A Moderate Defence of the Use of Thought Experiments in Applied Ethics
- First Online:
- Cite this article as:
- Walsh, A. Ethic Theory Moral Prac (2011) 14: 467. doi:10.1007/s10677-010-9254-7
- 268 Views
Thought experiments have played a pivotal role in many debates within ethics—and in particular within applied ethics—over the past 30 years. Nonetheless, despite their having become a commonly used philosophical tool, there is something odd about the extensive reliance upon thought experiments in areas of philosophy, such as applied ethics, that are so obviously oriented towards practical life. Herein I provide a moderate defence of their use in applied philosophy against those three objections. I do not defend all possible uses of thought experiments but suggest that we should distinguish between legitimate and illegitimate uses. Their legitimate uses are determined not so much by the modal content of any actual thought experiment itself, but by the extent to which the argument in which it is nested follows basic tenets of informal logic and respects the fundamental contingency of applied ethical problems. In pursuing these ideas, I do not so much provide a set of criteria for their legitimate use, but more modestly present two significant ways in which their use can go awry.
KeywordsThought experimentsApplied ethicsMethod
1 Introduction: Should We Stop Imagining?
Thought experiments have played a pivotal role in many debates within ethics—and in particular within applied ethics—over the past 30 years. A case in point would be Bernard Williams’ story of Poor Jim who is forced to choose between killing one Amazon Indian to prevent twenty being murdered (Smart and Williams 1973, 97–99). Or think of Phillipa Foot’s trolley problem where an out-of-control vehicle is racing down a tram track towards five people working on the track and there is nothing you can do to stop it; however you can steer the tram so that it shifts onto another line on which there is only one person working (Foot 1978). Would you kill one person in order to save five? For a more bizarre example of the genre think of Judith Jarvis Thomson’s article ‘A Defense of Abortion’ in which various forms of abortion are defended through a series of thought experiments involving, amongst other things, people-seeds that grow in one’s carpet and rapidly expanding infants who crush their mothers to death (Thomson 1971). Thought experiments such as these have been highly influential on the way topics in applied ethics have been pursued.
This increased occurrence of thought experiments can be viewed as part of a more general movement in analytic philosophy away from the conceptual analysis of the 1950’s and 1960’s. For instance, in ethics it is now far less common to find philosophers analysing the meaning of notions such as ‘goodness’ and ‘justice’. Gerald Massey in ‘Backdoor Analyticity’ goes so far as to suggest that thought experiments are contemporary analytic philosophy’s main modus operandi, the modern surrogate for meaning analysis (Massey 1991). We shift from conceptual analysis to what we might think of as a ‘possible worlds analysis’ (Fodor 2004). In order to test our general principles and claims we need to consider whether they are true in all possible worlds and to determine this we need to imagine what other possible worlds there might be. Thought experiments obviously play an important role in such an endeavour.
Nonetheless, despite their having become a commonly-employed philosophical tool, there is something odd about the extensive reliance upon thought experiments in areas of philosophy, such as applied ethics, that are so obviously oriented towards practical life. Given the practical orientation of applied ethics, why discuss people seeds and their like? I consider three lines of objection to their use that follow from such concerns. First, it might be objected that modally outlandish stories such as Thomson’s cannot have a legitimate role to play in a form of inquiry so clearly oriented towards actual social life. Second, it might be argued that thought experiments are typically so under-described as not to bear, except in a tenuous sense, on the complicated problems dealt with in applied ethics. Finally, it might be objected that there are some things we simply should not contemplate and that all too often these are the kinds of things such thought experiments ask us to think about. Might these objections provide grounds for constraining the kinds of thought experiments that are admissible in applied ethics? Might they provide grounds for rejecting thought experiments altogether as legitimate tools of philosophical analysis? Should we agree with Henry Shue’s more general suggestion that artificial cases make bad ethics? (Shue 1978)
Herein I provide a moderate defence of their use in applied philosophy against those three objections. I do not defend all possible uses of thought experiments but suggest that we should distinguish between legitimate and illegitimate uses. As we shall see below, their legitimate uses in applied ethics are determined not so much by the modal content of any actual thought experiment itself, but by the extent to which the argument in which it is nested follows basic tenets of informal logic and respects the fundamental contingency of applied ethical problems. (Respecting that contingency involves inter alia avoiding what I shall later refer to as the ‘Fallacy of Changing the Subject’.) In pursuing these ideas, I do not so much provide a set of criteria for their legitimate use, but more modestly present two significant ways in which their use can go awry.
2 Thought Experiments, Arguments and Applied Ethics
Tamar Gendler, whilst considering thought experiments in the philosophy of science, suggests that: “to conduct a thought experiment is to make a judgement about what would be the case if the particular state of affairs described in some imaginary scenario were actual” (Gendler 1998, 398). Following Gendler’s line of reasoning, we might define a thought experiment in ethics to consider what would be the case morally if the particular state of affairs described in the imaginary scenario were actual. In effect we are asked to determine the moral status of that hypothetical state of affairs. That much is fairly uncontroversial.
How best to understand the relationship between thought experiments and arguments is a subject of some considerable dispute. In the philosophy of science there has been considerable debate about whether we should regard a thought experiment as an argument. Some philosophers, such as John Norton, have argued that thought experiments are nothing more than dressed-up arguments (Norton 1991, 1996). Others, such as Rachel Cooper, argue that there are many thought experiments for which this simply cannot be true (Cooper 2005, 332). Cooper uses Hume’s example of the missing shade of blue to substantiate her claim. In A Treatise of Human Nature Hume asks us to consider whether someone could imagine what a missing shade of blue would look like without ever having seen it (Hume 1978, 6). On its own, this does not sound like an argument. Equally, if we turn from the philosophy of science to ethics we see that thought experiments on their own will not count as arguments. To see this think of Foot’s trolley problem which we mentioned earlier. Without the background context of the debate over the moral permissibility of consequentialism, the trolley problem might appear to be nothing more than a tale (perhaps of an ironic nature) regarding the different fates that sometimes befall people. On its own, and without some context, it is certainly not an argument. I shall assume for the rest of this paper that thought experiments are not arguments, but simply play a role within arguments or as Soren Haggqvist puts it they are “operations providing premises for arguments” (Haggqvist 1996, 136).
Within ethical arguments thought experiments typically function as un-stated premises of an argument or else as evidence for the premise of an argument. In other cases, a thought experiment functions as a mechanism for framing an argument. Of course, there will be occasions where we draw conclusions simply from a story told to us, but that is a consequence of the surrounding context in which it is embedded. In such cases we should regard the thought experiment as the only explicitly articulated element in what is an enthematic argument.
My intention here is to identify ways in which the use of thought experiments might be illicit or pernicious (Gale 1991). If we accept the idea that thought experiments function either as elements of an argument or to frame an argument, then there are two obvious ways in which thought experiments might go wrong. First their use is illicit when they are employed in arguments where the conclusions drawn from them are unwarranted on the basis of the evidence provided by the thought experiment and surrounding premises. Second, the use of thought experiments is illicit when they frame an argument in a misleading way. In either case we might say that the use of the thought experiment is pernicious or illegitimate.
Being illegitimate should not be confused with the idea of imaginative failure that covers cases where a thought experiment does not have any imaginative grip. For instance, if you ask me to imagine a square circle, I have no real sense of what it is I am meant to be imagining. Rachel Cooper makes a closely related criticism in her discussion of Bernard Williams’ thought experiment that asks us to imagine people who split like amoeba. She suggests, “we are unable to answer the necessary ‘what if?’ questions” required by this story and, hence it fails as a thought experiment (Cooper 2005, 342).1 Equally, the moral evaluation required by some thought experiments will be beyond our imaginative capacities.2
A consequence of this approach is that how we determine the legitimacy of the use of a thought experiment is not to be understood in terms of any intrinsic features of that experiment but in terms of the role it plays in the argument in which it is embedded. Usage will be illicit when the conclusions drawn from it are not warranted or when it frames a questions in a misleading way. It is not that we can determine from the content of the thought experiment alone that it is necessarily illicit; rather such evaluations will depend upon the argumentative context. What I am gesturing towards is what we might think of as an argumentative pragmatics of thought experiment whereby the argumentative context determines the legitimacy of any particular use.3 Unlike some other commentators in the area, such as Haggqvist, I make no attempt to provide a single characterisation of what makes for a successful thought experiment (Haggqvist 1996,152). In line with my ‘pragmatics’ approach, any systematic work here would begin with the identification of common fallacies in the use of thought experiments.4
In considering the relevance of this to applied ethics we should note at the outset that the argumentative context in this sub-discipline differs from that of the greater part of philosophical inquiry. Applied ethics primarily involves the application of moral philosophy to actual everyday solutions (Almond 1995). In applied ethics the aim is to arrive at determinate conclusions about the correct course of action or the correct public policy in the world we currently inhabit: we are then aiming to determine the rightness and wrongness of a restricted domain of action. Accordingly, the choice-set we confront is constrained by contingent features of the world, such as what we know about current human capacities, what we know about how human beings are likely to act and what technological and social constraints are before us and so on. The conclusions we draw are thus less concerned with what would be the correct course of action in all possible worlds—that is, the necessarily correct course of action—than with what is the correct course in our world as it presently stands.
This distinguishes it from much philosophy—especially metaphysics—where philosophers seek that which is necessarily true. In applied ethics, given its practical orientation, contingent features of the world are more significant than they would be in many other areas. This does not preclude counterfactual reasoning within applied ethics, for as we shall see such reasoning often has an important role to play in the chains of reasoning that lead us to our conclusions, but it does constrain the range of worlds to which our conclusions must accommodate. In this, whilst perhaps not sui generis, applied ethics is distinct from mainstream philosophical theorising.
3 A Taxonomy of Functions of Thought Experiments in Applied Ethics
Let us assume that thought experiments are typically employed as elements within arguments or in framing an argument. If this is true then it should be possible to categorise them in terms of the differing roles they play in an argument. We can identify four legitimate roles that thought experiments play. Whilst it is a partial list, nevertheless it does capture what I take to be the central cases.5
First, some thought experiments function as either counter-examples or reductio ad abdsurda in philosophical disputations. One philosopher might present a theory or a definition that is intended to be either necessarily or universally true and, in response, the person’s opponent attacks the position either by providing a counter-example or by demonstrating that the theory has absurd consequences (Gale 1991, 297). Roy Sorenson refers to these thought experiments as ‘refuters’, although not all refuters involve thought experiments (Sorenson 1992, 153).6 Thus, by way of example, in responding to the claim that ‘featherless biped’ provides an adequate definition of the human species, an opponent of the claim might, by way of counter-example, point to a shaven rooster (Gale 1991, 298). The use of such refuters is a commonplace in moral philosophy. For example, in order to refute a claim that it is always wrong to lie, one might direct our attention to the permissibility of some lies and do so by providing a case where most people would admit lying was acceptable (such as when telling a so-called ‘white lie’).
The second category of thought experiments involves what I shall call intuition pumps.7 This term has some currency in the literature, although, as far as I can tell, no standardised meaning.8 Indeed in some areas of philosophy it often functions simply as a synonym for ‘thought experiment’, whilst in other areas it is used to refer to pernicious thought experiments (although the latter is I would guess a less common usage).9 Given this divergence, I shall define it stipulatively, restricting the term to a particular class of thought experiments that aim to lead us, via our reactions to a single thought experiment, towards some general kind of conclusion.10 Trolley cases are often used in this way. We are meant to infer from the fact that we would choose to save the five rather than the one person on the track that numbers do count morally.
The third category of use to which thought experiments are put is as clarificatory devices. Perhaps the most widespread of these are what we might call commitment cleavers, that is cases where thought experiments are used to enhance our understanding by teasing apart distinct, but easily conflated, principles. This I take to be the import of Plato’s story of the Ring of Gyges in Book II of the Republic that occurs in the midst of a debate about the nature of justice (Plato 1974, 36). A commonplace of human discourse is that it is right to act justly. Yet it is not clear whether those who endorse it, do so on grounds of prudence (i.e. if we act unjustly we will typically suffer social sanctions) or because they regard acting fairly to be a fundamental moral obligation that we have regardless of the social consequences. In the dialogue, Glaucon tells the story of a ring that makes the bearer invisible and in this way forces his opponents to choose between justice chosen on prudential grounds and justice valued for its own sake. Through his use of this dramatic device, Plato’s characters are required to be more specific about their ethical commitments. Is it prudence or the intrinsic rightness of acting justly that underpins their commitment to justice?
Another common commitment cleaver is the Last Man example in environmental ethics. In this thought experiment—originally devised by Richard Routley—we are asked to imagine a hypothetical situation in which the last person in the world acted so as to eliminate all other living things and to destroy all landscapes after his demise (Routley 1973). Would it be wrong to do so? In other words, we are being asked to decide whether the environment matters extrinsically solely for the opportunities it furnishes humans to flourish or intrinsically for its own sake. This again is an experiment that aims to force us to reveal our fundamental moral commitments. As C.L. Ten notes such thought experiments help us to determine whether a particular principle or commitment is “fundamental or subordinate” (Ten 1987, 21).
My fourth and final category involves what we might call re-imaginings. There will be some ethical debates where over-familiarity with the issue can lead disputants no longer to engage in real dialogue with each other. The abortion debate is a clear example of such a stalemate. Abortion is a topic about which most people will have an opinion, and be prepared to argue about. However, often people’s views are so firmly fixed that there is no space for meaningful interaction. Here thought experiments provide a useful way of restarting dialogue.11 (I take Judith Jarvis Thomson’s violinist tale to be an excellent example of a thought experiment that functions as a re-imagining device.12)
Just as Brecht famously urged the theatrical spectator to suspend identification with the theatrical characters and their lives, in order to scrutinise the represented situation from a critical practical perspective, so good philosophy often gets us to do this with ourselves and our own lives. (Nussbaum 2000, 253)
Ethical theory can allow us to see relationships that ‘have eluded us in our daily thinking”(Nussbaum 2000, 253). For my part I think it is not just a matter of ‘defamiliarising’ but of reigniting genuine debate. Such re-imaginings can also functions as means of shifting the focus of a philosophical debate.
This taxonomy of legitimate uses of thought experiments in applied ethics is not intended to be exhaustive; however it does capture the central cases.13 Before leaving the topic of how they are employed, we should also note the rhetorical power (and hence the rhetorical function) of many of the more bizarre thought-experiments. Oftentimes the strange circumstances invoked in the story—such as, for instance, the idea of people-seeds floating around in the air—masks what role the thought experiment is intended to play in the argument in question. In many cases readers will have great difficulty determining what their intuitions about the story might be, let alone deciding whether or not the claim being made in the story is part of a valid line of reasoning. In such cases the bizarre nature of the tale hinders us from exercising critical faculties we might otherwise employ. As Daniel Dennett notes in Consciousness Explained, the highly imaginative scenarios of some thought experiments sometimes distract us from a thorough examination of what is being argued (Dennett 1984, 440). This is sometimes part of their rhetorical power (Brendel 2004, 107).
4 Should Proximity be our Guide? The Objection from Modality
One might naturally object to the use of thought experiments in applied ethical problems when the experiments involve modally bizarre situations, i.e., situations that are not possible in worlds nearby to ours. Let us call this the “Objection from Modality”. It involves a repudiation of those thought experiments that involve modally fanciful examples.
An obvious target for this objection would be Judith Jarvis Thomson’s people-seeds thought-experiment mentioned earlier in which the reader is asked to imagine a world in which ‘people-seeds’ float freely about and if left to their own devices will grow in our carpets. Thomson asks whether it would be permissible to remove any growing plants that we might find in our carpets, if we are responsible by leaving our windows open. Do we have any moral obligations to these future people if we allowed them in? And if not in this case, why, by way of analogy, do women in our world have such obligations when contraception fails? If we take these criticisms of bizarre cases seriously, then we should regard Thomson’s use of the people-seeds case as illegitimate, since it involves counterfactuals that invoke possible worlds very distant from our own. The idea is that the use of thought experiments in applied ethics can be legitimate, but only in so far as they do not involve modally remote worlds. Might proximity to our world be a way of distinguishing between licit and the illicit uses?
If we attend closely to the variety of thought experiments—and more specifically the variety of uses to which they are put—it soon becomes apparent that this way of drawing the distinction is mistaken. Firstly, on some occasions the bizarre examples provide useful re-imagining devices. The great advantage of a thought experiment like Thomson’s violinist is it allows for genuine dialogue to recommence (albeit about a context very different from that with which the dialogue began).
Second, such a line of criticism overlooks the role that general moral principles play in many discussions of important topics such as the justice of war, abortion and euthanasia and the role that thought experiments play in assessing those principles. (One need not be a ‘principlist’ to see the important role that thought experiments play in evaluating such principles). In attempting to determine the rightness or wrongness of a case, philosophers often appeal to some kind of ethical generalisations. These principles are presented as universal truths; indeed, part of their argumentative force comes from their status as moral universals. For instance, pacifists in discussions of the just war might appeal to a principle that says intentional killing is always wrong. In considering the merits of the argument, it is quite reasonable to ask, given its presentation as a universal, whether the premise that ‘killing is wrong’ is true in all possible worlds.14 Discussants might wonder whether there are any counter-examples to the case. These are legitimate avenues of inquiry. Accordingly, while applied ethics deals with ‘real world’ cases, if an argument regarding such a case relies centrally on a general moral principle, then it is warranted to test its generality against a range of possible scenarios, which may or may not be modally bizarre.
At the same time, I do not mean to imply that our use of bizarre thought experiments cannot go awry. Legitimacy here is determined with reference to the role that the thought experiment plays in an argument; this is the significance of adopting an ‘argumentative pragmatics’ approach to thought experiments. Sometimes the use of modally-bizarre examples is inappropriate, but not simply because the tale is bizarre, rather in such cases the bizarre will be illicit because the conclusions drawn are not warranted by the ‘evidence’ provided by the thought experiment or because they ignore the argumentative context. By the same lights, there will be far more quotidian cases that fail because of the structure of the argument in which they are embedded. (I will say a little more about this in the final section of the paper).
James Rachels’ ‘Smith and Jones’ thought-experiment is arguably a good example of this (Rachels 1975). This is a tale of two men, one Smith and the other Jones, both of whom have a cousin they would like to see dead. In the first case Smith drowns his cousin. In the second case Jones walks into the bathroom with the intention of drowning his cousin, but the very moment he intends to do so, the cousin bumps his head and drowns without Jones laying a finger on his cousin. Both intend to terminate the life of their respective cousins, but only one needs to do so. We are asked whether their actions are in any sense morally distinct.
Let us assume with Rachels that there is only the bare difference between actively bringing about X as opposed to not intervening to prevent X (and no genuine moral difference) in the case of the drowned cousin. All that this demonstrates is that there is not always a moral difference to be had between the active and the passive. However, a great deal more work is required if he is to demonstrate that there is no difference between active and passive euthanasia. The Smith and Jones case undermines any argument that says there is an intrinsic moral difference between active and passive euthanasia because killing is always morally distinct from letting die. But it does not, as Rachels seems to suggest, provide grounds for the stronger claim that there is no intrinsic difference between active and passive euthanasia. If the intention is to show that there is no intrinsic difference between the two, then the argument fails and the use of the thought experiment is, in this sense, illicit. Thus it is not only the bizarre that can lead us astray.
In short, then, in order to determine whether a thought experiment is legitimate we should begin by examining the role that it plays in the argument in which it is embedded, not by whether it belongs to a close possible world.15
I turn now to two objections that can be levelled not simply at the use of thought experiments in applied ethics, but rather at their use in ethical inquiry more generally.
5 The Objection from Under-description
Some critics—often claiming inspiration from Wittgenstein or from Wittgensteinian sources—have argued that such experiments are illegitimate because they are necessarily so under-described as to have no applicability to real life cases. We might call this the “Objection from Under-description”.16 The suggestion is that thought-experiments in ethics are so de-contextualised and under-described that they cannot bear on the real life cases applied ethicists wish to solve. Thus we might agree that in a certain hypothetical circumstance it would be wrong to ø, but the case is so lacking in the fine-grained detail of the cases with which we must deal, that there is no reason to believe it bears on our problems. If one introduces other details into the case, we might come to believe that in fact it is permissible to ø. According to the Objection from Under-description it is not possible to switch from the abstract hypothetical example to our real-life case because there is so much detail missing from the former. Only if we have all the detail can we know what to say about a case. But if we have all of the detail, then it will differ in terms of detail from other cases we wish the example to illuminate.
This line of reasoning is often associated with the ethical Wittgensteinianism of the Swansea School where the objection to thought experiments is part of a more general animus against theory in ethics. Instead of theory what is required is close attention to the details of real-life cases. In her article ‘The Power of Example’ Onora O’Neill notes that in this Wittgensteinian tradition examples are conceived “neither as incidental, let alone sketchy, illustrations of moral theory, nor as models for action, nor just as morally educative”. They are instead the “pivot of moral thought”. As O'Neill notes, “[I]nstead of schematic, possibly unimportant illustrations of principle and theory, Wittgensteinian writers provide elaborate and extended discussions of serious moral vicissitudes (O'Neill 1989, 170).” O’Neill goes on to criticise this approach for leading to a kind of ‘moral connoisseurship’ that treats moral life as a spectator sport instead of as a guide to action (O'Neill 1989, 176).
Recently, Cora Diamond has queried whether we should follow O’Neill in thinking that a concern with “real or realish” cases should really be viewed as at the heart of a Wittgensteinian-inspired ethics (Diamond 2002). (One can understand her motivation here since Wittgenstein, with his green Martians and rubbery rulers, was clearly not a man averse to bizarre examples). Diamond is particularly interested in defending the kinds of abstraction that one finds in stories such as The Ring of Gyges. Although she is Wittgensteinian in the sense that she does not think that what we might learn by thinking about a hypothetical case can be determined in advance by general principles, she is nonetheless of the opinion that hypothetical abstract examples are useful (Diamond 2002, 248).
But let us put exegetical questions to one side, for the Objection from Under-description stands alone as an objection to thought-experiments, regardless of its status as a piece of genuine Wittgensteinianism and regardless of what alternative ethical methodologies one might wish to endorse. The Objection in itself does not prescribe alternative ethical methodologies.
There are good reasons for defending abstraction in ethics, for one person’s under-description is another person’s legitimate and sophisticated abstraction. Here two forms of abstraction in particular are worthy of endorsement. First, there are cases—which in the previous section I labelled ‘re-imaginings’—where one necessarily abstracts so as to remove a debate from an intellectual stalemate. By abstracting, as opposed to under-describing, one shifts the discussion to a position where it is possible to have an actual debate. Second, in rejecting abstraction as under-description, proponents of this objection would rule inadmissible necessary processes of theoretical clarification. Think once again of our commitment cleavers, such as provided by Glaucon and his Ring of Gyges. The story of the Lydian shepherd, Gyges, with a magical ring, is a story that is designed to test our normative commitments and as such it works well. If one were to reject these kinds of processes of abstraction then our capacity to deal with complicated normative issues would be much diminished. For these reasons we should reject the general claim that thought experiments being under-described are thereby illicit.
Of course, this does not mean that the details of a thought experiment cannot be questioned because they are not reported, but stipulated by the thought-experimenter. In this I differ from Roy Sorenson who defends the idea that stipulation is enough on the grounds that it is after-all the thought-experimenter's experiment (Sorenson 1992, 259). But whilst that view has some initial appeal, it should be rejected since it overlooks the duty on the part of the thought-experimenter to provide, on request, a justification for the introduction of an imaginary case into a debate. For instance, if questioned, Glaucon would have had an intellectual duty to demonstrate the relevance of the story of the Lydian shepherd to the preceding debate on justice. The experiment must be relevant to whatever question is under review. Stipulation is not enough.
6 Could there be Morally Objectionable thought Experiments?
….what you ought to do if you had to move forward, and stepping with your right foot meant killing 25 fine young men while stepping with your left foot would kill 50 drooling old ones. (Anscombe 1957, 267)
Her point is that such thought experiments treat morally serious issues in a frivolous manner. Such thought experiments are morally objectionable because they involve frivolous examples and this is all the more galling when we are dealing with important ethical and social issues. Anscombe expresses a similar ‘dumbfounded’ reaction to modern moral philosophy in general—and not just thought experiments—when she suggests that it is futile to engage in philosophical dialogue with philosophers who think in this way (Anscombe 1958, 17). By their very willingness to entertain the idea that it is an open question whether or not procuring the judicial execution of the innocent is permissible, Anscombe believes such philosophers demonstrate the corrupt nature of their minds.
Obviously the right thing to do would be to jump and polish off the lot. (Anscombe 1957, 267)
One can only assume that Anscombe would be equally horrified by a great deal of contemporary applied ethics for the literature abounds with examples that might be thought to be morally frivolous and objectionable for being so. An example of this might be Michael Tooley’s super-kittens case in which cats are given a drug such that they have the rational capacities of persons (Tooley 1983, 191). Tooley’s claim is that if we are concerned about abortion because a foetus is a potential person, we should equally be concerned about killing cats because all are potential super-cats and thus potential persons. One might argue thought experiments like these encourage us to think about non-trivial issues, such as abortion or euthanasia, in a highly insouciant or casual manner. One might even argue that merely contemplating the scenarios envisaged by such thought experiments is morally corrupting.
Another way of explicating this idea that many thought experiments are morally objectionable is that they prevent authentic imaginative moral engagement with the issues they are meant to illuminate. The suggestion is that the highly artificial nature of many thought experiments leaves little space for genuine choice; they stifle our moral creativity and stymie the development of appropriate moral sensibilities. This line of objection would most plausibly be directed at what I label intuition pumps, for in these thought experiments a single case is constructed with the aim of directing us towards the experimenter’s favoured conclusion, often by forcing us to choose between the experimenter’s favoured choice and some entirely unpalatable alternative. In such cases by failing to offer a genuine choice, the thought experiment fails to foster genuine moral engagement. They are so structured as to lead us towards a specific conclusion about a case.17
Neither of these claims about the morally objectionable character of thought experiments, however, provides grounds for a complete repudiation of thought experiments, bizarre or otherwise. If we consider the accusation of moral frivolity more closely, we see it accords a great deal of moral significance to what it is we imagine. In doing so, it overstates the importance of what we imagine. It is our genuine and considered moral attitudes that are of greatest importance, and what we imagine is conceptually distinct from those attitudes. Such critics need to demonstrate a causal relationship between the tone of what it is we imagine and our moral modes of regard. They need to show that these thought experiments infect the way we regard other people. But, this line of reasoning underestimates the resilience of many of our genuine moral attitudes—merely talking in a frivolous manner about an issue like abortion is unlikely, by itself, to lead us to treat it with a lack of due seriousness when considering its permissibility or otherwise. Furthermore, in some cases separating ourselves from our ingrained moral views—if need be through a slightly jocular tone—is necessary if we are genuinely to attend to the case. A lack of moral seriousness is rarely the problem when topics such as euthanasia or stem cell research are at issue. Indeed, more often than not the real obstacle is our dogged unwillingness to engage in genuine dialogue with those who hold differing views. If a thought experiment, through its jocular tone, enables us to stand back from the case and re-examine our views, then this can only assist in genuine moral engagement.
Turning to the claim that thought experiments are morally objectionable because they stifle moral creativity, I suggest that such an objection only has any grip if one views thought experiments as having greater power than they do. The assumption is that by ‘consorting’ with a highly loaded thought experiment, one immediately concedes to the perspective of the thought-experimenter. But this is simply not the case: engaging with a thought experiment—no matter how loaded—does not mean that one has to accept the conclusions drawn by the thought experimenter. Instead there are many ways of not conceding. Consider Thomson’s violinist example. One might concede that in this case it would be justifiable to free oneself from the violinist, but this does not carry across to the case of abortion since there might be a number of important disanalogies between it and that of the woman who falls pregnant through rape. (This is precisely the approach taken for example, by Michael Davis in his 1983 article). Or one might develop an alternative case where our intuitions seem to run in the other direction. So long as we do not regard the particular thought experiment as the final word, but as an opportunity for further dialogue, then imagining what it is the thought experimenter asks us to imagine need not stifle our moral creativity.
7 The Fallacy of Changing the Subject—A Potential Trap in Applied Ethics
As stated, the three objections considered do not provide grounds for abandoning the use of thought experiments in applied ethics. Indeed, to the contrary, the discussion shows a number of ways in which thought experiments provide us with extremely valuable intellectual resources. If some other line of criticism were to introduce genuine concerns, then it would need to demonstrate that the problems it raises significantly outweigh the intellectual benefits that thought experiments bring. However, given the wide variety of uses to which they are put, it is hard to imagine a criticism that could successfully demonstrate that all the different varieties are illicit. Any plausible criticisms would be of particular uses in specific forms of arguments, not of their use in general. According to my account, thought experiments are illicit when they fail to provide the inferential basis for the conclusions that philosophers intend them to support.
However, at this point I want to complicate the picture somewhat and identify a distinctive way in which the use of thought experiments in applied ethics can lead us astray. The use of thought experiments leads to a ‘natural error’ when we ignore the contingency of the problems with which applied ethicists characteristically deal and draw conclusions that attempt to accommodate a wide range of merely possible cases rather than the actual case before us. This ‘natural error’ I shall refer to as the Fallacy of Changing the Subject. It picks out a natural way for us to misuse thought experiments in cases in which the focus shifts from what should be done in an actual circumstance, to what might or might not be necessarily morally permissible in all possible circumstances.
At this stage an illustrative example might be helpful. Imagine a debate concerning the problem of whether or not men should have an equal say in choosing whether a pregnancy should be terminated. A proponent of equal rights for men might suggest that it is possible to imagine circumstances in which artificial wombs become so developed that gestation can proceed entirely ex utero. (This, of course, is not entirely fanciful). In this situation envisaged by the thought experiment one might well argue that men should have equal rights in decisions about whether gestation should continue since the implications for the health and inconvenience of men and women are no different (at least as the story is told). However, in the world we currently inhabit, the implications of pregnancy are vastly different for the women involved than for the men. To use such a thought experiment to justify equality in this case at the present time is, I would suggest, illegitimate because it involves changing the subject. (That is not the prejudge the question of whether or not men should have equal right but simply to say that in this case the use of the thought experiment is illegitimate). Although, as I noted above, we should not reject bizarre thought experiments when they, for instance, illuminate the general principles we employ or provide commitment cleavers, nonetheless when we are engaged in applied ethics then the arguments in which these thought experiments are ‘nested’ must respect the ‘contingency’ of that context and not change the subject.
What we have here is a form of contingency constraint on the use of thought experiments in applied ethics. One (mistaken) reading of this constraint would be that it involves a restriction of the range of thought experiments that are admissible within applied ethics. However, this is not what I am defending; indeed such a reading would sit rather oddly with the earlier defence of the modally bizarre. The narrower interpretation of the contingency constraint that I have in mind is that it restricts the kinds of counter-examples (as opposed to thought experiments more generally) one might raise against any applied conclusion concerning the right course of action. This contingency constraint has no bearing on many other kinds of uses of thought experiments (such as commitment cleavers) nor does it restrict the use of counter-examples when considering the general moral principles that often form the basis of our conclusions in applied ethics.
To see the difference, reconsider our earlier illustrative example. On the wider interpretation the rejection of artificial wombs as providing a relevant case involves a modal restriction on currently impossible scenarios. On my preferred narrower interpretation, the rejection of the counter-example is because it involves answering a different question. The question shifts from “Should men have the same rights to make decisions about pregnancies as women at the present time?” to “Should men have the same rights in some possible future circumstances?”. When we ask whether men should have the same rights as women, this is a form of what David Lewis calls ‘restricted speaking’ since in this case we are not asking whether they have those rights in all possible circumstances.18 We need to be mindful of the particular context of the debate into which any thought experiment is being introduced. This is ultimately a counsel of informal reasoning.
We need also to distinguish this contingency constraint from a particularist response to the use of thought experiments in ethics. For a particularist, like Jonathan Dancy (1993), moral judgements are particular to the cases in question and we cannot generalise from other cases. He is especially suspicious of so-called ‘switching arguments’ that he believes force us to a view of the case before us “by appeal to some feature of another case or some comparison between this one and that” (Dancy 1993, 65). Dancy claims such switching arguments are far more unreliable than moral generalists suppose since reasons often change their ‘moral polarity’ in different contexts.19
Space obviously does not permit an extended discussion of the relative merits of Dancy’s views. Suffice it to say that his view is remarkably different from the one on offer here and his range of relevant cases will be far more restricted. On my account we can in fact generalise across cases. My restriction is simply that the domain of relevant cases that will count as counter-examples to a conclusion in applied ethics will not include all logically possible ones. In choosing a right course of action we can in fact learn from other relevantly similar cases, it is simply that our conclusion do not have to hold necessarily.
My general point, then, is that in applied ethics we are dealing with problems of choice in very restricted domains of action (although the domain of relevant cases is much broader than the particularist would allow). Typically we are concerned with whether an action in a specific context is right or wrong for these kinds of beings.20 To raise the possibility of artificial wombs as a means of answering the question of who should determine whether a pregnancy should continue is to ‘change the subject’. Similarly, if the topic under discussion is whether it is morally permissible to gaol people without trial, then it would be ‘changing the subject’ to point to creatures who did not mind, or even enjoyed, being incarcerated, since it is assumed in the debate before us the prisoners do not want to be in jail.
My account of this contingency constraint, then, does not rule out bizarre thought experiments, but simply advises us to be cautious of the use of those thought experiments that lead us to change the topic. It does, however, involve a repudiation of the philosophical nostrum that any genuine ethical claim, in order to be genuine, must hold true in all logically possible worlds.21
8 Concluding Remarks
One of my children recently asked me the following riddle. Imagine that you are sealed in a large metallic room with no windows, no doors and no other means of exiting. How would you get out? The answer he gave, after rejecting a number of my suggestions, was “Stop Imagining!”.
Analogously, we can ask whether applied ethicists should also stop imagining? Should we stop using thought experiments? Are thought experiments but another form of befuddlement encountered in philosophy? While it is entirely understandable why some philosophers might want to refuse to engage with thought experiments—especially the bizarre imaginary cases—such a response is ultimately mistaken, for it would deny us access to an important intellectual resource. Thought experiments have an important role to play in applied ethics, in focusing our attention on morally-salient features of a moral problem, providing counter-examples to general moral principles and allowing us to re-imagine moral problems that have become stale.
Although there is nothing intrinsically wrong with any thought-experiment in terms of their content, this is not to suggest their use is always legitimate. Firstly, their employment is illicit when they are elements of invalid arguments. This of course is a generic concern with the use of thought experiments. But there is also a way of going wrong that is specific to applied ethics. When thought-experiments fail to respect the contingent nature of the problems considered in applied ethics then their use is also illicit. This does not rule out bizarre thought experiments. However, it does involve a rejection of the use of thought experiments that lead us to change the topic under examination.
The appropriate response, if one finds a particular thought experiment pernicious, is to seek out the point where the argument goes awry or explain how the framing of the question might be misleading. Thus a proper reaction to, say for example, Thomson’s violinist or her people-seeds case is to think of relevant disanalogies between her stories and the circumstances of abortion. One should not refuse to engage with such thought experiments, but instead deal with them simply as elements in a chain of reasoning, just as we would in any other dialectical situation. It would be a mistake to stop imagining.22
Brendel argues that ‘thought experiments in which we are asked to apply a concept in an unfamiliar situation can be problematic, since we often have no stable intuitions that could guide us in finding a justified answer’ (Brendel 2004, 106).
The analogy here is with pragmatics in linguistics where the focus of analysis is on the identity and intentions of the speaker and hearer and the context of discussion.
Sorenson (1992) takes a similar path in the last chapter of his book on thought experiments where he provides a list of fallacies associated with the use of thought experiments.
Readers should note the difference with Sorensen’s account in which the official single role of thought experiments is to test modal consequences. Sorensen’s leavens this narrow conception of function with the comment that the ‘apparent narrowness of its function eases once we realise that there are many kinds of necessity’. (Sorenson 1992, 6). On my account there is no such single function, no matter how broadly construed.
We also find this ‘destructive’ aspect of thought experiments discussed by Thomas Kuhn (1997) in his article ‘A Function for Thought Experiments’.
James Robert Brown (1991, 124) distinguishes, in a similar vein to my distinction between counter-examples and intuition-pumps between destructive and constructive thought experiments.
Brendel (2004, 106) uses the term to denote thought experiments that misuse intuitions and lead us to believe in an unjustified conclusion.
We might reasonably think of it as a form of inductive reasoning.
This has some resonance with what Sorensen (1992, 5) has to say about the cleansing functions of thought experiments.
Nancy Davis (2001) has an insightful discussion of the various literary techniques used by Thomson that engage our attention..
For a different (although not incompatible) way of cutting up the functions and purposes of thought experiments, Brendel, 2004, 92.
Elke Brendel (2004, 106) makes a similar suggestion noting that “….if an analysis or a definition of a concept is regarded as universally valid, then far-fetched imaginary scenarios that cast doubt on this general validity are legitimate”.
Interestingly, Quine (1972, 490) notes that ‘The method of science fiction has its uses in philosophy but….I wonder whether the limits of the method are properly heeded.”
The objection from under-description can only be directed plausibly at intuition pumps where we are trying to derive some more general conclusions.
This is presumably John Stuart Mill’s point when he writes: “A test of right and wrong must be the means, one would think, of ascertaining what is right and wrong, and not a consequence of having already ascertained it.” (Mill 1972, 2).
Lewis (1986, 3) distinguishes between restricted and unrestricted speaking using the example of the claim that all the beer is in the fridge. The restricted view ignores all of the beer that there is in the world that is outside of the fridge, that is all of the beer that exists simpliciter while the unrestricted view, in answering the question, does not.
Dancy (1993, 64) notes that he does not reject all switching arguments, rather it is that the particularist allows far fewer to be sound than does the generalist.
Martha Nussbaum (1993, 248) notes that Aristotle defends a similar contingency constraint on philosophical theory more generally. She writes that in the Politics, Aristotle “insists that only human, and not either animals or gods, will have our basic ethical terms and concepts…..because the beasts are unable to form the concepts and the gods lack the experiences of limit and finitude that give a concept such as justice its point [Politics 1, 2, 1253a1-18]”.
I would like to thank Peter Forrest for his very helpful comments on this paper as well as two anonymous referees from this journal.