The difference between preference and hedonic utilitarianism

Utilitarianism is the view that an act is morally good insofar as it maximizes utility. The best act in any given context is that act which maximizes utility, the worst act is that which minimizes utility, and each other act can be ranked somewhere between best and worst. The preference and hedonic variants of utilitarianism disagree on the nature of utility. Preference utilitarians think utility is sum total preference satisfaction (minus sum total preference frustration); hedonic utilitarians think utility is sum total pleasure (minus the sum total suffering). The disagreement regards the source of goodness and badness in the world: whether they are intrinsic to the satisfaction/frustration of agent goals, or whether they are intrinsic to certain experiential states. It is an axiological dispute.

At first glance it is odd to ask what differentiates preference from hedonic utilitarianism because it is obvious: one cares about preferences, the other about pleasures and sufferings. All we need to do to show the difference between these views is consider a case where satisfying somebody’s preference, all else equal, yields less pleasure than not satisfying somebody’s preference. Suppose somebody prefers vanilla icecream over chocolate even though the latter provides them more gustatory pleasure. In that case, hedonic utilitarianism would counsel choosing the chocolate and preference utilitarianism would counsel choosing the vanilla – the two frameworks straightforwardly differ.

The hedonic utilitarian, at this juncture, will point out that the agent choosing ice cream should (and would) come to the conclusion that they prefer chocolate to vanilla ice cream, were they better informed. Given the intrinsic goodness of pleasure, says the hedonic utilitarian, an agent who tried both chocolate and vanilla ice cream and found the former more pleasurable would (and should) prefer chocolate to vanilla ice cream. As such, the hedonist continues, the thought experiment as initially formulated is malformed – there is no case where a rational agent would prefer a less to a more pleasurable experience, all else equal.

From actual to ideal preference utilitarianism

To further motivate the preference utilitarian to accept this hedonic conclusion, the hedonist will point out that surely the preference utilitarian does not accept actual preference utilitarianism. Actual preference utilitarianism is the view that utility is sum total actual preference satisfaction. By actual preferences I mean the preferences which agents actually have. Suppose an agent has the mistaken belief that the glass of clear poison in front of them is a glass of water, and forms the preference to drink the liquid in the glass. Actual preference utilitarianism would counsel that, all else equal, the world would be better off if the agent satisfied their preference to drink the liquid – the agent should drink the liquid. But clearly this is false, because if the agent satisfied their preference they would experience severe pain and die.

The preference utilitarian should instead adopt ideal preference utilitarianism. Ideal preference utilitarianism is the view that utility is sum total ideal preference satisfaction. The nature of ideal preferences is contentious, but a workable definition is that they are preferences agents would have if they were rational and had deliberated about all the relevant facts. The goal is to avoid endorsing cases like drinking poison on the basis of mistaken beliefs.

Taxonomizing hedonic and preference utilitarianism

It is at this juncture that hedonic and preference utilitarianism seem to blend together. Because the hedonist thinks pleasure just is intrinsically good (and suffering just is intrinsically bad), the hedonist will argue that an agent with ideal preferences (i.e., an ideal agent) will just form preferences for actions which maximize pleasure and minimize suffering. Agents will do this, the hedonist claims, because experiences of pleasure just are the basis for goodness and experiences of suffering just are the basis for badness, so it is no surprise that an ideal agent will prefer to maximize net pleasure experiences. For the hedonist, there is no other ultimate basis for making a rational decision based on all the facts. At this juncture the preference utilitarian has three options:

1. Deny that ideal agents should (or do) form preferences based on pleasure-maximization and pain-minimization.
2. Accept that ideal agents should (or do) form preferences based on pleasure-maximization and pain-minimization, but claim that intrinsic goodness derives from the satisfaction of those preferences rather than the experiential quality of pleasure- and pain-states.
3. Same as 2, but accept that intrinsic goodness derives from the experiential quality of pleasure- and pain-states.

3 just is hedonic utilitarianism, so the interesting options are 1 and 2.

Option 1 can be further broken down depending on whether the proposed alternative basis for preference-formation involves experiential qualities. The preference utilitarian might disagree with the hedonist because they think there are other experiential qualities which are intrinsically good and bad apart from pleasure and suffering, and that ideal agents should form preferences on the basis of the intrinsic normative qualities of those experiences. Call this 1a. I think the hedonic utilitarian will be quite satisfied if the preference utilitarian takes option 1a. The disagreement between the hedonic utilitarian and the 1a preference utilitarian, if there even is a disagreement, is whether there are experiential states with intrinsic value apart from states of pleasure and pain. Really, this is no longer a disagreement between a preference and hedonic utilitarian – it is an internecine dispute between two hedonic utilitarians who disagree on which experiential states are intrinsically valuable.

The other variant for Option 1 — call it 1b — is that the alternative basis for preference-formation does not involve experiential qualities. This is a real dispute, because a hedonist must think that normative value ultimately derives from experiential qualities. In this sense 1b agrees with 2 that intrinsic good does not derive from any experiential quality.

The new taxonomy is in terms of experiential versus non-experiential bases for intrinsic value or ideal preference formation. Hedonic utilitarians (1a and 3 above) argue only experiential states are intrinsically valuable, and that ideal preferences are formed on the basis of those experiential states. Non-hedonic utilitarians (1b and 2 above) argue experiential states are not intrinsically valuable, and ideal preferences are formed on the basis of something else (or, alternatively, the satisfaction of ideal preference is good even if that satisfaction is not accompanied by any positive experiential states whatsoever).

Motivating hedonic over non-hedonic utilitarianism: the case of P-zombies

Here the hedonist, I think, has the decisive argument. Consider first the case of P-zombies, which are entities identical in all respects to humans except lacking any experiences. They look and behave just like people, but are “dark inside.” P-zombies, just like humans, have preferences – they set goals (e.g. winning the writing competition) and prefer that those goals be satisfied (e.g. they form a plan and invest effort into writing a winning entry). The argument I have in mind asks us to consider whether there is any value in P-zombies satisfying their preferences (and the conclusion I will suggest is no, there is not). Getting to this argument first requires some conceptual machinery.

Distinguishing psychological and phenomenal mental states

The reason p-zombies can have preferences is because preference is, at least in part, a psychological rather than a phenomenal state. David Chalmers usefully distinguishes between psychological and phenomenal states of mind. A state of mind is psychological insofar as it plays the right causal role in explaining behavior. A paradigm psychological state is learning – for an organism to learn something just is “for it to adapt its behavioral capacities appropriately in response to certain kinds of environmental stimulation.” (Chalmers, The Conscious Mind pg. 11). A state of mind is phenomenal insofar as there is a certain way it feels. A paradigm phenomenal state is seeing red or smelling a rose.

The case of pain and nociception can be useful in disambiguating these two different concepts of mentality. Strictly speaking, pain is a phenomenal state: it is the unpleasant experience often accompanying tissue damage. Nociception is a psychological state: it is the neural process of encoding and processing noxious stimuli (like tissue damage). While pain and nociception often present together, and are both subsumed under the colloquial concept of “pain,” they can come apart empirically and are distinct conceptually. (See Loeser and Treede, “The Kyoto protocol of IASP Basic Pain Terminology,” Pain 137 (2008) 473-477). For clarity, I will use “pain” to refer to the colloquial mental concept of pain (which conflates or combines its functional and phenomenal properties), “phenomenal pain” to refer to the unpleasant experience of pain, and “nociception” to refer to pain as a psychological state.

The reason pain is bad, argues the hedonist, is because it feels bad. Pain is only bad insofar as it is phenomenal pain. Actually, pain is quite good (instrumentally) insofar as it is a psychological state: it causes and trains our body to avoid harmful stimuli. A world where people were only ever in the psychological state of pain, and never had any experience of phenomenal pain, would be better than our world where the two are mostly coupled: people would avoid harmful stimuli just as well as they do now, except their avoidance would not be accompanied by any bad feelings.

The moral value of satisfying P-zombie preferences

What is so good about satisfying preferences? The hedonist will argue that, if anything is good about satisfying preferences, it is the positive experience – pleasure – which accompanies preference-satisfaction (and the avoidance of the negative experience – suffering – which accompanies preference-frustration). The preference utilitarian, unless they are happy to accept 3 above and become hedonic utilitarians, must deny that the goodness of preference satisfaction stems from any experiential quality of preference satisfaction.

The distinction between phenomenal and psychological mental states introduced in the previous section is helpful in understanding this dispute between the hedonic and preference utilitarian. The hedonic utilitarian thinks that preference-satisfaction is valuable insofar as it is a phenomenal state. The preference utilitarian thinks that preference-satisfaction is valuable insofar as it is a psychological state. So, for the hedonic utilitarian, there is no moral value in P-zombies satisfying their preferences; for the preference utilitarian, there is moral value in P-zombies satisfying their preferences.

Which view is correct? At this point, we may have reached intuitional bedrock. As for myself, I have the strong intuition that preference-satisfaction is good because it feels good, and preference-frustration is bad because it feels bad. This is analogous to my strong intuition that pain is bad insofar as it is a phenomenal state (i.e., insofar as it feels bad), intrinsically morally neutral insofar as it is a psychological state (i.e., insofar as it re-orients an organism’s behavior and cognitive state to avoid the pain-causing stimulus), and instrumentally morally good insofar as its psychological properties might help an organism avoid future phenomenal pain states.

More generally, it seems to me that psychological states have no intrinsic moral value whatsoever. When a living person is unconscious – not experiencing anything, as in a dreamless sleep or under total anesthesia – there are a number of psychological states that person is in. They still have memories, beliefs, and dispositions insofar as those are cognitive features of their brain-states. But there is nothing intrinsically valuable about those unconscious memories, beliefs, and dispositions. By analogy, consider a basic handheld calculator. A calculator also has memory and can perform certain cognitive operations. But, assuming for the sake of argument that the calculator has no phenomenal states – it does not experience anything – it does not seem there is any intrinsic moral value in the calculator’s performing various operations on the numbers stored in its memory. A P-zombie is just a much more complicated version of the handheld calculator, and it is unclear why added functional complications would generate intrinsic moral value.

If, after reading this, the preference utilitarian still insists that a P-zombie has identical intrinsic value to its experiencing counterpart, I still have two more moves. First, I would ask them to consider whether they would accept the following deal: they can either receive a hundred dollars and become a P-zombie (losing all future phenomenal states they might otherwise experience), or pay a hundred dollars to avoid becoming a P-zombie. In the former case, they will surely be able to satisfy at least one more preference than in the latter case – they will effectively have two hundred more dollars to work with! Under preference utilitarianism, the former case should strictly dominate the latter case. And yet, I get the feeling they would be willing to pay more – a lot more – to avoid becoming a P-zombie, and for good hedonic reasons.

The second move is to consider the case of the Unlucky Competition Winner.

The Unlucky Competition Winner – why psychological preference-satisfaction doesn’t matter

Imagine G has entered a writing competition with the goal of winning. They prefer to win the competition, and work very hard to write a good story. They submit the story on the deadline before the competition closes, and the judges have a week to return their verdict. On the last day of judging, the judges decide G has won the competition. They publish their results online at Time 1. Straightforwardly, G’s preference has been satisfied at Time 1 because G’s preference was to win the competition, and G has won the competition.

At Time 1, G does not yet know they have won the competition, because they have not yet checked the competition results. Suppose at Time 3, G checks the competition results and discovers they have won. They are elated that their hard work has paid off, and feel great about their achievement.

Now suppose instead that G is unlucky, and dies at Time 2 before checking the competition results. Call this the Unlucky Case. Here is the question: is G’s life better because their preference has been satisfied? It seems the preference utilitarian must say: yes! But this seems very strange indeed. Because consider the case where G has not won the competition, and similarly dies at Time 2 before learning about the results. Call this the Unluckier Case (it’s even unluckier to lose the competition and have died). This case is identical, from G’s perspective, to the previous case. G has lived exactly the same life in the Unlucky Case and the Unluckier Case. And yet the preference utilitarian must say G’s life has gone better in the Unlucky Case than in the Unluckier Case, even though – from G’s perspective – there is absolutely no experiential difference between the two cases. How can somebody’s life have gone better for them as a result of something they never experience?

Recall that we are talking about intrinsic value here. From one perspective, G’s life has gone better in the Unlucky Case – namely, from the perspective of those who learn about G’s accomplishment and think better of G’s literary talent posthumously. But what we are concerned with is G’s perspective, because we are concerned with what makes a life intrinsically more valuable to the person living it: preference-satisfaction or pleasure.

The hedonist has a ready answer to the Unlucky Case and the Unluckier Case. They deny that G’s life is better just because their preference has been satisfied at Time 1, since at Time 1 G has not yet learned they have won the contest. Therefore, they can reject the unintuitive conclusion that two experientially equivalent lives can have different intrinsic values.

The preference utilitarian might say that G’s preference is satisfied only when G learns that G’s preference is satisfied. This is false. G’s preference is to win the contest, not to learn that they have won the contest. If G learns they have won the contest, but later founds out this was an erroneous announcement, they will not think their preference has been satisfied.

The preference utilitarian might revise their theory to say that what is really intrinsically valuable is not satisfying one’s preference, but learning that one has satisfied a preference. At the very least, this is much less plausible than their original theory, and they would have to offer good reasons for why learning about one’s preferences being satisfied is intrinsically valuable. These reasons could not involve any phenomenal state of pleasure accompanying learning that information, because then their theory would just amount to hedonic utilitarianism.

Conclusion

Hedonic and preference utilitarianism disagree on the ultimate source of the good. Chalmers’ distinction between phenomenal (experiential) and psychological (cognitive-functional-causal) mental states helps clarify their dispute. Hedonists think the ultimate source of the good is in positive phenomenal states, or pleasure. Preference utilitarians think the ultimate source of the good is in psychological states – specifically in states of preference-satisfaction. I have argued that preference utilitarians are committed to the moral value of experienceless P-zombies. They are also committed to the view that two lives can be experientially identical (as between the Unlucky Case and the Unluckier Case), but one can be intrinsically more valuable than the other. Because these conclusions are implausible, hedonic utilitarianism is more plausible than preference utilitarianism.