in terms of probabilities that are either 0 or 1, instead of taking uncertainty into account.

This tendency means that when a hypothesis is unlikely, people think its chances of occurring are nil, even while agreeing that the hypothesis is not negligible. Similarly, a plausible hypothesis is usually taken for granted, even when there is considerable uncertainty. So someone might dismiss a given s-risk because they mistakenly think its probability is zero.

Belief digitization can also lead us to focus on a very narrow range of s-risks, since we may be overconfident that they will occur (for example, because they are the risks that most catch our attention, something that has to do with salience bias).

3.6. Conclusions

Given these biases, it is no surprise that few people are concerned about the long-term future and, above all, about s-risks. As we have seen, even people concerned with the long term tend to neglect the issue of s-risks. This neglect is a reason to invest resources in raising awareness of s-risks and reducing them⁵⁶. Given that this topic has hardly been addressed, we can make significant progress on it.

An important strategy for trying to deal with the influence of biases in general, and especially in this case, is to try to remain open-minded and not close ourselves off from thinking about controversial and counterintuitive issues, and to practice changing our position if new evidence and arguments emerge that present good reasons for changing our priorities and strategies. In the next chapter we will explore this question in more detail.