In the last posting we had reached Parfit's understanding of how to comprehend the two parts of the Formula of Humanity and the introduction of examples that occurs next is part of his way of testing the understanding of the formula at this point. Essentially Parfit produces now some classic "trolley problem" arguments of the type that Judith Jarvis Thompson initially introduced into moral philosophy. As originally formulated by Thompson these problems literally refer to potential outcomes when faced with runaway trains. Parfit introduces three examples that fairly closely parallel Thompson's originals. Firstly he refers to the "Lifeboat" example in which one person (termed "White") is stuck on a rock somewhere whilst five others are elsewhere and our lifeboat can either rescue White or the five others but not both. Secondly, Parfit invokes "tunnel" in which a runaway train is heading towards five people but could be redirected away from them, at, however, the cost of hitting and killing poor old White. Finally, in "Bridge" the train is again aiming at the five but there isn't a parallel track but we could, using a trap-door, get White off a Bridge above the track into the path of the on-coming train thus saving the five. (The second two are pretty transparent replicas of two of Thompson's originals.)
In terms of "outcomes" alone all three examples have the same structure of enabling us to act in such a way that we save the five at the expense of the one. However, the manner in which this outcome is produced is different in each case as, in "Lifeboat", it's simply the case that, given the constraints of time, there's no way to save the five and the one and so a trade-off between them seems plausible in which we justify our action through a classic commissions/omissions defence. In "Tunnel" the saving of the five happens more directly through a fore-seen harm being produced to the one which we have directly brought about (so that the commissions/omissions defence we could give in "Lifeboat" isn't here available). Finally, in "Bridge" we, even more directly, aim at harming White in order to save the five and often it is with this example that intuition rings an alarm bell leading many to reject this action despite plausibly having sacrificed White in the two previous cases.
If "Bridge" is a case where many object, some draw a line earlier and rule out the action even in "Tunnel" on the grounds that we have an over-riding duty not to directly harm so that White can't be made the victim there either. If you don't draw the line at "Tunnel", however, there are less obvious reasons than might be thought why you should do so in the case of "Bridge". Now, Parfit's point in bringing in these thought-experiments is not to repeat the arguments that have swirled ever since Thompson set out the originals of these cases but rather to see what difference application of the two sub-principles he takes the Formula of Humanity to comprise of, make to the cases under consideration.
Recall that Parfit divided the Formula of Humanity into two sub-parts that he described as incarnating a Rational Consent notion and a Mere Means principle. Rational Consent taken alone implies treating people in ways to which they could, in principle, give rational consent even if, in the actual cases under consideration, they don't manifest the tendency to give such consent (or perhaps wouldn't if we were in a position to ask them). With regard to this principle Parfit assumes that White could give Rational Consent to sacrificing themselves with the proviso that this sacrifice would (or would at least tend) to save the other five. So the Rational Consent principle alone, on Parfit's construal, would not rule out taking action that produced this outcome.
It might next be asked whether the addition of the Mere Means principle would make any difference to Parfit's verdict here but he takes it to be the case that it wouldn't affect the verdict since, if we treat people in accord with the Rational Consent principle, then we wouldn't be violating the Mere Means principle. Taking the examples in turn, "Lifeboat" doesn't seem problematic since here we can see what would be meant by "Rational Consent" in terms of how White might, assuming a Rawslian veil of ignorance, adopt the position of thinking that the five should here be saved were they to be asked about the example without knowing whether they were a member of the five or the one left, unfortunately, to the fate of the waves.
Given this argument Parfit also takes it to be true that there would be no relevant difference between the case of "Lifeboat" and "Tunnel". Viewed in relation to "Rational Consent" alone Parfit also thinks that there is no relevant difference again between "Lifeboat" and "Bridge" though this is, as indicated above, certainly a counter-intuitive result. However, at this point we may still be unconvinced that the application of the Mere Means principle could validate the same outcome in the case of "Bridge" at least as Parfit's suggestion about Rational Consent has done.
In considering this objection Parfit mentions one understanding of the Mere Means principle that runs counter to the one he has adopted and that is the view of Robert Nozick who, in Anarchy, State & Utopia, presents the Mere Means principle as a basis for appeal to deontological side-constraints that overcome the consequentialist reasoning that Parfit has adopted up to this point. On Nozick's view the side-constraint in question enables one to deny that it is right to sacrifice someone in order to achieve an end they have not consented to the adoption of. Noticeably, Nozick's reading requires not just that the Mere Means principle is understood differently from how Parfit has presented but also alters the understanding of what is involved in reference to consent as here actual consent makes an appearance rather than Parfit's ideas of "rational" consent.
Parfit states that the Nozickian view requires us to view the Mere Means principle in such a way that it it becomes sufficient for someone to be treated as a mere means if we act towards them in a way that they have not actually consented to (and subsequently harm them in the process). If this is the implication of Nozick's understanding of Mere Means then Parfit has a reply ready to hand as it may be necessary (as he gives an additional example to show) that someone be injured (in ways to which they have not actually consented) in a relatively minor way in order that someone else be saved from a much worse fate and yet we wouldn't generally regard that as sufficient to say that the injured person had been treated merely as a means. This riposte to Nozick's case is, however, surely insufficient since, in the examples under consideration the level of harm involved is life-threatening and not only does that undermine actual consent being available but it destroys all conditions of agency as such.
Parfit considers this riposte but not, I think, in its full sense since he views it only in terms of limitation of harm, not in terms of protecting the conditions of agency as such. Later Parfit does arrive at consideration of such a case which is considered in a very classic consequentialist way. This is the case he dubs "Catastrophe" where we can prevent some awful event occurring only at the cost of killing some innocent person. This does directly involve consideration of undercutting of the conditions of agency and it is treated in traditional way through maximisation of the good.
If Parfit's pattern of reasoning is here somewhat predictable he does use it to modify Nozick's objection through formulation of a "harm principle" that states that it is plausible to harm people without their consent in order to achieve a good aim so long as the harm in question is not "disproportionate" with regard to the aim. Parfit does give a final consideration to the objection from agency I have mentioned stating that Thompson assumes that there is some absolute value attaching to it such that it cannot be over-ridden whatever the supposed good outcome in question. Assuming that someone maintains this view Parfit proposes a modified version of his 'harm principle' that allows lesser harms to be inflicted on people without their consent assuming the good end is the basis of this.
Part of the point of this whole "trolley" discussion on Parfit's part has been to bolster his argument that we cannot apply the Mere Means principle directly to our evaluation of actions. Rather than applying it to the evaluation of actions it should apply, on Parfit's conception, only to the evaluation of attitudes. This produces the outcome that the second half of the Formula of Humanity is now understood in one way when it applies to evaluation of attitudes (where the Mere Means principle applies) and another at the level of evaluation of actions (where it is replaced by the Harm principle). However there is another element to the Formula of Humanity which has yet to be considered and that is its reference to "respect" for rational nature and the way in which this is treated in the first of Parfit's 2002 lectures will be the subject of my next posting on Parfit.