1. Introduction
An increasingly popular way for philosophers to argue against opponents who prefer status quo options, in this or that domain of inquiry, is to try to explain away their preferences as resulting from ‘status quo bias.’ Dorsey (Reference Dorsey2010) suggests that welfare-relevant preferences are often shaped by status quo bias. De Brigard (Reference De Brigard2010) suggests that preferences for real experience over virtual experience, as expressed in ordinary reactions to Nozick’s experience machine thought experiment, are often motivated by status quo bias. Wilkinson (Reference Wilkinson2009) suggests that much of the opposition to opt-out approaches in medicine is driven by status quo bias. Kahane and Savulescu (Reference Kahane and Savulescu2015) suggest that opposition to human enhancement is motivated by status quo bias, and there are many more examples of explanatory appeals to status quo bias that can be found in the recent philosophical literature.
The mere fact that someone prefers the status quo to change does not show that they are in the grip of status quo bias. Sometimes, in some circumstances, the status quo option is the best available choice and would, therefore, be the choice of an unbiased, rational person. To substantiate the charge that a particular preference results from status quo bias, we need to offer supporting evidence over and above a mere demonstration that status quo bias might possibly have shaped that preference. In a much cited article, Bostrom and Ord (Reference Bostrom and Ord2006) present a test, which, they argue, can be used to provide corroborating evidence for the presence of status quo bias.Footnote 1 This is the reversal test. The reversal test has been appealed to by many, including Wilkinson (Reference Wilkinson2009, 236), Sandberg (Reference Sandberg, Savulescu, ter Muelen and Kahane2011, 83), Rippon (Reference Rippon2012, 350–354), and Kahane and Savulescu (Reference Kahane and Savulescu2015, 138).Footnote 2
Bostrom and Ord describe the reversal test as follows:
Reversal Test: When a proposal to change a certain parameter is thought to have bad overall consequences, consider a change to the same parameter in the opposite direction. If this is also thought to have bad overall consequences, then the onus is on those who reach these conclusions to explain why our position cannot be improved through changes to this parameter. If they are unable to do so, then we have reason to suspect that they suffer from status quo bias. (Reference Bostrom and Ord2006, 664–665)
The reversal test can be used to substantiate the assertion that a preference for a status quo option, which can be represented as the value of a parameter, results from status quo bias. According to Bostrom and Ord,
The rationale of the Reversal Test is simple: if a continuous parameter admits of a wide range of possible values, only a tiny subset of which can be local optima, then it is prima facie implausible that the actual value of that parameter should just happen to be at one of these rare local optima.
Therefore, they argue, ‘… the burden of proof shifts to those who maintain that some actual parameter is at such a local optimum’ (Reference Bostrom and Ord2006, 665). The reversal test shifts the burden of proof by showing up the prima facie irrationality of favoring the status quo without providing good reason for doing so. Bostrom and Ord (Reference Bostrom and Ord2006) discuss four ways in which one might go about finding reasons that are sufficient to meet that burden of proof: appealing to evolutionary adaptation, appealing to transition costs, and appealing to arguments that they refer to as ‘the argument from person-affecting ethics’ and ‘the argument from risk’ (Reference Bostrom and Ord2006, 665–70).
Despite the widespread application of the reversal test, Bostrom and Ord’s (Reference Bostrom and Ord2006) discussion of it has been subjected to very little criticism, and what little it has been subjected to is mostly indirect criticism.Footnote 3 Nordmann (Reference Nordmann2007) is critical of the way in which Bostrom and Ord (Reference Bostrom and Ord2006) appeal to status quo bias to explain opposition to human cognitive enhancement.Footnote 4 He argues that Bostrom and Ord (Reference Bostrom and Ord2006) encourage their readers to assume that they must accept either that humans have a nature that just happens to be at optimal parameter values, or that humans should be cognitively enhanced (Nordmann Reference Nordmann2007, 38–39). But this is a false dichotomy. Nordmann (Reference Nordmann2007) understands the reversal test to play a part in Bostrom and Ord’s (Reference Bostrom and Ord2006) setup for this false dichotomy, but it only plays a part and it is not a necessary one. Nordmann would make the same criticism of their appeal to status quo bias to explain opposition to human cognitive enhancement even if it did not involve the reversal test. His major line of criticism of Bostrom and Ord (Reference Bostrom and Ord2006) is not a criticism of the reversal test. He does not argue that all uses of the reversal test encourage acceptance of the false dichotomy he warns us about, and he does not dispute that the reversal test can be an effective test of status quo bias.Footnote 5
Nebel (Reference Nebel2015) also discusses the reversal test. The main focus of his paper is status quo bias. Whereas most people, including, it appears, Bostrom and Ord (Reference Bostrom and Ord2006), simply assume that those who are in the grip of status quo bias are less than fully rational, Nebel (Reference Nebel2015) argues that status quo bias can be consistent with full rationality. In particular, he argues that on some theories of the rationality of preferences, especially (but not only) on subjective theories, it is not always irrational to be biased toward the status quo. Nebel is a proponent of the reversal test. But while Bostrom and Ord (Reference Bostrom and Ord2006) present the reversal test as a test for status quo bias, Nebel holds that the reversal test should be understood as a test for irrational status quo bias (Nebel Reference Nebel2015, 453–455).Footnote 6 As we will see, the reversal test works by identifying instances of status quo bias which are, prima facie, irrational, so the reversal test tests for both status quo bias and irrational status quo bias.
None of the authors who have critically discussed, or applied the reversal test appear to stop to consider whether and when it is an effective test of status quo bias, and whether it might not be refined. My concern in this paper is with these unaddressed issues. Whether or not the reversal test is best characterized as a test for status quo bias (which, following Bostrom and Ord (Reference Bostrom and Ord2006) will be my working assumption), or irrational status quo bias, will not matter for my purposes. I’ll take issue with Bostrom and Ord’s (Reference Bostrom and Ord2006) wording of the reversal test and with their characterization of the burden of proof it imposes. I’ll show that there is a way to try to meet that burden of proof which Bostrom and Ord do not consider. I’ll demonstrate that they do not succeed in showing that opponents of human cognitive enhancement – their key example – fail to meet the burden of proof imposed by the reversal test; and more generally, that they overestimate the range of cases in which the reversal test can be usefully applied. I will not, however, dispute that the reversal test can sometimes be used to show up the irrationality of preferring status quo options, and can therefore be an effective test of both status quo bias and irrational status quo bias.
In the next section of the paper, I’ll briefly explain why psychologists and cognitive scientists, such as Kahneman (Reference Kahneman2011), are convinced that ordinary thinking is prone to status quo bias. In the following section, I’ll explain how Bostrom and Ord (Reference Bostrom and Ord2006) think that the burden of proof imposed by the reversal test can be met. I’ll also introduce the key example that Bostrom and Ord (Reference Bostrom and Ord2006) use in their discussion of the reversal test: opposition to human cognitive enhancement (Reference Bostrom and Ord2006, 656). In the fourth, fifth, and sixth sections I discuss the problems that I discern in Bostrom and Ord’s (Reference Bostrom and Ord2006) discussion of the reversal test. Section Seven contains concluding remarks. Throughout the discussion, I’ll assume Bostrom and Ord’s definition of status quo bias: ‘… an inappropriate (irrational) preference for an option because it preserves the status quo’ (Reference Bostrom and Ord2006, 658).
2. Status quo bias
Psychologists and cognitive scientists who work on ‘status quo bias’ do not claim that there is a single factor that biases our thinking in favor of the status quo. Rather, they hold that a persistent bias in our thinking, toward the status quo, results from the combined influence of a number of distinct psychological effects. One of these is the pervasive ‘endowment effect, ’ which leads people to ascribe additional value to items when they come to regard these as their possessions. To exchange what we possess for an alternative involves departing from the status quo, so by deterring people from making exchanges, the endowment effect causes status quo bias.Footnote 7
Another contributor to status quo bias is loss aversion. Many of the opportunities we are presented with are ‘mixed, ’ involving the possibility of gain and the risk of loss. It seems that when we decide to accept or reject such opportunities, we typically do not weigh potential losses and potential gains equally. We feel the pain of a loss more than we feel the satisfaction of a gain. So, we tend to avoid mixed opportunities unless these are markedly in our favor (Kahneman Reference Kahneman2011, 284). Our tendency to forego opportunities that are only marginally in our favor leads us to avoid making many changes to our circumstances which we would otherwise make and thereby contributes to status quo bias.Footnote 8 There are also various other psychological effects that contribute to status quo bias including omission bias (Ritov and Baron Reference Ritov and Baron1992) and regret avoidance (Anderson Reference Anderson2003).Footnote 9
3. The burden of proof imposed by the reversal test and opposition to human cognitive enhancement
In Section Five, I will dispute Bostrom and Ord’s claim that in order to meet the burden of proof imposed by the reversal test good reason needs to be provided to suppose that we are currently at a local optimum with respect to the value of a parameter (Reference Bostrom and Ord2006, 665). However, they are surely right to think that good reasons for preferring the current value of a parameter to changes in either direction should be required of the proponent of the status quo. Status quo bias provides a prima facie explanation of a preference for the current value of a parameter, so the reasons for preferring the status quo to changes in either direction should be the basis for an explanation that is at least as good an explanation as a competing explanation of a preference for the current value of a parameter that involves appeal to status quo bias. Otherwise, the most plausible conclusion to draw is that status quo bias is the best available explanation of a preference for the current value of a parameter.
Bostrom and Ord (Reference Bostrom and Ord2006) work through the example of opposition to human cognitive enhancement, in some detail, to illustrate the application of the reversal test. They ask us to suppose that a medically safe, affordable means of enhancing human cognition has been developed. They then consider whether or not this new technology should be used. They set aside deontological constraints and ask, ‘do we have reason to believe that the long-term consequences of human cognitive enhancement would be, on balance, good?’ (Reference Bostrom and Ord2006, 656). Their answer is yes, but, as they note, many people’s answer is no (Bostrom and Ord Reference Bostrom and Ord2006, 662–663). Many believe that there are unacceptable risks involved in allowing cognitive enhancements to be used, even if these are medically safe and affordable. It has been suggested that in a world in which inheritable genetic alterations were widely available, humanity might split into two sub-species, a post-human sub-species and an unenhanced sub-species; and that members of the post-human sub-species might enslave or exterminate unenhanced humans (Annas, Andrews, and Isasi Reference Annas, Andrews and Isasi2002, 154).Footnote 10 It has also been suggested that enhanced humans might lack a property that Sandel (Reference Sandel2004) refers to as ‘openness to the unbidden.’ And it has been suggested that in a world in which people could be genetically enhanced, the ability of parents to determine the genes of their children might lead to various undesirable consequences, as parents are liable to make shortsighted choices on behalf of those children (Fukuyama Reference Fukuyama2002, 93–100).Footnote 11
The long-term consequences of human cognitive enhancement might be deleterious, all things considered, but it is also possible that they will be beneficial for us, overall. Nick Bostrom imagines that overall beneficial consequences have been obtained from cognitive enhancement in his ‘Letter from Utopia.’ In this letter ‘your possible future self’ writes from a future in which post-humans enjoy a highly fulfilling life, every second of which ‘… is so good that it would blow our minds had their amperage not been previously increased’ (Bostrom Reference Bostrom2010, 8). Bostrom and others in the pro-enhancement camp do not deny that there is some chance that things will not turn out as they hope, and do not deny that the widespread use of cognitive enhancement techniques could lead to one or more of the undesirable outcomes that opponents of enhancement warn us about. Similarly, most of those in the anti-enhancement camp do not deny that there is some chance that the negative outcomes that they worry about will not transpire, or will not be as worrisome as they anticipate. The disagreement is over competing estimates of the likely overall balance of costs and benefits arising from human cognitive enhancement.
If we could perform an accurate cost-benefit analysis, determining the various harms that human cognitive enhancement is liable to cause, and the likelihood that these will occur, and weigh this against a determination of the various benefits that human cognitive enhancement is liable to provide, along with accurate estimates of the likelihood that these will occur, then perhaps we could settle the debate between opponents and proponents of human cognitive enhancement. But we cannot do this. It is very difficult to determine the harmfulness of some of the potential hazards of human cognitive enhancement. How harmful is a loss of openness to the unbidden? It’s very hard to say. Also, we cannot currently acquire accurate information about the likelihood that cognitive enhancement will lead to the various benefits and harms that have been mooted occurring. People’s current convictions about the likelihood of these occurring are the result of intuitive judgments; and these are grounded on incomplete information and subject to cognitive biases (Gilovich, Griffin, and Kahneman Reference Gilovich, Griffin and Kahneman2002; Kahneman Reference Kahneman2011). This is not a point that Bostrom and Ord (Reference Bostrom and Ord2006) dispute. They describe assessment of the overall consequences of human cognitive enhancement, as well as other assessments of consequences that impact on ‘human lives and social systems, ’ as judgments that ‘… rely also – crucially and unavoidably – on subjective, intuitive judgment’ (Reference Bostrom and Ord2006, 657). Despite acknowledging the crucial role that subjective, intuitive judgment plays in assessment of the overall consequences of human cognitive enhancement, Bostrom and Ord assure us that application of the reversal test shows that objections to human cognitive enhancement are affected by status quo bias and that ‘When the bias is removed, the objections are revealed as extremely implausible’ (Reference Bostrom and Ord2006, 658).
Opponents of human cognitive enhancement are hardly ever in favor of cognitive dis-enhancement. So, it seems that they are committed to defending the view that average human cognitive abilities just happen to be at a local optimum, which seems prima facie unlikely, given all the different values at which average human cognitive abilities might be found. So, the onus is upon the defender of the status quo, with respect to human cognitive abilities to demonstrate that we are at a local optimum. Bostrom and Ord attempt to apply all four of the possible ways of meeting the burden of proof imposed by the reversal test, which they identify, to the case of opposition to human cognitive enhancement (Reference Bostrom and Ord2006, 665–672). They argue that none can be successfully applied to this case. I’ll go on to dispute the conclusion of this argument in Section Five, but first I’ll raise an issue about their wording of the test and consider a way of meeting the burden of proof it imposes that Bostrom and Ord (Reference Bostrom and Ord2006) do not consider.
4. Backfiring and the reversal test revised
Bostrom and Ord’s question: ‘do we have reason to believe that the long-term consequences of human cognitive enhancement would be, on balance, good?’ (Reference Bostrom and Ord2006, 656) is ambiguous. They might be asking us to imagine a future in which many humans have been cognitively enhanced to a significant degree and then asking us to try to judge whether or not life in this future would be preferable, all things considered, to life now. Or they might be asking us to imagine all the possible consequences of allowing significantly many humans to attempt to become cognitively enhanced, and then asking whether, when we weigh the likelihood of any or all of these consequences eventuating, we think it is better to allow, or to try to prevent, human cognitive enhancement? On the second interpretation of their question, a form of risk will need to be considered that it is not necessary to consider on the first interpretation. This is the risk that our attempts to enhance ourselves could fail and cause unintended deleterious outcomes – which is to say that they could backfire – and lead to a future that is neither one in which cognitive enhancements are widely used, nor one that is superior to the status quo.
Bostrom and Ord do not consider the possibility of backfiring, so it seems reasonable to understand them as construing their question narrowly. But it is possible that our attempts to create a cognitively enhanced future for ourselves will backfire and this possibility concerns some opponents of human enhancement. We can try to use the reversal test to investigate whether or not intuitions about the significance of this possibility are driven by status quo bias, but it will require revisions to the wording of the test for us to be able to do so. These revisions should be made anyway, because any attempt to shift the value of a parameter in a particular direction could backfire, and defenders of the status quo will be able to appeal to the possibility of backfiring to defend the status quo in many different contexts. A reversal test that fails to make room for this possibility is significantly less useful than one that does.
Bostrom and Ord’s wording of the reversal test includes the phrase: ‘When a proposal to change a certain parameter is thought to have bad overall consequences, consider a change to the same parameter in the opposite direction’ (Reference Bostrom and Ord2006, 664). Their wording is, in effect, an invitation to consider the consequences of a proposed change to a parameter that has succeeded, along with the consequences of a change that has taken place in the opposite direction. What we should be trying to consider are all the possible consequences of attempting to change a parameter in one direction, and all the possible consequences of attempting to change it in the other direction, including the possibilities of attempts to change the parameter in either direction failing. Here is a revised version of the reversal test that allows for the possibility that attempts to change the value of a parameter may fail (while taking account of earlier revisions), and so is broad enough to allow for consideration of the possibility of backfiring. New wording has been underlined:
Reversal Test (Revision One): When an attempt to change a certain parameter is thought to have bad overall consequences, consider an attempt to change the same parameter in the opposite direction. If this is also thought to have bad overall consequences, then the onus is on those who reach these conclusions to explain why our position cannot be improved through attempts to change this parameter. If they are unable to do so, then we have reason to suspect that they suffer from status quo bias.
The propensity of human schemes aimed at improving society to backfire is a persistent theme in conservative political thought (Buchanan Reference Buchanan2011, 146). Given that many of the opponents of human enhancement are politically conservative, it should not be surprising that some are concerned about potential backfiring. Both Fukuyama (Reference Fukuyama2002, 3–10) and Kass (Reference Kass2003, 11) worry that attempts to enhance humans may inadvertently lead to the creation of a society that resembles the dystopic future society depicted in Aldous Huxley’s Brave New World (Reference Huxley[1932] 2007). This looks to be a concern about an attempt to create an utopian society backfiring and leading to the creation of a society in which a future world government uses scientific techniques to cognitively dis-enhance large sections of the population.
Brave New World is a society controlled by a world government in which selective breeding, and interventions in fetal development, lead to the creation of distinct human castes – Alphas, Betas, Gammas, Deltas, and Epsilons – bred to play specific roles in a world economy. The developing fetuses of Gammas, Deltas, and Epsilons are cognitively dis-enhanced by being deprived of oxygen and doused with alcohol to ensure that they have artificially low levels of intelligence (Reference Huxley[1932] 2007, 4–11). The sperm and ova that is used to create Alphas and Betas is selected to ensure that their intelligence is at a suitable level for their designated economic roles, but there is no attempt to manipulate the genetic material of these castes to produce higher-than-natural levels of intelligence mentioned in the book. Brave New World is far from the scientifically advanced world that Nick Bostrom imagines in his Letter from Utopia (Reference Bostrom2010). Scientific advances may have led to Brave New World, but it is a world with a government that prohibits serious scientific research, on the grounds that scientific breakthroughs could threaten social stability (Huxley Reference Huxley[1932] 2007, 198).
Neither Fukuyama (Reference Fukuyama2002) nor Kass (Reference Kass2003) try to explain how the cognitive enhancement of humans might backfire, leading to a Brave New World-style scenario in which systematic cognitive dis-enhancement becomes commonplace; and Huxley himself does not spell out how he supposes that our world might become the dystopia of his imagination.Footnote 12 Perhaps Fukuyama and Kass fear that after the means to alter human cognitive capacities is developed, it may fall into the hands of a malevolent sub-group of humanity who use it not to enhance anyone, but as part of a scheme to create an oligarchic world government, dis-enhancing the cognitive capacities of the people who might object in the process?
Could an argument against human cognitive enhancement, based on fears that attempt to cognitively enhance humans will inadvertently lead to a Brave New World-style dystopic future, be adapted to become an argument that meets the (revised) reversal test? To apply the (revised) reversal test we need to consider the consequences of attempts to either increase or decrease the cognitive powers of significantly many humans. If it is considered reasonable to fear that attempts to enhance cognitive capacity will lead to a dystopic future, in which humans fall under the control of a world government that manipulates individual cognitive capacities, then it also seems reasonable to fear that attempts to dis-enhance cognitive capacity could lead to a similar dystopic future involving a world government manipulating individual cognitive capacities. This might involve successful cognitive dis-enhancement, as in Brave New World, or it might involve attempts to dis-enhance backfiring, but still leading to power being concentrated in the hands of a malevolent oligarchy. So, it looks like at least one of the arguments mounted against cognitive enhancement, based on the fear of backfiring, can be adapted to become an argument for the cognitive status quo, which looks like it might meet the (revised) reversal test. More generally, it seems that any attempt to shift the value of a parameter away from the status quo is susceptible to the possibility of backfiring, and appeals to this possibility have the potential to meet the burden of proof imposed by the (revised) reversal test.
5. The argument from risk
Bostrom and Ord introduce the reversal test and immediately specify what needs to be done to meet the burden of proof that it imposes: provide ‘some good reason’ to think that we are at a local optimum with respect to a parameter (Reference Bostrom and Ord2006, 665). They then go on to sketch four ways in which one might meet the burden of proof imposed by the reversal test. When they come to discuss meeting that burden of proof by employing the ‘argument from risk, ’ it becomes apparent that Bostrom and Ord (Reference Bostrom and Ord2006) are willing to consider a way of meeting that burden that does not involve providing good reason to suppose that we are at a local optimum with respect to a parameter. They write:
Even if it is agreed that we are probably not at a local optimum with respect to some parameter under consideration, one could still mount an argument from the risk against varying the parameter. If it is suspected that the potential gains from varying the parameter are quite low and the potential losses very high, it may be prudent to leave things as they are. (Reference Bostrom and Ord2006, 669)
However, they go on to assert, mere uncertainty about consequences is not good grounds for sticking with the status quo, ‘Only if the expectation value of the hypothetical negative results is larger than the expectation value of the hypothetical positive results does the uncertainty favor the preservation of the status quo.’ (Reference Bostrom and Ord2006, 668–669).
What Bostrom and Ord (Reference Bostrom and Ord2006) appear to be suggesting is that if it can be shown that the potential harms of increasing the value of a parameter outweigh the potential benefits, and if it can be shown that the potential harms of decreasing the value of a parameter outweigh the potential benefits, then we have sufficient reason to retain the status quo; and we do not need to also demonstrate that we are at a local optimum with respect to a parameter. So, in effect, they are proposing a less demanding burden of proof than the requirement that we demonstrate that we are at a local optimum with respect to the value of a parameter. Bostrom and Ord (Reference Bostrom and Ord2006) appear to intend that this less demanding burden of proof only be applied when a version of the ‘argument from risk’ is employed. But the less demanding burden of proof seems appropriate to apply whenever we are employing the (either revised or unrevised) reversal test. If we can demonstrate that the potential harms of attempting to increase the value of a parameter outweigh the potential benefits, and the potential harms of attempting to decrease the value of a parameter outweigh the potential benefits, then we have explained why our position cannot be improved through changes to a parameter, which is what the reversal test asked of us (Bostrom and Ord Reference Bostrom and Ord2006, 664–665).
Bostrom and Ord’s (Reference Bostrom and Ord2006) less demanding burden of proof is highly relevant to the discussion of their key example, opposition to human cognitive enhancement. Many opponents of human cognitive enhancement think that the potential harms of attempts to cognitively enhance humans outweigh potential benefits; and they take it as given that the potential harms of attempting to cognitively dis-enhance humans outweigh potential benefits. If they are able to demonstrate that they are right about both of these claims, then it looks like they will be able to meet the burden of proof imposed by the (either revised or unrevised) reversal test, without also having to demonstrate that human cognitive capacity fortuitously happens to be at a (locally) optimal level.
I’ll look at Bostrom and Ord’s (Reference Bostrom and Ord2006) discussion of the risks involved in human cognitive enhancement in some detail in the next section, but before I do so I want to raise another issue about their wording of the reversal test that is particularly relevant to a discussion of the argument from risk. Their wording of the reversal test places the onus on those who favor the status quo with respect to a parameter ‘to explain why our position cannot be improved through changes to this parameter’ (Reference Bostrom and Ord2006, 664). But appeal to the argument from risk to meet the burden of proof imposed by the reversal test does not involve showing that our position cannot be improved by attempting to make changes to a parameter, it involves showing that our position is more likely to be made worse than improved through attempting to make changes to the parameter (in either direction). Therefore, the reversal test should be revised as follows. Again, changes from the original wording are indicated via underlining:
Reversal Test (Revision Two): When an attempt to change a certain parameter is thought to have bad overall consequences, consider an attempt to change the same parameter in the opposite direction. If this is also thought to have bad overall consequences, then the onus is on those who reach these conclusions to explain why our position is more likely to be made worse than improved through attempts to change this parameter. If they are unable to do so, then we have reason to suspect that they suffer from status quo bias.
6. The risks involved in cognitively enhancing humans
As we saw earlier, Sandel (Reference Sandel2004), Annas, Andrews, and Isasi (Reference Annas, Andrews and Isasi2002), and Fukuyama (Reference Fukuyama2002) all argue, in different ways, that we should refrain from cognitively enhancing humans because of the risks of deleterious consequences. They all offer arguments against human cognitive enhancement, rather than arguments for the status quo, with regard to human cognitive ability, but their respective arguments could easily be transformed into arguments for the status quo, by being conjoined with arguments against human cognitive dis-enhancement. Human cognitive dis-enhancement is usually regarded as a very unattractive option, so such arguments should not be hard to provide.
As we have already noted, Bostrom and Ord (Reference Bostrom and Ord2006) suggest that what is needed to determine whether or not a particular instance of the ‘argument from risk’ meets the burden of proof imposed by the reversal test is an assessment of the overall consequences of departing from the status quo with respect to the value of a parameter. There will be many contexts in which we will be unable to provide fully objective assessments of these consequences and will have to settle for somewhat subjective ones. Assessments of the potential costs and benefits involved in human cognitive enhancement are such cases. According to Bostrom and Ord, as we saw earlier, our assessments of all risks to ‘human lives and social systems’ rely on subjective intuitions about risks, which cannot be objectively weighed (Reference Bostrom and Ord2006, 657).
Having explained that the way to assess the argument from risk is to weigh all of the potential costs and benefits of human cognitive enhancement and dis-enhancement, one might expect Bostrom and Ord to attempt an assessment of the overall costs and benefits of human cognitive enhancement and dis-enhancement, but they do not do this. Instead, they present us with an argument for the conclusion that we may have systematically underestimated the expected benefits of human cognitive enhancement, as well as a listing of some of its most significant potential benefits (Reference Bostrom and Ord2006, 668–670). They are not clear about their motives, but it seems plausible to think that they intend to shift the balance of other people’s assessments of the expected costs and benefits of human enhancement by presenting their argument and their listing.
Bostrom and Ord’s argument for the conclusion that the benefits of human cognitive enhancement may have been systematically underestimated involves a thought experiment aimed at showing how hard it is for us to conceive of many of the potential benefits of cognitive enhancement (Reference Bostrom and Ord2006, 69). They ask us to consider the difficulties faced by a tribe of Australopithecus debating whether or not to enhance their intelligence to modern human levels. It would be extremely difficult for the Australopithecus to conceive of many of the benefits of human-level intelligence, such as having the abilities to create and appreciate art, literature, music, and poetry. Just so, they argue, we find it hard to foresee the various benefits of significantly enhanced cognition (Bostrom and Ord Reference Bostrom and Ord2006, 669). The significant potential benefits of human cognitive enhancement they list are enabled by increased cognitive capacity. With increased cognitive capacity, we are more likely to be able to find cures to diseases, answer scientific questions, invent solutions for poverty, and solve environmental problems than we would be if our cognitive powers remain at their current level (Bostrom and Ord Reference Bostrom and Ord2006, 669).
Having presented the above considerations, Bostrom and Ord acknowledge that some of their opponents will not accept that risks should be managed by weighing the potential costs and benefits of change (Reference Bostrom and Ord2006, 669). Some urge a risk-averse approach under which (consistent with loss aversion) potential costs weigh more heavily than benefits.Footnote 13 Bostrom and Ord (Reference Bostrom and Ord2006) think that even if we adopt such a risk-averse approach the case for human cognitive enhancement is still strong. This is because, they argue, increased human cognitive capacity can enable us to reduce various serious threats to humanity more effectively than we would be able to otherwise (Reference Bostrom and Ord2006, 670). All things being equal, we are more likely to find cures to new pandemic diseases, more likely to be able to prevent asteroids from colliding with the Earth, more likely to be able to prevent invasion by hostile aliens from other solar systems, and so on, if we are cognitively enhanced than if we are not.
Bostrom and Ord’s discussion of cognitive biases that are liable to lead us to underestimate the likely benefits of human cognitive enhancement is relevant to the question of whether or not opponents of human cognitive enhancement can meet the burden of proof imposed by the (revised) reversal test, however, cognitive biases cut both ways. Bostrom and Ord’s debating Australopithecus would have a hard time conceiving of the benefits of art, literature, music, and poetry. But they would also have a hard time conceiving of many of the dis-benefits of modern life. For example, they would be unlikely to conceive of the various diseases that were not a threat to humans until we began to live in concentrated urban communities, and commenced farming, such as tuberculosis, smallpox, and influenza.Footnote 14 The debating Australopithecus would also have a hard time anticipating the many health problems caused by modern diets and would probably find some of the frustrating features of modern life, such as the difficulties involved in having to complete many years of formal education, having to dealing with modern bureaucracies, and having to function in a modern workplace, simply unimaginable. It’s not clear that if our debating Australopithecus were able to comprehend all of the potential benefits and harms of having the intelligence of modern humans that they would choose to have their intelligence increased.Footnote 15 Mutatis mutandis, if modern humans were able to comprehend all of the potential benefits and harms involved in becoming cognitively enhanced post-humans.Footnote 16
The other two lines of argument that Bostrom and Ord (Reference Bostrom and Ord2006) develop concern benefits to humanity that we are more likely to enjoy if we are cognitively enhanced than if we remain unenhanced, and harms to humanity that we are more likely to be able to avoid. Bostrom and Ord’s (Reference Bostrom and Ord2006) discussion of these lines of argument is shot through with a sunny-eyed optimism about the propensity of cognitively enhanced individuals to devote their energies to providing benefits to, and reducing the risk of harms to, the rest of humanity. If we do not share this optimism then the conclusions that Bostrom and Ord (Reference Bostrom and Ord2006) draw look very shaky. Unless they are provided with the right incentives, selfish cognitively enhanced people will not be motivated to cure diseases that do not endanger them, share scientific knowledge, alleviate the poverty of others, and protect parts of the environment that they do not interact with. Furthermore, selfish cognitively enhanced people may create and spread diseases that harm or kill others, prevent scientific knowledge from being shared, cause others to live in poverty, and cause damage to the environment, if it is in their interest to do these things. They can be relied on to use their enhanced intellectual powers to prevent pandemics, asteroid strikes, and alien invasions, when these place them in danger, but if they can protect themselves from such threats without also protecting others, then they are liable to do so.Footnote 17
The result of the thought experiment involving the debating Australopithecus is inconclusive. The other two lines of argument for human cognitive enhancement can only be accepted as good arguments if we accept Bostrom and Ord’s optimistic implicit assumption about the propensity of the cognitively enhanced to use their cognitive powers to aid the rest of humanity. As we have already noted, many of the opponents of human enhancement, such as Fukuyama (Reference Fukuyama2002) and Kass (Reference Kass2003), are political conservatives and are likely to reject this optimistic assumption. Traditionally, political conservatives have taken a jaundiced view of human motivation, stressing the propensity of selfish motives and emotions to undermine pro-social tendencies (Buchanan Reference Buchanan2011, 147).Footnote 18 So, many opponents of human cognitive enhancement are unlikely to be swayed by Bostrom and Ord’s (Reference Bostrom and Ord2006) counterarguments to their arguments from risk against human cognitive enhancement. These opponents of human cognitive enhancement are unlikely to be able to persuade Bostrom and Ord, or other proponents of human enhancement, to shift their position either. The competing assessments that proponents and opponents of human cognitive enhancement make, about the likely balance of costs and benefits involved with attempts to cognitively enhance humans, are very different and because these assessments rely significantly on subjective intuition there is no clear way to adjudicate between them.
We lack accurate data to ground an assessment of the potential benefits of human cognitive enhancement. We also lack accurate data to ground an assessment of the potential costs involved with successful attempts to cognitively enhance humans, as well as the potential costs associated with attempts to cognitively enhance humans that backfire. Until such time as we have accurate data, to ground our assessments of all of the potential costs and benefits involved in attempts to cognitively enhance humans, proponents and opponents of human cognitive enhancement are going to have to continue to rely on their subjective intuition about costs and benefits, and disputes between the two sides are not going to be rationally resolved.
Do contemporary opponents of human cognitive enhancement do enough to meet the burden of proof imposed by the (revised) reversal test right now? Those who have pro-enhancement intuitions will be likely to conclude that opponents of human cognitive enhancement do not do enough to meet the burden of proof imposed by the (revised) reversal test. However, those who have anti-enhancement intuitions are likely to think that at least some opponents of human cognitive enhancement succeed in meeting the burden of proof imposed by the (revised) reversal test. They will be convinced that the risks of harms resulting from (both successful and backfiring) attempts to cognitively enhance humans outweigh potential benefitsFootnote 19; and also convinced that the risks of harms resulting from (both successful and backfiring) attempts to cognitively dis-enhance humans outweigh potential benefits. Because disagreement about whether the burden of proof imposed by the (either revised or unrevised) reversal test has been met or not, comes down to competing subjective intuitions, it will not be amenable to rational resolution. Contra Bostrom and Ord (Reference Bostrom and Ord2006), application of the (either revised or unrevised) reversal test does not provide a clear result that tells against opponents of human cognitive enhancement and it does not reveal that objections to human cognitive enhancement are ‘extremely implausible.’ It produces different results depending on whether opponents or proponents of human cognitive enhancement run the test.
A general lesson we can take away from Bostrom and Ord’s application of the reversal test to opposition to human cognitive enhancement is this: it is pointless to apply the (either revised or unrevised) reversal test to circumstances in which there is significant disagreement about the risks involved in varying the value of a parameter away from the status quo, and no prospects of rationally resolving that disagreement, as is the case when disagreement turns on competing subjective intuitions about those risks. In such circumstances, there cannot be an objective, impartial determination as to whether the burden of proof imposed by the test has been met or not.
7. Concluding remark
I have argued for four conclusions in this paper. First, that the wording of the reversal test should be revised as per the wording set out at the end of Section Five. Second, that the burden of proof needed to meet the (either revised or unrevised) reversal test is less demanding than that described by Bostrom and Ord (Reference Bostrom and Ord2006, 665). My clarification of the burden of proof needed to meet the reversal test was provided in Section Five. Third, that there is an additional way to meet the burden of proof imposed by the test, which Bostrom and Ord (Reference Bostrom and Ord2006) fail to recognize. This is by appealing to backfiring, as was discussed in Section Four. Fourth, the (revised) reversal test should not be applied to circumstances, such as opposition to human cognitive enhancement, where there is significant disagreement about the risks involved in varying the value of a parameter away from the status quo, and where that disagreement is not amenable to rational resolution. The argument for this conclusion was developed in Section Six.
I have been critical of Bostrom and Ord (Reference Bostrom and Ord2006). But despite the various shortcomings I have discerned in their (2006) paper, we should not lose sight of what Bostrom and Ord have achieved. The reversal test needs to be refined, and it needs to be applied more judiciously than they recognize, but nevertheless, it can be an effective test of status quo bias. Given the propensity of contemporary philosophers to try to explain away preferences for status quo options by appealing to status quo bias, it should be recognized that Bostrom and Ord (Reference Bostrom and Ord2006) have made an important contribution to philosophy by developing the original version of the much-used reversal test.
Funding
This work was supported by the Australian Research Council under Discovery Grant [DP130103658].
Acknowledgment
Thanks to Matthew Kopec, Neil Levy, Toby Ord, Russell Powell, and two anonymous referees for helpful comments on earlier drafts of this paper as well as audiences at the University of Oxford, Charles Sturt University, the University of Helsinki, and the Australasian Association of Philosophy annual conference in 2015.
