Word count: 4,163

1. Introduction

If we want to increase happiness it follows we need to understand what makes people happy and for how long. In this paper I explore the purpose and mechanisms of hedonic adaptation – the process by which the affective intensity of constant of repeated stimuli diminishes over time – and argue these suggest some peculiar and unappealing implications both for how to increase happiness and morality as a whole. A key challenge is that our folk psychological understanding of happiness is not matched by the evidence available. Many of the things we would expect to have a long-term impact on our happiness do not. These include, but are not limited to, becoming richer, becoming paraplegic, being diagnosed with HIV, being incarcerated and even being incarcerated then put in solitary confinement. On the other hand, we show an inability to hedonically adapt to other experiences such as noise, bad smells, tedious commutes and chronic pain (for summaries see Frederick and Loewenstein, 1999 and Lyubormirsky, 2011). I argue quite a lot may turn on how hedonic adaptation works. It is therefore surprising that, to best of my knowledge, the problem seems to have received no philosophical scrutiny so far and social scientists who have written about it have not recognised the stark implications of the topic.

I first provide a rough theoretical explanation of the mechanisms and evolutionary purposes of hedonic adaptation. It’s not entirely clear how hedonic adaptation is supposed to work or where its limits lie, so I generate 3 hypothetical models for how it might function. I then show each of them yields highly unintuitive and unhelpful results for how to maximise happiness in whatever the domain is of things that we hedonically adapt to.  I suggest this implies some surprising revisions of morality, including that, under some conditions, intentionally making someone miserable in the early part of their life is a good thing to do, and that these sorts of conclusions demonstrate this area requires further philosophical scrutiny.

2. Why does hedonic adaptation occur?

Before discussing the difficulties adaption generates for efforts to increase happiness, it will be helpful to get a clearer understanding of what adaptation is and what purpose it serves. From Frederick and Loewenstein (1999):

Adaptation [in general] refers to any actions, process or mechanism than reduces the effects (perceptual, physiological, attentional, motivational, hedonic and so on) of a constant or repeated stimulus. Adaptation can occur at several different levels – from overt behaviours that reduce exposure to a stimulus, to molecular changes at the cellular level that diminish the perceived or experienced intensity of an objective stimulus.

For example, when you get out of bed and pull open the curtails, the light outside seems painfully bright compared to the darkness your eyes had adjusted to. By a combination of things like shielding your eyes and having your pupils contract to let in less light, you will soon be able to see ‘normally’ and sufficiently well to navigate the world. Whilst there are limits to adaptation – we can’t see in total darkness, or look at the sun – such adaptation allows the human eye to see in environments with light intensities a million times that of others, a rather useful and remarkable feat.

Hedonic adaptation refers to adaptation to stimuli there are affectively relevant, which is to say they make us feel good or bad. As Frederick and Loewenstein (1999) point out, this could happen in a number of ways. For instance, hedonic adaptation to a foul smell might be sensory: as a direct result of sensory adaptation a less intense odour will seem less unpleasant. It could also by physiological: a paraplegic gets stronger arms that allow him to move his wheelchair more easily. However, hedonic adaptation is often cognitive: our interests, values, goals or attention changes. For example, a recent paraplegic might decide to play chess instead of his previously adored tennis or put his thoughts elsewhere so he’s reducing the quantity, although maybe not the intensity, of his negative thoughts. He might decide the tragedy is a valuable learning experience. Alternatively, the neurotransmitters in his brain might work to oppose the negative affect he feels and so reduce it: for example, there’s evidence that high-level use of narcotics may diminish the reward pathways in the brain (Cassens et al. 1981)

The next question is why hedonic adaptation occurs. Put another way: why can’t we simply live lives filled with constant pleasure? The most plausible explanation is an evolutionary one: happiness is nature’s[1] way of getting us to do the things that help us survive and propagate our genes. However, intense sensations are evolutionarily costly and hedonic adaptation increases our fitness to survive. (see Frederick and Loewenstein, 1999, Rayo and Becker, 2007, Perez-Truglia, 2012)

1. Intense sensations are costly in terms of energy. The brain of an average adult human represents approximately 2% of the total body weight while still consuming approximately 20% of the energy (Perez-Truglia, 2012).
2. Intense sensations diminish awareness. For instance, if eating an apple put us into an intense period of euphoria, we would not be able to focus properly and would be an easy target for predators.
3. We a have limited range of perceptual sensitivity based on the stimulation and firing of nerves cells, and hedonic adaptation helps us to make better choices by being better able to discriminate between the options available (Rayo and Becker, 2007)

As Perez-Truglia (2012) argues the optimal solution to this problem, from nature’s perspective, is to adjust the individual’s utility function by her expectations so that experienced utility depends on the difference between what she expected and what actually occurred. This avoids the above three costs whilst at the same time allowing the individual to still be rewarded for things that aid gene propagation.

As far as providing an explanation goes, this is probably sufficient for our purposes. Given the paucity of evidence on which particular things people do and don’t adapt to, it’s illuminating to have a sense of what our hedonic experiences are for. The picture we’re left with is that we should expect hedonic adaptation to occur wherever it can in order to reduce the 3 costs I associated above with intense hedonic sensations. Put another way, we should expect making people very happy or unhappy for very long will be nearly impossible.

3. How should we model hedonic adaptation?

Given our interest in being happier, the next two question that follow are 1. How we should model the process of hedonic adaptation? and 2. On the basis of those models, how should we optimise for happiness amongst the things we hedonically adapt to? I’ll have to leave the question of how to optimise for happiness in general – presumably we adapt to some things and not others – for another time, but here I’ll focus just on how to optimise for hedonically adaptive happiness.

As far as I know, three models for hedonic adaptation have been proposed (Frederick and Loewenstein, 1999, Lyubormirsky 2011). We might call these the ‘lifetime average’, ‘short-term’ and ‘major event’ models. As it’s not clear which is correct, my approach will be to spell out how each of them works briefly before discussing their implications in general.

3.1 Lifetime average model

Perhaps the simplest model of hedonic adaptation, this assumes that the level of a stimulus that elicits no response is the average of all past stimulus levels (from Helson 1947). In other words, if a stimulus is higher than average you feel good, if lower, you feel bad.

Here’s a stylised example of how this might work with an overall happy life. Take the units in the ‘stimulus level (SL)’ column to refer to quantities of something, such as number of hats a hat salesman sells, income in pounds divided by 10,000, or perhaps a given dosage of cocaine. I assume for simplicity the starting adaptation level is at 5 for the 0^th period. Thereafter the adaptation level is the mean average of all the stimulus level at previous time periods. Happiness is the hedonic experience brought about by the gap between the stimulus level (SL) and the adaptation level (AL). Time period can refer to whatever time length seems most relevant.

Note how the adaptation feature is captured by the extending average. So in the 4^th period even though the stimulus level was the same as in the 3^rd period, the 4^th period was less unhappy in contrast because adaptation level had gone down.

Suppose we want to maximise total happiness on this model. As the stimulus level of an earlier period counts in each subsequent period, therefore lower stimulus in the first part of life contributes to a lower average that allows greater happiness overall. If this is the case, the happiest possible life one could hope for follows this pattern:

In this the first half of the life has as little of the stimulus as possible as the second half as much as possible (I’m presuming ‘10’ is the maximum possible stimulus). I’ve again assumed the adaptation level starts at 5, which means the person is unhappy until time period 3, but I could just as easily have assumed the starting adaptation level was 0. In that case the person here wouldn’t count as unhappy in the first half of their life but neutral because, so to speak, they wouldn’t know what they were missing out on. The person then gets the maximum stimuli at time period 3 – let’s say he sells 10 hats – and experiences great happiness which then reduces in the subsequent time periods as he sells the same number of hats again.

• Short-term adaptation

Suppose we don’t find the first model particularly compelling, and instead assume that more recent events are going to determine the adaptation level whereas more historically distant events will cease to be relevant.[2] In the example below I’ve assumed for simplicity it’s just the last time period that determines the adaptation level of the pervious time period, where the length of the time period can be anything.

Notice here that because the stimulus level of a time period time becomes the adaptation level of the next time period, and this goes on and on, all the events apart from the first and last cancel each other out. Hence the total happiness experienced is just the stimulus level in the last time period minus the original adaptation level.

If this is the model of hedonic adaptation it implies nothing that happens in between the first and last periods of your life effects your overall happiness: you can only have a pleasant experience if it was better (i.e. higher stimulus) than the last one, which then makes it harder to have another pleasant experience because a higher stimulus level is required. Conversely, having an unpleasant experiences makes it easier to have a pleasant experience next time. On the presumption that you’re not able to control what happened at the start of your life, because that is now all in the past, the only way to increase your hedonically adapting happiness is to try and have as happy an experience as possible directly before your demise.

• Major event model

The third alternative that seems to exist between the first model – all your experiences contribute to hedonic adaptation – and the second model – only some amount of your more recent experiences contribute – is to say that some of life’s extreme or memorable events might be very significant for the setting your standards. For example, a holocaust survivor might reflect on their horrific experiences and see the rest of the world in a more positive light. Alternatively, as Parducci (1995) notes, Russian emigres who fled the revolution of 1917 felt poor for the rest of the lives because they took their pre-emigration level of wealth as the relevant standard, instead of altering their standards thereafter.

I won’t produce a table for this model but, building on the discussion so far, I hope it’s intuitively clear that, if happiness is the result of comparisons to major good or bad events, the happiest life will be one with a major bad event or events very early in life. The worse the event, the better other things seem in contrast, and the earlier the event, the more time there is to experience those comparative improvements.

It’s not clear on this model what it is that makes events major and this will make a difference for optimisation: is it just the strong hedonic experience they produce, or are they memorable for some other reason? If an event becomes ‘major’ if it moves over a hedonic intensity threshold, that means we had better stay away from really happy moments until close to the very end of life. For instance, it implies an intensely pleasurable experience – perhaps an evening staring into the eyes of an apparently perfect partner who later spurns you, or the mountaineer who considers summiting Everest to be the best moment possible moment he could have – might actually cause a substantial reduction in your happiness thereafter because nothing seems as good in comparison. Obviously having the best experiences at the end of life avoids this problem because there is no time for other experiences to feel worse. The best strategy in this case seems to be to have very bad experiences early on which comparatively improve your later experiences, then as many good ones as possible that are sub threshold, leaving the best until last. If major events are not triggered by the strength of the hedonic experience, it will be very important to know what that thing is and how to trigger or avoid it at the appropriate moments.

• Implications of the adaptation models

I now want to make a few general comments on adaptation and the adaptation models. I take it all of the models provide peculiar results for how to maximise one’s own happiness in the realm of things that one hedonically adapts to. I’ve not said exactly what the realm is as I don’t know the answer, although I will gesture towards it later. I think it’s worth spelling out and contrasting those with our ordinary assumptions.

There are whole range of experiences we enjoy that we do not expect to detract from our future happiness. However, where hedonic adaptation occurs it means that events which appear good at the time may, in fact, be positive, negative or neutral depending on the model of adaptation and the time of life at which the experience occurs. Take the following example:

Bad stay at Fawlty Towers: you spend an evening in Fawlty Towers, a famously bad hotel, and discover its terrible service, awful food and verbally abusive owner. You have a terrible time whilst there and leave hurriedly the next morning.

Ordinarily, we would think this event has detracted from your happiness. But suppose that all your experiences are things which you can hedonically adapt to and change your hedonic expectations for the future. Let’s consider this on each of the three models above.

On the Lifetime average model, your stay at Fawlty Towers will have increased your lifetime happiness if it occurred in the first half of your life and was below whatever your lifetime happiness average will be, and decreased it if it occurred in the second half of your life and is above your lifetime happiness average.[3] As highlighted earlier, this is because bad experiences pull down the average, so bad, or rather low stimulus, events early in life increase overall happiness. In contrast if you are approaching the end of life it is less of concern to experience joyous events which then increase your average, as you will soon be dead and unable to hedonically adapt further.

On the short-term model, Fawlty Towers will make no difference to your overall lifetime happiness unless it is that last experience you have. If it is any earlier than that, it simply has the effect of reducing your adaptation levels for your next experience.

On the major event model, and if we presume that Fawlty Towers does constitute a major event, it may increase your overall happiness depending on how close it is to the end of your life and which other major events you have experienced. Suppose you find your stay horrendously unpleasant and more unpleasant than your other major events then, when you reflect on other experience in the future, they will then seem better in contrast. If you have enough of a future to reflect on this contrast in, Fawlty Towers will have increased your happiness.

I take it this sort of analysis is a surprise because we do not typically expect events which make us miserable to increase our overall happiness. However, this is what hedonic adaptation implies because bad experiences can change our expectations. The converse also applies: good experiences can reduce overall happiness. Quite a lot seems to rest on the model of hedonic adaptation, which is why I think this problem requires more scrutiny, both from empirical and philosophical perspectives. The life-time average model suggests experiencing the early years of your life with minimal experience of the things that could make you happy, but that you can adapt to, is good for your happiness. If true, this would bear out the common assumption we have that making life too easy for children really does spoil them for later life. The short-term model suggests nothing really matters, amongst the experiences you can adapt to, apart from the end. The major event model suggests having really terrible experiences early in life may bring about more long-term happiness than having had a happy childhood. Before you go and torture your children though, I would point out this only applies to whatever is within the realm of hedonic adaptation. Some experiences – rape, war, etc. – are so severe we are unable to adapt to them and they result in Post-Traumatic Stress Disorder (PTSD). So we really need to know when bad experiences are good for you, and when bad experiences are bad for you.

This line of thinking may have some implications for morality. If we suspected that Basil Fawlty made his guests miserable because he sadistically enjoyed their suffering, we would be inclined to call him immoral. However, on the crucial supposition that the entire experience will be involved in the hedonic adaptation process and so effect the experiences of his guests’ future happiness, the consequences of Basil’s sadism may well have been neutral or positive overall. I’m not sure quite what to make of this – potentially we could blame Basil for his malice – so simply note it.

4. Conclusions – obvious problems and further research

I can’t pretend this provided the complete picture of happiness, only a partial investigation into one area that suggests that area is not well understood. In closing, I want to list a set of areas for further work.

The first important area that I’ve not attempted is to produce an understanding what the limits of adaptation are. It seems unlikely that all our experiences are entirely subject to one of the three models of hedonic adaptation I’ve sketched. As Perez-Truglia (2012) argues, whilst we should expect hedonic adaptation to occur for lots of things, we should not expect it to occur for things which form a warning/defence role. Sensations depending on warning/defence roles should always produce a strong hedonic reaction, otherwise they would fail to keep us alive. For instance, if the pleasure of eating an orange was so strong than, when I walked into a fire, I failed to notice the fire, I would do a very poor job at surviving. In support on this he cites Sternbach (1963) who notes that modern individuals with a rare congenital syndrome which makes them indifference to pain are almost all dead by their mid-thirties: many young children chew off their fingers or burn themselves against hot stoves.

Related to the first, not having given a clear account of what hedonic adaptation does and doesn’t apply to, I’ve been unable to give anything like a final view on how to maximise happiness. My suspicion is that focusing on non-adaptive phenomenon, which we might always expect to feel good or bad, such as pain, sadness, love and friendship, may be a promising route if hedonic adaptation exhibits the peculiar properties above. However, given that, in the long-run, we’re all dead, it’s possible there will be ways to increase happiness we’ll nevertheless adapt to that are more cost-effective than targeting the happiness we don’t adapt to. It may also turn out that much of hedonic adaption relates how people think – perhaps this is why some people are un-affected by events that traumatise others – and by thinking differently, there might be things we can do to extend the adaptation to positive events and speed it up to negative events (see Lyubomirsky, 2011 for a discussion of this), thus increasing happiness overall. This suggests focusing on changing how people think is a promising over the long-term in a way providing good stimuli which people can adapt to is not.

Absence from my discussion has also been any worries about methodology. Perhaps those who say they have adapted to paraplegia are really much less happy, but have simply changed their scales of reference. For reasons Perez-Truglia gives, this seems unlikely because it is evolutionarily advantageous to have warning/defence sensations that are intense in an absolute, and not just a relative sense, hence we should expect things like pain to be roughly comparable across human given our similarly functioning brains. However, I have not engaged in this topic here.

Further, I’ve not had a chance to discuss complications to the picture such as ‘sensitisation’. Far from hedonically adapting to some sensations, it seems some become more acute over time, the obvious example because the increasing, not reducing, irritation we feel when living or working with other people who annoy us. Neither have I discuss the concept of ‘feed-forward’ where our expectations about what will happen in the future can determine how we feel now. For instance, there’s evidence prisoners are most dissatisfied just before the leave prison, presumably because they are comparing their current incarceration with the expected freedom (see Frederick and Loewenstein, 1999 for a discussion of these).

Finally, I don’t think hedonic adaptation is the only major problem which requires analysis before understanding how to maximise happiness. There’s also the worry of ‘hedonic comparison’ that some things which increase my happiness, either adaptively or non-adaptively, bring about a reduction in the happiness of others. The most obvious example here might be status: part of what I enjoy so much about driving my imaginary Ferrari is that other people don’t have one, and presumably they feel bad about this fact in some sort of proportion to how good I feel about it.

Here I have just focused on how to understand and respond to hedonic adaptation for whatever realm of phenomenon hedonic adaptation covers. The limited conclusion I’ve reached is that hedonic adaptation, which seems to have been treated largely as a curiosity by those how have discussed it, rather than a serious impediment to happiness, is not only a serious impediment but, however understood, has some very peculiar implications.

Bibliography

Cassens, G., Actor, C., Kling, M. and Schildkraut, J.J., 1981. Amphetamine withdrawal: effects on threshold of intracranial reinforcement.Psychopharmacology, 73(4), pp.318-322.

Frederick, S. and Loewenstein, G., 1999. Hedonic adaptation. in Kahneman, D., Diener, E. and Schwarz, N. eds.. Well-Being: Foundations of Hedonic Psychology: Foundations of Hedonic Psychology.

Helson, H., 1947. Adaptation-level as frame of reference for prediction of psychophysical data. The American journal of psychology, 60(1), pp.1-29.

Lyubomirsky, S., 2011. Hedonic adaptation to positive and negative experiences. Oxford handbook of stress, health, and coping, pp.200-224

Parducci, A., 1995. Happiness, pleasure, and judgment: The contextual theory and its applications. Lawrence Erlbaum Associates,

Rayo, L. and Becker, G.S., 2007. Evolutionary efficiency and happiness.Journal of Political Economy, 115(2), pp.302-337

Sternbach, R.A., 1963. Congenital insensitivity to pain: A critique.Psychological Bulletin, 60(3), p.252.

[1] Obviously this is no agent ‘nature’, this is simply an easy way of referring to the process of evolution. We should expect our hedonic set up to be ancient, rather than recent: whilst the Genus Homo has existed for 2 million years, and agriculture for 10,000, agriculture has only been practised by the majority of the world’s population for 2000 years, a relatively brief time period to for selection to occur.

[2] At alternative for this model is what Frederick and Loewenstein (1999) call “time sensitive satiety” which refers to the idea that the happiness something gives us may depend on how recently we’ve experienced eat. For example, the enjoyment of a fancy dinner may depend on whether we had another such dinner last week or last year. I’m not sure how to model this, or whether it really counts as adaptation at all, so will have to investigate further how to make sense of it.

[3] In other circumstances it’s harder to say: we would need to know the details of the rest of the life and where happiness occurred in it.