Why the Experience Machine is Evidence for Hedonism
Explaining a subtle error in a great paper
Weijers and Schouten have, arguably, the most up-to-date treatment of the experience machine. It’s a great paper, though there is, I believe, one crucial error. While the paper ends up providing a solid response to the experience machine objection, the error makes the paper less decisive than it should have been.
The authors begin by explaining why the deductive version of the experience machine does not work. Their arguments here are, I believe, decisive. They then turn to the abductive version of the experience machine. They summarize it in the following way.
AP1.
In terms of the internal aspects of our experiences, an experience machine life would be much better than a life in reality. (Stipulated in thought experiment)
AP2.
When instructed to ignore their responsibilities to others, the vast majority of reasonable people report preferring reality over a life in an experience machine. (Empirical claim)
AP3.
The best explanation for AP2 is that reality matters intrinsically to the vast majority of reasonable people
AP4.
Inference to the best explanation: If a hypothesis is the best explanation of an observation, then it is rational to believe that hypothesis is true. (Standard methodological premise)
AC1
Therefore, it is rational to believe that reality matters intrinsically to the vast majority of reasonable people. (Modus ponens AP3, AP4)
AP5.
The best explanation for reality mattering intrinsically to the vast majority of reasonable people is that reality has intrinsic prudential value
AP6.
Inference to the best explanation. (Standard methodological premise)
AC2.
Therefore, it is rational to believe that reality has intrinsic prudential value. (Modus ponens AP5, AP6)
AP7.
If internalist prudential hedonism is true, then the internal aspects of pleasure and pain are the only things of intrinsic prudential value (or disvalue) in a life. (Stipulated definition)
AC3.
Therefore, it is rational to believe that internalist prudential hedonism is false. (Modus tollens, AC2, AP7)
They then provide a series of criticisms of the various premises. In response to AP5, the authors write
Crisp and Hewitt also argue that AP5 is false because reality mattering intrinsically to the vast majority of reasonable people is better explained by evolutionary and psychological explanations.Footnote39 Both Crisp and Hewitt use specific examples to argue that many of our judgments about which goods matter intrinsically to us could have developed because they aided our ancestors’ procreative fitness and provided us with pleasure throughout our personal development. Crisp mainly discusses our preference to accomplish real achievements and Hewitt focuses on our preference to establish real interpersonal relationships. Since both accomplishing real achievements and establishing real interpersonal relationships require us to live in reality, and on the assumption that evolutionary and psychological mechanisms best explain why accomplishing real achievements and establishing real interpersonal relationships matter to us, Crisp and Hewitt both conclude that evolutionary and psychological mechanisms best explain why reality matters intrinsically to us.
They argue, however, that these aren’t sufficient to refute AP5 — just to undercut our confidence in it a bit.
They then argue against AP3. First, they say
J. Bronsteen, C. Buccafusco, and J. Masur worry that our intuitive reactions to the experience machine thought experiment might be based on more than just an isolated prudential value comparison between reality and how our experiences feel to us on the inside: “…we must look behind whatever visceral aversion to the machine we might have and assess (i) whether that aversion relates to welfare and (ii) whether the aversion springs from rejecting the rules of the hypothetical example.”Footnote42
Indeed, many potential causes of negative visceral sensations, especially fear, have been identified in the experience machine thought experiment; it’s often described as scary, unrealistic, unreliable, and unsafe to the extent that a reasonable person could be forgiven for not wanting to connect to an experience machine for fear of “catastrophic, unimaginably horrible consequences of malfunction or abuse.”Footnote43 Adam Kolber notes that the fears of machine underperformance or failure are exacerbated by the troubling irrevocability of the experience machine and our general fear of the unfamiliar: “We are hardly comfortable enough with our own world to risk life under totally foreign circumstances.”Footnote44 Christopher Belshaw is also concerned that the thought of spending our whole lives in an experience machine “stacks the odds in favour of the cautious”, deterring many of us from connecting.Footnote45 Both Dan Weijers and J. Mendola also emphasise how worries about the machine feed into a general fear of the unknown.Footnote46 These fears lead Crisp to dismiss the question about whether people would or should choose to connect to an experience machine because their choices are likely to be affected by “differing attitudes to risk.”Footnote47 All of these worries about the machine are highly relevant to our well-being, but they should not enter into a direct comparison between the prudential value of reality and how our experiences feel to us on the inside. Therefore, if these worries significantly affect people’s preferences in the experience machine thought experiment, AP3 starts to come under pressure.
However, only a few of the authors mentioned argue that these irrelevant fears might amount to a refutation of AP3. This is because, although these fears plausibly affect people’s preferences in the experience machine thought experiment, they don’t obviously provide a better explanation for them than reality mattering intrinsically to us. For this reason, many of the philosophers who argue that these fears might provide a good explanation for the widespread preference for reality over a life in an experience machine also provide a further objection to AP3.
Next, they say
That further objection to AP3 is that people can have preferences for things because they promote several different kinds of value (not just prudential value). This objection argues that people might report preferring reality over a life in an experience machine because they prefer to promote aesthetic, moral, or other non-prudential values, which they could not achieve while connected to an experience machine. If the widespread preference for reality over an experience machine life is best explained by the widespread desire to promote moral values, then AP3 is false and the experience machine thought experiment does a very bad job of isolating a prudential value comparison between reality and how our experiences feel to us on the inside. If AP3 is false for this reason, then the experience machine thought experiment doesn’t tell us much at all about prudential value and, therefore, it can’t provide the basis for a reasonable objection to prudential hedonism.
Finally, they respond to AP2
Responses to AP2
Another group of responses to the experience machine objection to hedonism deny AP2. The contemporary denials of AP2 are often subtle and indirect, but at least one is incredibly straightforward. Tännsjö takes a direct approach by claiming that he would not choose reality over an experience machine life.Footnote51 He also points out that lots of people take drugs and argues that this is the pharmacological equivalent of connecting to an experience machine.Footnote52 Combined with his worry about people’s fear of abuse while connected to an experience machine, this evidence forms the basis for his claim that “it is far from clear that … the claim that we would not plug in, is true.”Footnote53 In a similar vein, Belshaw argues that many people, especially those whose remaining life years look to be somewhere between relatively dull and excruciatingly painful, might connect to an experience machine.Footnote54 However, this kind of approach will do very little to convince anyone who has presented the experience machine thought experiment to students (or any other group of people) that AP2 is true because these people have had first-hand experience of the vast majority of people reporting that they would prefer reality over a life in the experience machine.Footnote55
A much more promising approach to denying AP2 is to construct an alternate version of the experience machine thought experiment that does not elicit a widespread preference for reality over a life in an experience machine, while remaining essentially the same in all relevant ways. Based on the worry that the potential causes of people’s preference for reality over an experience machine life, such as irrational fear, are irrelevant to assessing prudential value, several philosophers have developed new experience machine scenarios. These new scenarios attempt to eliminate from consideration all of the factors that are irrelevant to an isolated prudential value comparison between reality and how our experiences feel to us on the inside. If other versions of the experience machine thought experiment can equally or better isolate this prudential value comparison, and people’s judgments about them significantly diverge, then either AP2 or AP3 might be false. Kolber, Felipe De Brigard, and WeijersFootnote56 have all created new experience machine scenarios that attempt to reduce interference from irrelevant factors.Footnote57 What is most notable about these new scenarios is that they are designed to reduce, or even reverse, the negative impact of unfamiliarity with, and fear of, experience machines by framing being connected to a machine as the status quo – you are asked if you want to disconnect from the machine that has generated all of the experiences that you remember.
For example, here is an excerpt from one of De Brigard’s reversed scenarios:
“I am afraid I have some disturbing news to communicate to you” says Mr. Smith. “There has been a terrible mistake. Your brain has been plugged by error into an experience machine created by super duper neurophysiologists. All the unpleasantness you may have felt during your life is just an experiential preface conducive toward a greater pleasure… we’d like to give you a choice: you can either remain connected to this machine (and we’ll remove the memories of this conversation taking place) or you can go back to your real life. By the way, you may want to know that your real life is not at all as your simulated life. In reality you are a prisoner in a maximum security prison in West Virginia. What would you choose?”Footnote58
What might be said about this case is that even if the vast majority of people would choose to go into the machine, all we have shown is that sometimes people are willing to trade reality for a significant improvement in the quality of their experiences (but we have not shown that they place no value on reality).Footnote59 But what is important about this case is that it is structurally analogous to Nozick’s original experiment. In both cases, the reader is being asked to choose between a life with much greater hedonic value (and much less connection to reality) and a life with much less hedonic value (and a much greater connection to reality). There is one important difference, though; in De Brigard’s case, the status quo is the first kind of life rather than the second (the experience machine life instead of reality). De Brigard found that the vast majority of people presented with this scenario preferred a life connected to the machine.
One plausible explanation for the judgments reported by De Brigard, then, is that they reflect a preference for continuing the current state of affairs (the status quo). If this is true, then AP3 is highly questionable because people’s preferences in the original experience machine case might be explained better by their preference for the status quo (what is most familiar to them) than by reality mattering intrinsically (or non-prudential values mattering intrinsically) to them. However, there is a potential problem for Kolber, De Brigard, and Weijers; their scenarios might have introduced new biases and other confounding factors that were not present in Nozick’s original scenario.
For example, in De Brigard’s case, we are choosing between a life with negative net hedonic value (and a strong connection to reality), and a life with net positive hedonic value (and a weak connection to reality). Whereas in the original experience machine case, we are asked to choose between a life that will nearly always have net positive hedonic value (and a strong connection to reality) and a life with much greater net positive hedonic value (and a weak connection to reality). What might explain De Brigard’s result, then, is that we are willing to sacrifice reality to avoid negative net hedonic value, but not willing to sacrifice it for hedonic improvements to lives that already have net positive hedonic value. A plausible explanation might be that hedonism is false – reality does have prudential value – but that sometimes hedonic considerations trump our preference for reality.
Nevertheless, for anyone who endorses the experience machine objection to hedonism and found themselves preferring a life connected to an experience machine over a life in reality in one of these new scenarios, the onus should be on them to point out the confounding feature of the new scenario(s).
A safer response for proponents of the experience machine objection to hedonism is analogous to the direct approach taken by Tännsjö above; to simply deny that they would choose a life connected to an experience machine over a life in reality in the new scenarios. While this approach is safer, it is not necessarily enough to prevent the denial of AP2 because these authors have begun to empirically test their thought experiments and it is no longer clear what the vast majority of people believe about experience machine scenarios.Footnote60 This means that attempts to deny AP2 by creating a new scenario that produces a different result from Nozick’s scenario requires a sound empirical result before defenders of internalist prudential hedonism stand a good chance of convincing stubborn proponents of the experience machine objection that AP2 or AP3 is false.
Recent work by Weijers attempts to provide grounds for an empirical refutation of AP2 with his test on a group of university students. Weijers has attempted to neutralise the status quo by stipulating that:
A stranger, named Boris, has just found out that he has been regularly switched between a real life and a life of machine-generated experiences (without ever being aware of the switches); 50% of his life has been spent in an Experience Machine and 50% in reality. Nearly all of Boris’ most enjoyable experiences occurred while he was in an Experience Machine and nearly all of his least enjoyable experiences occurred while he was in reality. Boris now has to decide between living the rest of his life in an Experience Machine or in reality (no more switching). You have had a go in an Experience Machine before and know that they provide an unpredictable roller-coaster ride of remarkable experiences. When in the machine, it still felt like you made autonomous decisions and occasionally faced tough situations, such as striving for your goals and feeling grief, although you didn’t really do these things. Your experiences were also vastly more enjoyable and varied in the machine. You also recall that, while you were in the Experience Machine, you had no idea that you had gotten into a machine or that your experiences were generated by a machine. Boris’ life will be the same length in an Experience Machine as it would in reality. No matter which option Boris chooses, you can be sure of two things. First, Boris’ life will be very different from your current life. And second, Boris will have no memory of this choice and he will think that he is in reality… Ignoring how Boris’ family, friends, any other dependents, and society in general might be affected, and assuming that Experience Machines always work perfectly, what is the best thing for Boris to do for himself in this situation?”Footnote61
Over half of the respondents to this survey (55%, 42/77) indicated that they thought Boris should choose the experience machine life.Footnote62 If Weijers’ result is valid, and he has not inadvertently introduced further confounding factors, then AP2 is false; when instructed to ignore their responsibilities to others, the vast majority of reasonable people do not report preferring reality over a life in an experience machine. But additional tests of Weijers’ scenario on other sample groups would be required before AP2 could confidently be rejected. Furthermore, proponents of the experience machine objection to hedonism should analyse Weijers’ scenario for new potentially bias-eliciting features and conduct surveys to test any hypotheses to that effect.
For example, the Boris scenario might be priming respondents to give greater weight to Boris’ experiences than to his values: the scenario explicitly mentions Boris’ enjoyable experiences, but makes no mention of his desires regarding reality, or what it is that he really wants out of life. In the case of the original thought experiment, a plausible explanation of the fact that most people would refuse to connect is that they have desires that can only be fulfilled in reality, or that they have values which could not be realized in the experience machine scenario. Given the structure of Weijers’ case, readers could assume one of two things about Boris:
1.
He has no desires or values that connecting to the machine would frustrate (and in this he differs from most ordinary people).
2.
He does have desires or values that connecting to the machine would frustrate, but he should choose the experience machine anyway.
So one possible explanation of the responses to Weijers’ scenario is simply that most people believe that other people should do whatever it is that best satisfies their desires – and if you happen to be a person who cares merely about experiences, then you should choose the life that leads to the best experiences overall. A variant of the case which stipulates that Boris either does (or does not) have particular desires or values might well get different results.
For example, here is a possible case: Boris really cares about having children. In both the real world and the experience machine, Boris has children. His real-world children are disappointments, while his experience machine children make him proud and happy. Now, because it is true that Boris cares about his children, he gets much more pleasure out of his experience-machine children than his real children, and he also gets more pleasure from his experience-machine children than someone who did not care much about their children would. Given that we have stipulated that Boris cares not just about his own experiences, but also about his children, it is possible that a significant number of people would say that it would be better for Boris if he chose not to connect to the machine.Footnote63
So, while AP2 should not be rejected outright, it is certainly on shaky ground because of the uncontested empirical evidence that the vast majority of reasonable people do not report preferring reality over a life in an experience machine. Furthermore, Weijers provides some evidence that the status quo bias and certain other confounding factors, such as irrelevant worries about whether the experience machine would work properly, are elicited by Nozick’s experience machine scenario. If this is correct, then AP2 might well be true for Nozick’s scenario, but AP3 would come into serious doubt; the best explanation for AP2 might not be that reality matters intrinsically to the vast majority of reasonable people (because people’s preferences might be directed more by confounding factors than by reality mattering intrinsically to them). The task, then, for defenders of the experience machine objection to hedonism is to offer a version of the thought experiment that is free of confounding factors and generates the result that the vast majority of reasonable people would not connect to the machine. If such a case cannot be offered, then it seems like either AP2 will be false or the best explanation of AP2 will not be an explanation that refutes hedonism (i.e. that AP3 is false). Again, further testing is required to empirically assess the extent of the bias elicited by Nozick’s and Weijers’ scenarios before anything definitive can be said about the fate of AP3 for Nozick’s scenario.
But here’s the error that I think they make — they ignore the dialectical context. This is being presented as an argument against hedonism. Thus, I think AP5 is false.
As I argue here, when we present many of the objections to utilitarianism in a bayesian way, they crumble. This case is no exception. Given the dialectical context, the best explanation of reality mattering to lots of people is not that reality matters intrinsically and hedonism is false.
If utilitarianism were true, we’d expect it to sometimes be counterintuitive. Moral disagreement and the poor track record of moral intuitions show that our moral intuitions, while often correct, are not infallible. Thus, we’d expect there to be cases like the experience machine that arise — we’d expect hedonism to seem unintuitive. Given this, the fact that hedonism seems unintuitive is no evidence against hedonism. The best explanation of the experience machine is hedonism being unintuitive sometimes because of our moral intuitions — rather than hedonism being wrong.
This motivates the following general principle.
Wide scope anti single counterexamplism For theories that apply to a very large number of cases, single counterexamples do not give us a good reason to reject them.
However, if it turned out that there were no good arguments for biting the bullet in the case of the alleged counterexample, then that would be evidence against the theory. We’d expect the correct theory’s judgments to be more plausible upon reflection. Given that they’d be correct, we’d expect careful reflection to lead to turn up defeaters for the counterexample.
But in the case of the experience machine, defeaters for the abductive argument present themselves. This is what we’d expect on hedonism. Thus, not only is the experience machine not evidence against hedonism, it’s evidence for hedonism. Hedonism would be expected to be unintuitive sometimes — but it would be odd if every time it’s unintuitive, there were independent arguments for the claim that hedonism was correct. The experience machine shows a case where this is true.
Thus, hedonists can reverse the force of the experience machine objection. It is not, in fact, evidence against hedonism.