How the Experience Machine Works

Robert Nozick’s Experience Machine counterexample to hedonism is one of the most famous thought experiments in contemporary philosophy. It has convinced many that there is more to prudential value than the felt quality of our experiences. Yet it is often misunderstood, and too easily dismissed. Most recently, Felipe de Brigard’s ingenious experimental study, whose results he and others, including Josh Knobe, take to cast aspersions on the Experience Machine argument, is based on a misunderstanding of what is at stake in it, as I will argue below. The key points concern the structure of Nozick’s argument and the nature of the relevant comparison. (I’m afraid this’ll be rather long, but it does meet my criterion for a blog post, namely being written in the course of a day in a fit of inspiration.)

1. What is the experience machine argument?

Let hedonism be the thesis that what is intrinsically good for us is nothing but the balance of pleasure over pain. It doesn’t matter how we understand pleasure, as long as it is a non-factive psychological state. If someone’s life contains as much pleasure and as little pain as it is possible for that person, there is no way to make that person’s life better for her. It is only the hedonic quality of experience that ultimately matters for well-being or prudential value. For hedonism, how those experiences are brought about is irrelevant.

An ancient counterexample to hedonism says that I’m worse off if I falsely believe that my partner is faithful to me (and perhaps take pleasure in this) than if my partner really is faithful to me. Nozick’s Experience Machine thought experiment mines the same vein, but generalizes the point and thereby eliminates some distractions (such as the influence of the thought that I might find out about the cheating and feel bad because of that). I won’t quote the passage here, as it is familiar to everyone. The basic point is that the Experience Machine produces the perfect illusion of leading whatever kind of life one desires, while one floats about in a tank. Nozick asks two different questions about plugging in to such a machine: “Should you plug into this machine for life, preprogramming your life’s experiences?” and, later, “Would you plug in?”. I think the point about preprogramming is a distraction – we can imagine Experience Machine 2.0 that infers your preferences from your virtual choices and keeps reprogramming your illusion to maximize your pleasure – so I will ignore it.

But the difference between these Nozick’s two questions is important, and too often neglected. The first question is a normative one: what should one do? I presume that it presupposes an answer to an evaluative question – what is the (prudentially) best kind of life for one? – so that if hedonism is true, one should (in one’s self-interest anyway) plug in. The second question, on the face of it, is a question about preferences, rather than about what is genuinely valuable. Nozick wants to argue that hedonism is false, so why does he ask the reader about her preferences?

I believe the answer is found by considering two different argumentative strategies in ethics. I’ll call them indicative and dialectical appeals to intuition. Here’s a schema for indicative arguments, which take our intuitions to function as indicators of how things are:

1.     The considered conviction of competent moral judges who fully understand scenarios A and B is that A is better than B for x.

2.     The considered convictions of competent moral judges are reliable guides to evaluative facts.

3.     Hence, we have good reason to believe that A is better for B than x.

(For premise 2, compare W. D. Ross, The Right and the Good: “The moral convictions of thoughtful and well-educated people are the data of ethics just as sense-perceptions are the data of a natural science”.)

Here’s the alternative schema for the relevant kind of dialectical appeals:

1.     If you would prefer A to B, abstaining from moral considerations, you regard A as better for you than B.

2.     If you regard A as better for you than B, you can’t in good faith rely on the premise that A and B are equally good for a person in moral argument.

When Nozick asks the reader if she would plug in (rather than whether it would be better for her to plug in), he is at least making the dialectical move against the would-be hedonist. This is, in a sense, a cheap move, as it avoids commitment to people’s reactions in general or to licensing an inference from preferences to evaluative truth. It thus can’t be rebutted by general sceptical arguments or empirical studies of what people in general think. It is perfectly legitimate, I think, to address the reader in this way in philosophical argumentation.

But Nozick is, I think, also arguing for the stronger conclusion of the indicative type of argument by appealing to preferences. This requires the additional premise that people’s preferences – or more precisely, considered hypothetical preferences – track their considered convictions about value. Assuming that most people are competent judges who fully understand the relevant scenarios, we can extract the following non-dialectical argument with an empirical premise:

The Minimal Experience Machine Argument

1.     Faced with the choice of a life in the Experience Machine and a qualitatively identical life in reality, most people would prefer the latter.

2.     People’s considered preferences reflect their considered convictions about value.

3.     People’s considered convictions about value are reliable guides to value, and thus good evidence about what is genuinely valuable.

4.     Hence, we have good evidence that other things besides the quality of experience are genuinely valuable, and hedonism thus false.

(I’ll return to the truth of these premises in a while.) I call this the minimal version of the argument, since it aims to show nothing more than that hedonism is false. Nozick’s actual presentation is more ambitious and arguably misleading, since he expects the reader to prefer their actual life to a qualitatively superior life in the experience machine. The argument appears to go like this:

Nozick’s Suggested Experience Machine Argument

1.     Faced with the choice of a life in the Experience Machine and a qualitatively inferior life in reality, most people would prefer the latter.

2.     People’s considered preferences reflect their considered convictions about value.

3.     People’s considered convictions about value are reliable guides to value, and thus good evidence about what is genuinely valuable.

4.     Hence, we have good evidence that other things are more valuable than hedonic quality of experience.

This is unnecessary to defeat hedonism, and may confuse some people, who dismiss the argument for wrong reasons. Premise 1, after all, is eminently questionable. Surely its truth depends on how painful most people’s actual lives are. But it is important to note that rejecting it is consistent with, and indeed in certain circumstances predicted by, non-hedonistic objectivism about value. After all, if pleasure is a good – and I certainly wouldn’t want to argue that it isn’t – and goods aren’t lexically ordered so that it is at the bottom of the list, it is possible for enough pleasure to outweigh other values, in particular values that involve being in touch with reality. If one’s actual life is bad enough in every respect, and the illusion supremely pleasant, an objectivist about the good may well believe that the balance of reasons points to plugging into the machine.

We could, to be sure, construct comparisons that would provide support for the thesis that other things are more valuable than the hedonic quality of the experience. The way to do this would be to give a choice between a life outside the machine that is stipulated to be less pleasant to some degree, but rank higher in terms of the value in question to the same degree. (This is not a trivial comparison, and incommensurability would make it even tougher.) Here the thesis to defeat is that pleasure is equally good as, say, meaningfulness. If we prefer a more meaningful but less pleasant life to a less meaningful but more pleasant life, we value meaningfulness more than pleasure. I’d wager that competent judges – and here we might apply Mill’s criterion requiring experience with both kinds of life – would indeed agree that other things than pleasure are more important to faring well in life.

Let’s return to the premises of the Minimal Experience Machine Argument. Premise 3 is much disputed in value epistemology, though I suspect most ethicists make use of it in some form. In any case, it raises no special issues for the Experience Machine, so I’ll let it pass. Premise 2 is not true in the general case – many other things than our value convictions influence our preferences, most obviously in cases of weakness of will. But preferences about hypothetical scenarios may be less subject to such interference, and in any case it is always possible to ditch it and ask directly about considered convictions, as in the first (indicative) schema above.

Premise 1 is the empirical assumption of the argument. It is not a simple prediction of survey results, as it refers to competent judges who fully understand the scenarios, which is a non-trivial condition (as we’ll soon see). But let’s say the relevant comparison was put to people. We’d ask whether they would prefer to lead the life of their dreams – say, be a Nobel-prize winning rock star – in reality, or have the perfect illusion of leading such a life while lolling about in a tank. Recently, when I put this question to my students, one raised her hand and said: “When you put it that way, it’s obvious.” Well, I think so too. I think it would be pointless to run a survey. (Indeed, asking a question with such an obvious answer might prompt many to suspect it’s a trick question and answer differently.) I’m willing to wager a lot that anyone who understands the question and seriously considers it prefers actual friends and success to merely illusory ones.

So I think the conclusion holds, and we have good reason to believe that hedonism is false. The argument doesn’t show which other things besides pleasurable experiences are valuable, only that actually doing things is necessary to realize those other values. (It is a mistake to conclude, as Nozick seems to, that contact with reality is itself intrinsically valuable.) I think plausible candidates include intimate relationships and carrying out meaningful projects. Neither of these is possible if we in reality just lie in a tank, unable to interact with real people or make a real difference to the world. (Note that I’m not talking about a preference for direct bodily presence: it is possible to interact with real people virtually, as in Second Life, and make a difference to the real world in virtual reality, as in Ender’s Game. But these are not experience machine scenarios.) I doubt if being in touch with reality is itself intrinsically valuable, though perhaps it is, if knowledge amounts to being in touch with reality and is an intrinsic value, as some say.

 

2. Plugging Out vs Plugging In

Now, what about Felipe’s study, and his hypothesis that it is status quo bias that explains the preference not to plug in (or out)? (De Brigard, Felipe (2010) 'If you like it, does it matter if it's real?', Philosophical Psychology, 23: 1, 43-57) Felipe’s clever idea is to invert the situation, and ask whether people would plug out of an experience machine, having lived there for their whole lives. Here’s his neutral vignette:

It is Saturday morning and you are planning to stay in bed for at least another hour when all of the sudden you hear the doorbell. Grudgingly, you step out of bed to go open the door. At the other side there is a tall man, with a black jacket and sunglasses, who introduces himself as Mr. Smith. He claims to have vital information that concerns you directly. Mildly troubled but still curious, you let him in. ‘‘I am afraid I have to 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 superduper neurophysiologists. All the experiences you have had so far are nothing but the product of a computer program designed to provide you with pleasurable experiences. All the unpleasantness you may have felt during your life is just an experiential preface conducive toward a greater pleasure (e.g. like when you had to wait in that long line to get tickets for that concert, remember?). Unfortunately, we just realized that we made a mistake. You were not supposed to be connected; someone else was. We apologize. That’s why 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.’’

What would you choose? [Please circle only one option]

Remain connected Go back to reality

Different scenarios vary the nature of the real life: the subject is either told that she is a prisoner (negative) or a multimillionaire (positive). It turns out that about half the people are unwilling to plug out in neutral and positive scenarios, and most people unwilling to plug out in the negative one. Felipe explains this in terms of a form of risk aversion, namely status quo bias, and suggests the same mechanism may explain responses to the traditional thought experiment.

You will be unsurprised to hear I find the study deeply problematic. The design itself is odd: if someone is told their whole life, as they have experienced it, is an illusion, what does it mean for them to “go back to reality”? They’re not asked to go back anywhere, but to abandon what they took to be reality for something else. (The other neutral scenario just says ‘disconnect’ and that life outside the machine is very different.) That is sure to be a bizarre and disconcerting thought. In fact, I don’t believe we’re psychologically capable of seriously entertaining the hypothesis that our whole life has been an illusion. (This is a version of the problem of imaginative resistance, I suppose.) Think for a while what it would mean. You never actually had that first kiss, or any other kiss for that matter. You never heard music. You never jumped in a lake. For all you know, there are no lakes, no people, no cars. What you took to be your parents or children never existed, except in your mind. For all you know, you have no head. Are these possibilities that you could seriously entertain? There’s also a major worry about personal identity. If we subscribe to a psychological continuity theory of some sort, in what sense is the body in the tank yours? Who am I – this guy who grew up there and wrote this and that, or someone who never did any of those things (such as a multimillionaire artist living in Monaco)? What are we talking about when we say that I am really some person who lacks any psychological continuity with me?

So there’s a serious question whether we can ever make the comparison we’re asked to. But in any case, it’s the wrong comparison. The question to ask would be: would you be indifferent to your life, as you have experienced it so far, to have been an illusion, or for its events to have happened just the way you think they have? (The scenario: Mr Smith comes to you and says that you may be a person whose life has hitherto been a computer program; would you prefer his suspicion to be unfounded?) If you prefer to have actually gone to the prom or heard your child’s first cry, you’re not a hedonist.

So the important issue in a backward-looking case is not whether people would prefer to plug out, but whether they would prefer not to have been plugged in. And it’s easy to see why they wouldn’t want to plug out, if they’ve been plugged in – after all, that would amount to giving up all the projects and relationships they’re invested in (and whose reality they are probably psychologically unable to doubt). That’s not a bias, unless you call wanting to keep the things that make life worth living a bias.

Maybe it comes down to this: our relationships to the past and the future are not symmetrical; one is what it is, and the other is not yet. It’s not the same thing at all to give someone the choice of two alternative futures, given the actual past, and give someone the choice of two alternative futures, given a counterfactual past. The former is a choice we can understand, the latter isn’t. There’s thus very good reason to be suspicious of the results of Felipe’s experiment that is absent in the Minimal Experience Machine Argument. Whatever explains those results won’t explain the standard intuition, unless people flat out refuse to accept the terms of the scenario.. Hence, we’ve no reason to be sceptical about the intuitions about the original scenario.

Comments

[Sven Nyholm]

Hi Antti,

nice post.

Are you thinking of the experience machine argument as being aimed to produce a general conclusion about what is prudentially good for everyone, independently of their personal attitudes? Or are you understanding it as a test individuals can use to get clear on what their own conception of the good is?

I'm asking because it seems to me that whether or not survey results are relevant to the question of whether to accept hedonism seems to depend on just how universal (if at all) our ideas of the good life are intended to be. If, for example, we take a view on which what's good for an individual, or what makes her life go well for her, chiefly depends on what she herself (perhaps in certain more or less idealized conditions) desires or values, then how most people respond to this type of thought experiment seems rather irrelevant. What is relevant, from each particular agent's point of view, seems then only to be how the person herself responds to the thought experiment.

(What I just said may not be any kind of objection to Felipe's argument/view since Felipe's paper, unless my memory fails me, is not so much about the question of whether hedonism is a good view or not, instead being about what exactly is going on when people are faced with the experience machine thought experiment.)