What is actually good for us?

In contemporary discussions of hedonism, the most influential thought experiment has become Nozick’s “experience machine”. The thought experiment is regarded as establishing that we attribute intrinsic value to more than just pleasure for ourselves and, hence, hedonism’s "time is up" (Hewitt, 2010, p. 332). However, in this essay, via juxtaposing desire fulfillment theory (DFT) and hedonism (specifically Mill's Qualitative theory of hedonism), I will argue that the experience machine does not straightforwardly refute hedonism. I will first outline the scenario created by the experiment, then I will explore how DFT and hedonism differ in their respective responses to the imagined situation. I will argue against DFT's claim that “what is prudentially good must be actually experienced” by claiming this is an intuition formed upon cognitive biases. Finally, I will argue that with a clearer understanding of the types of pleasure the machine would provide (drawing on Mill’s theory of qualitative hedonism), appropriately informed people would, in fact, be largely inclined to enter the machine. Hence, I argue, the experience machine actually supports hedonism.  


Theories of wellbeing are generally divided into three main categories: hedonism, desire fulfillment theories (DFT) and objective list theories (Fletcher, 2016). I will focus on the first two theories. Hedonism states that all and only pleasure is good for us, as a theory of prudential well-being. Crucially, hedonists do regard things other than pleasure as good for us, but only in as much as they are instrumental towards the fundamental end of pleasure. Secondly, hedonism claims that pleasure is valuable irrespective of our desire for it. The basic formulation is as follows: 

1. All and only pleasure is (non-instrumentally) good for us. 

1. All and only pain is (non-instrumentally) bad for us. 

1. A person’s overall level of well-being is determined solely by the balance of pleasure and pain they experience. (Fletcher 2016, p. 8).  

For many, Hedonism's simplicity intuitively feels reductive, intuitively we do in fact value objects other than mere pleasure. DFT seems to satisfy that intuition by claiming that there is indeed an additional or alternative constituent to our wellbeing, the actual satisfaction of our desires. Crucially, what matters to DFT is desires legitimately being fulfilled, not merely the experience or feeling of them being fulfilled (Fletcher, 2016, p. 29). This entails that if a subject is unaware of a desire having been fulfilled, this can still be good for them. So, whilst a hedonist would say that a fully accurate simulation of pleasure is prudentially good, DFT would claim there is no intrinsic value in this experience as simulations cannot actually fulfill our desires.  



Nozick asks us to consider a situation in which "super-duper neuropsychologists" have created a tank capable of simulating a life of abundant pleasures and void of all suffering for those willing to “plug in” (Nozick, 1974, p. 42). The experiences within the tank would be entirely fabricated. However, once entered, subjects would have no knowledge of the simulated state they previously agreed to.  



Hedonism, of course, would instruct one to enter the machine, because guaranteed pleasure would maximize what is good for us. On the contrary, DFT would vehemently deny that entering the machine is better for us, since  a life in the tank would prohibit one from ever actually fulfilling desires, disaster.  The vast majority of us do indeed recoil from such an offer, Nozick himself likened entrance to suicide (Nozick, 1974, p.43). In fact, several empirical investigations observe that most participants claim they would not choose to enter. Weijers 2014 found that 84% of participants say they would not enter the tank (Weijers, 2014, para. 9). And thus, an argument emerges: 

• P1) Our robust and well-informed preferences provide evidence of the correct theory of wellbeing
• P2) If our well-informed preferences provide evidence of the correct theory of wellbeing, then: if we have robust and well-informed non-hedonistic preferences, hedonism is false
• P3) (From P1–2) If we have robust and well-informed non-hedonistic preferences, then hedonism is false
• P4) We have robust and well-informed non-hedonistic preferences (as established by the experience machine thought experiment)
• C) Therefore (from P3–4), hedonism is false ☐

This argument takes the descriptive statement that “people show reluctance to enter”, and derives (by implication) the normative consequence that “people should show reluctance to enter the machine”, at least from the prudential perspective (Buscicchi, 2022, Section 4). This leap is plausible, since we do seem to regard intuitions as valuable reflections of where intrinsic value lies. However, this leap should only be made when we are confident our intuitions are suitably informed and unbiased. 


I will argue that our intuitions about the experience machine are not fully informed, due to cognitive biases. In fact, when all these biases are removed, I argue that people's answers change so significantly, that experimentally observed preferences would support hedonism. And of course, as intuitions become more informed, we should follow the direction they point: if it appears that preferences reliably move towards hedonism or any other theory, we can become more confident in hedonism, even if people’s true preferences remain mixed. 



The biases that impede us when considering nozick's machine fall into many categories. First, we appear to be pretty bad at imagining things and fall into two main imaginative failings; imaginative resistance (where subjects reject a crucial stipulation of the experiment eg. that it is safe) or overactive imagination (where subjects imagine additional features of the machine) (Buscicchi, 2022, Section 5).


Secondly, we have strong biases for familiarity, "status quo bias", or as Fletcher puts it “the mistake of giving undue weight to the option they believe to be the current state of affairs” which again through tweaking the conditions of the experiment seems accountable for much of the observed reluctance (Fletcher, 2016, p. 16). Psychological evidence more broadly confirms the strength of status quo bias (Maden, 2024).


There may also be technophobic biases at play: we are surrounded by stories of mutant evil tech both in fiction and reality; and hence we may be inappropriately untrusting of “sci-fi” scenarios (Maden, 2024).Together, these irrational factors seem to account for a truly significant portion of observed reluctance.  



However I believe the most influential bias is something Wiejer did not explore and often goes under-examined, I shall refer to it as the “gluttony bias”. This is the idea that due to the connotations of “pleasure” we imagine pleasure to be a gluttonous rampage; perhaps of sex, food, and general excess. Of course, lived experience quickly reveals these things are in fact a very bad recipe for obtaining pleasure. As Mill and many others have observed, the "pleasure paradox" suggests that the relentless pursuit of pleasure is perhaps the most reliable path to an unpleasurable life (Hewitt, 2010, p. 4). Furthermore, our bodies are simply physically not equipped for this type of "extreme" pleasure (Hewitt, 2010, p. 4). And so it follows, that if we believe the machine will provide this type of gluttonous experience (be it consciously or subconsciously) we are rational to not wish to enter.  
   

However, I will argue that a machine truly committed to simulating the most "pleasurable" life would adopt a qualitative form of hedonism, and thus these reluctances are misinformed. As observed by Mill, "It would be absurd that while, in estimating all other things, quality is considered as well as quantity, the estimation of pleasures should be supposed to depend on quantity alone." To understand pleasure more accurately, he proposes we adopt a qualitative hedonistic view in which the quantity and quality of a pleasure constitute its score on our overall "hedonic level" (Fletcher, 2016,  p. 8) (Mill, 1863, Chapter 2). 


In agreement with Mill, I will argue the relative quality of a pleasure can be seen as the pleasures that adopt the agent's higher faculties. This understanding can explain why we do not envy a life of extreme pleasure if it does not employ our higher faculties or as Mill states, “It is better to be a human being dissatisfied than a pig satisfied” (Mill, 1863, Chapter 2). 



Notably, higher faculty pleasures describe a wide variety of experiences. These may appear as pride, love or dignity; but could also include a deep sense of nostalgia or even catharsis. We should be cautious not to equate sensory pleasure with pleasure in the sense of “those experiences which are good for us”; and hence we should profoundly widen our image of what is considered pleasurable. This does not mean that the machine would not also be full of ‘lower faculty’ pleasures —it simply means it would ensure both the balance and supply of the highest “quality” pleasure as well.



Thus, if we are fully informed on the machine's "anti-gluttonous" intentions, it is plausible that our intuitive responses would shift in the direction of entering the experience machine, and thus towards hedonism.



Consider the original argument I described for hedonism. Proponents of hedonism could (and often do) deny (P3), that “If we have robust and well-informed non-hedonistic preferences, then hedonism is false”. I am arguing that the hedonist can accept (P3), and instead deny (P4), that “We have robust and well-informed non-hedonistic preferences (as established by the experience machine thought experiment)”. This is because, plausibly, stated preferences around whether to enter the experience machine are not well-informed, but are rather subject to psychological biases and misapprehensions. Furthermore, if we were better informed, I speculate that we would be more inclined to enter the experience machine, and thus that well-informed preferences would in fact support hedonism as a theory of prudential wellbeing.