Tomkow.com

The Good, The Bad and Peter Singer

The Wall Street Journal reporting Peter Singer's forthcoming book tells us:

In his latest book, "The Life You Can Save,"  Mr. Singer argues that failing to donate money to help the roughly 1 billion people suffering from poverty and preventable diseases is a moral offense equivalent to standing by as a child drowns because you don't want to ruin a nice pair of shoes.

Equivalent. But how bad is that I wonder?  Given that Singer is on record saying there is "no intrinsic moral difference between killing and allowing to die" he would seem committed to saying that failing to donate is morally equivalent to drowning a child.  Pretty bad!

Of course, not everyone denies moral significance to the difference between killing and allowing to die. Some philosophers distinguish between "negative" duties (e.g., not to drown children) and positive ones, (e.g., to save children from drowning), holding that positive duties are not as onerous as negative ones.  Though, as Judith Jarvis Thomson recently observed,

"... it is one thing to say there is a difference in weight between positive and negative duties, and quite another to say what the source of that difference is.  I know of no thoroughly convincing account of its source, and regard the need for one as among the most pressing in all moral theory."

For anyone interested in taking up Thomson's challenge, Singer's equivalence poses a problem.  After all, this bystander who stands by while a child drowns (for the sake of his shoes!) is a bad man. Let us not mince words; he is a sonofabitch.  The positive moral duty not to behave like such a character must have significant "weight".  And if failing in positive duties makes us as bad as that guy, then the difference in "weight" and hence the significance of the positive-negative distinction itself, must be morally slight.

I give negligible amounts to charity. So is Singer calling me a sonofabitch? Apparently. And you too, if you fail to donate money to the starving billion when you could (and you know you could).

Is there any way to defend ourselves? Shelly Kagan, Peter Unger, and now Singer, have written whole books arguing that there is not.  Arguing, that is, that if you agree that you have moral obligations to do things like help that drowning child, you must concede that you are likewise obliged to do everything you can to help everyone, everywhere, all the time.  If that is so, then it seems we must admit that we are sons of bitches, you and I.

The problem for Singer's position is that none of us believe its conclusion.  Not even Singer. 

Singer tells the WSJ:

I give a third of my income to Oxfam and other organizations working in the field. I still feel that, as comfortably off as I am, I should be giving more. We still take family vacations to nice places. We could spend time somewhere less expensive. Also, I'm still prepared to have a bottle of wine or go to the theater or to some kind of concert. If you think about what that money can do for people in extreme poverty, it's hard to justify that type of spending.

Hard? Hard!  If having that Merlot is really equivalent to standing by while a child drowns (to save your shoes!)  isn't it impossible to justify that type of spending?

And yet Singer clearly does not expect us to think he is a sonofabitch.  Look at that picture!

We don't begrudge Singer a drink or show now and then.  We don't think of him as comparable to the sonofabitch who stands by while a child drowns, still less to someone who drowns a child.  Why not?

It seems to me that it must be because the children Singer lets die when he buys his ticket for the show  aren't nearby.  It is not as if Singer thinks those stage kids in Les Miz are really miserable   We assume that Singer, like the rest of us,  would be hard pressed to enjoy a drink or a show if he knew there was great misery nearby.  But it is easy for all of us to enjoy ourselves even while we know for a fact that there is enormous suffering elsewhere and out of sight.  It's easy provided we don't dwell on it.  And it is easy not to dwell on it: out of sight, out of mind. 

We think the sonofabitch is a bad guy--- whereas Singer, not so much--  because the sonofabitch is  letting a nearby kid die, not some out-of-sight abstraction.  It’s ignoring this vividly present suffering that makes him such an odious figure.   Not so Singer, or the rest of us,  who are far from the children we  allow to die . Which is  why we think that we are not so bad. 

Does this make sense?  After all as Singer et al never tire of pointing out, mere proximity can't make a moral difference.  What matters for the moral assessment of behavior are means and motive.  Singer has the means: he  could give his pin money to charity as easily as the bad guy could save that drowning child.  And while the sight of a drowning child may be psychologically harrowing Singer knows full well the consequences of his actions: he cannot plead ignorance.  He chooses self-gratification over saving lives.  How can we regard him as morally different from the sonofabitch bystander? 

We can't.

I think that Singer is right that failing to send money overseas is morally equivalent to allowing the  child to drown.   I think that because I think we have no moral obligation to either save the drowning child or the remote starving millions.  Absent prior duties imposed by promises or  parenthood, I maintain that we have no moral obligation to alleviate the sufferings or enhance the welfare of others.  There are no positive moral duties. 

I offer our good opinion of Singer and Singer's good opinion of himself (look at that picture),  despite his drinks and shows and vacations, as evidence for my view. 

Let me hasten to say that I agree that the bystander who watches the child die is a sonofabitch.  I am happy to say  he is  "a bad man".  But the sense of "bad" in which this is true is not, I think, a moral one.  

Moral right and wrong have to do with actions, with what people do.  But we do not think the bystander is a  bad guy because he does something morally wrong.  Suppose the bystander had stayed at home to polish his shoes.  Then he would never have encountered the drowning child and would never had the opportunity to save or refrain from saving it.  If you count failing-to-save as "doing a bad thing" then you should agree that, had he stayed home, the bystander would have done one less bad thing that day.  If you think the bystander wrongs this child by not saving it when he can, then you should agree that the bystander would not have wronged the child had he just not been standing by.  

But the world would not have been a better place -- no one would have been better or better off-- had the bystander stayed home.  The child would still be dead and the bystander would still be every bit as much of a sonofabitch as he is in the world where he can save the child but doesn't.  He would still be the kind of sonofabitch who would stand by and watch a child drown when he could save it.  His actual behavior with respect to the child is relevant only because it reveals what a sonofabitch he is. 

The  bystander is a sonofabitch because his behavior demonstrates that he has a bad character .  I suppose a practitioner of "virtue ethics" would say that his behavior demonstrates that he lacks the virtue of "charity" or perhaps "empathy". I don't entirely disagree:  I think charitableness, in its place,  is a virtue.  But I don't agree that it is a moral or "ethical" virtue. 

There are lots of virtues that aren't moral.  I think having a sense of humor is a virtue but I wouldn't claim it was a moral virtue (except as a joke).  To be convinced that a character trait is morally bad I would require evidence that it disposed its bearer to morally bad behavior.  And remember I don't think  that the bystander does anything morally bad by doing things like standing by while the child drowns.  If you do then you will have a problem with Singer's equivalence: if failing to donate all you can is as unvirtuous as letting that child drown then you are going to have to say that even Singer lacks moral virtue.  But look at that picture!

Virtue ethics goes wrong precisely when it aims to be a kind of ethics: as if we could decide the ethics of actions by examining the character of  admirable agents.  This is not a new idea:  Nietzsche took the Christian recipe for being a good person to be deeply repulsive.  Heroic charitableness, he thought, does not make you a Hero.  But because he insisted on treating his position as a moral one he took this to be a reductio of Christian norms of action (this, I think, is the core of his "critique of morality").   The problem with this is its corollary, which Nietzsche embraced, viz.  that True Heroes can literally do no wrong.  This runs Singer's mistake in reverse: whereas Singer supposes that behaving like a sonofabitch must mean that you are doing something morally bad, Nietzsche lets his superior men get away with murder.  Contemporary virtue ethicists threaten to  repeat the mistake--  though their preferred model of the Übermensch is less like Siegfried and  more like an Associate Professor of Women’s Studies .

The corrective to Nietzsche is not to deny the reality of heroism but to  acknowledge that heroic acting out is not always morally good;  just as the corrective to Singer is to note that some behavior, though swinish, is not morally wrong. 

So if I don't think the badness of the sonofabitch is moral badness, what sort of disvalue is it?

A.J. Ayer is reported to have once said, of a certain colleague, that he had  "...gone bad."  Ayer explained, " I don't mean morally bad.  I don't use moral language.  I mean he's gone bad like an orange goes bad!"  I think that's about right. The relevant sort of goodness and badness has more in common with aesthetic value than it does with moral right and wrong.  We think the bystander is an ugly customer.  He is, among people, as an ugly picture is among pictures.

In calling our evaluation of the bystander’s badness "aesthetic" I am not in the least trying to trivialize it.  I do not say (would never say)  "merely" aesthetic.  The measures by which we judge one person better than another are at the center of human life.  They are values by which we choose who to love, who to hate, who to befriend, who to celebrate and who to shun.  Few of us  make decisions as important as these by asking how much the other gives to Oxfam. 

To ask if someone is a good or a bad person is to ask a profoundly different question than to ask whether they do or are disposed to do morally good or bad things.  I call the former 'aesthetic' because the logic of 'good person'  seems to me to more like that of 'good picture' than 'do gooder'.

Moralists are in the business of dividing in twain: deontologists  between good and bad acts; consequentialists among outcomes.  But just as it is an aesthetic mistake to think that the job of the critic is to divide all art into two piles it is absurd to think that  people monotonically range from saints to sons of bitches.  There are good people and bad, just as there are good and bad pictures, but there is more to it than that.   Lot's more.  Though one will get little help from philosophical moralists in trying to sort it out.

Moral philosophers are characteristically committed to avoiding comparisons of the intrinsic value of persons.  The moral point of view, many hold, is constituted by seeing no person, even oneself, as more valuable than any other.   It is the "view from nowhere" and no one.  And that may indeed be the right stance for resolving questions of moral right and wrong.  But this point of view is blind to everything essential to answering a different sort of question: "What kind of person do I want to be?"  It is all very well to treat everyone as an end in himself, but we also have to decide how we ourselves want to end up.

Answering that question requires a keen eye for the varieties of persons there are and can be.  Again, one will get scant insight on this score from a moralist who can't see a difference between people who don't give to Oxfam and those who would let a child drown.  Traditionally  the job of describing human values-- understood as the project of saying what gives humans their different values-- belongs to the narrative arts and their critics.  And when we look to what they tell us about those values we find that they are largely orthogonal to moral virtue.

Good people can do bad things even when they act in character (the technical term for this phenomenon is 'tragedy').   Killing people is bad.  Hamlet kills lots of people and mostly for no good reason.  For sure Hamlet does a bad thing when he kills Polonius .  But Hamlet is not a bad guy; or if he is, its not just because he kills people and if he isn't, it isn't because he somehow makes up for it by increasing Danish foreign aid.

It seems to me that anyone who would stand by and let a child drown is simply a sonofabitch.  But that may only show my lack of imagination.  I suspect that Dostoyevsky could have closely described a character doing just that and made  him wholly sympathetic. Cormac McCarthy's work is full of sons of bitches  who would go out of their way to watch a child drown; but they aren't simple sons of bitches.

And If this seems frivoulous, we have Milgram and Arendt's empirical observations of the disconnect between the quality of character and action.  The most evil things can be done by people who no one would judge, apart from their actions, to be bad  people.  Thus, if you are planning a genocide and looking for staff you can probably find the kind of people you'll need to tend to the details by looking to the office of your local Dean of Arts.  Which is not to say that the people you will find there are morally bad, only that they are banal characters.  And lucky.  Theirs is the moral luck not to have been born in a time and place where -- characters unchanged-- they would placidly have administered atrocity.

It is one thing to avoid doing bad things, another much more difficult and complicated thing to be a good person.  I'm suggesting that we avoid confusing these by ceasing to call questions about the value of persons "moral questions".  Of course, the weight of tradition is against me in this.  As a compromise perhaps we could train ourselves to reserve 'moral' for one sort of goodness  and 'ethics' for the other.  I don't  much care which, but I do think it is important to keep them separate.  Confusion on this score can lead one to think that, say,  giving to Oxfam makes you a good person and, well... look at that picture.

Against Intentionality

Frege's mistake lay in assuming an appealing theory about why people say what they do. He thought people believe sentences true because they believe what those sentences say.  If so, then what someone thinks must exactly reflect what he would candidly assert.  The theory appeals because it assumes a even simpler picture of thinking: it is the picture of thinking as talking to yourself, sotto voce.

Frege didn't invent this way of thinking about thinking, but he was the first  to see how much must follow about thought and language if it is true.  If we take it as analytic that beliefs have the functional role Frege assumes for them, we can define Frege Beliefs:

x believes_F p =_df  x would believe ^┌p^┐ was true iff x understood ^┌p^┐

When contemporary philosophers talk about the "narrow", "notional", "opaque" or "internalist" sense of 'believes' they are talking about belief_F.

The theory extends to the other attitudes. Thus we have Frege Desires:

x wants_F that p = _df x would want ^┌p^┐ to be true iff x understood ^┌p^┐ .

And supposing that what you understand are meanings, we can define Frege Meanings :

x understands the meaning_F of ^┌p^┐ =_df x would believe ^┌p^┐ was true iff x believed_F p.

Frege called the meanings_F of sentences "thoughts" and the meaningsF of words "senses".  As Michael Dummett tells us:  "The sense is that part of the meaning of an expression which is relevant to the determination of the truth‑value of a sentence in which the expression occurs".  We can specify the Frege Meaning or sense of a term thus:

x understands the meaning_F of ^┌a^┐ =_df  x would believe ^┌..a...^┐ was true iff x believed_F ...a...

After Frege, when philosophers spoke about "meanings" (and after Frege, they spoke of little else) they were always talking about meanings_F.

These equivalences don't count as ampliative definitions since meaning_F and belief_F are defined reciprocally.  A proper analysis would have to define either 'belief_F' or 'meaning_F' in terms that did not assume the other.  However, despite generations of effort, finding a non-circular analysis has proved problematic.  It is one of those philosophical problems that has grown capital letters: it is " The Problem of Intentionality".

A word about the word "Intentionality".  As all the texts will tell you, the term was introduced as a philosophical technical term by Brentano, who intended it to pick out what was distinctive about the mental. So what is "intentionality"? Like so many philosophical technical terms, it can only be explained with a waving of hands.

" Intentionality has to do with the directedness or aboutness of mental states — the fact that, for example, one's thinking is of or about something."

"Intentionality is the power of minds to be about, to represent, or to stand for, things, properties and states of affairs."

This makes "intentionality" sound like some sort of spooky power.  But on reflection there seems nothing spooky, or particularly mentalistic, about "aboutness".  You can throw your arm about your friend.  The moon circles about the earth.  The price of gas is about four bucks a gallon. Nothing mental here.  These kinds of "aboutness" are just garden variety relations that one item can have to another.

What is distinctive about the "aboutness" that is supposed to characterize mental states is precisely that it is not a relation between a mind and what it thinks about.  Relations require relata.  You can't put your arm about something unless it exists.  On the other hand it seems you can think "about something" even if it doesn't exist.  Don't children have beliefs about Santa?  They must.  After all, they say things like “Santa is coming tonight."  Then, too, ordinary "aboutness" relations are relations between things, no matter what you call them.  A moon couldn't orbit about Hesperus without orbiting about Phosphorus; Hesperus is Phosphorus.  But apparently you can think about Hesperus without thinking about Phosphorus; after all, someone could think 'Hesperus has a moon' was true but deny 'Phosphorus has a moon'.

So what's special about the aboutness of intentional states is that these states manage to be "about " things without actually being... well, about things.  Which, come to think of it, is kind of spooky.

Frege's Mistake

In a recent post I told the story of Ralph, 'Hesperus' and 'Phosphorus'.  We saw there why Ralph might assert

('h≠p')    'Hesperus ≠ Phosphorus'

and how astronomical observation might lead him to change his mind. Remarkably, at no point in telling that story did we find any reason to deny that

(Bh=p)    Ralph believes Hesperus = Phosphorus.

It was not that we refused to take Ralph at his word.  Ex hypothesi, Ralph is a competent speaker of English and we took him to be sincere when he asserted ('h≠p').  We had to conclude that:

(Bh≠p)    Ralph believes Hesperus ≠ Phosphorus.

The principle at work here is an important one.  I call it (following Kripke) "The Disquotation Principle":

Lo, Hesperus. Lo, Phosphorus.

Here is Ralph. Ralph is a competent speaker of English. In particular, Ralph understands the words 'Hesperus' and 'Phosphorus' and knows what they refer to. When you are walking on the moors with him and he says "That's Hesperus!' or "There is Phosphorus!', you can be sure it will be Venus he's pointing at. Nevertheless, Ralph denies:

(HP)    'Hesperus = Phosphorus'

Though, of course, he would never deny:

(HH)    'Hesperus = Hesperus'

If we take Ralph at his word in denying (HP) we must conclude that Ralph believes that Hesperus ≠ Phosphorus in other words, that he believes Venus ≠  Venus. And given his endorsement of (HH) we may conclude that he also believes that Venus = Venus.

Ralph might puzzle us if we thought that psychological explanation must make sense of people in terms of what propositions they believe. But, as we have already observed, this is a mistake.

Psychology in Two Dimensions

It has been clear since the discovery of Twin Earth that there must be two different determinants of the propositional attitudes. Our Twin Earth doppelgangers are atom for atom identical with us.  Any psychological theory— any theory about how sensory inputs get translated into behavioral upshots— true of us must be true of them.  And yet, because of our different environments, we and our twins have different propositional attitudes.  Apparently, these differences in what we think and think about, must be irrelevant to psychology.  How can we reconcile this with our folk psychological conviction that people do what they do because of what they think?

Suppose that we one day have to hand The One True Psychological Theory of Human Beings.  Like any theory it will have a proprietary vocabulary– a way of carving up the world into psychological natural kinds– and a finite set of laws.  While we do not know what those laws are, we can be sure they will entail any number of nomolological regularities of the form:

(L)       ("x)( Ψ¹x at t₁ É Ψ²x at t₂)

where the ‘Ψ’s name individuals’ total psychological states; everything that can be said about them in the terms of the Theory.

The lesson of Twin Earth is that individuals might share the same total state but believe different things.  What propositional attitudes one has depends upon extra-psychological facts about the world in which one realizes that state and about one’s location in time and space in that world.   I’m going to call these locations in possible worlds situations (with applogies to Perry and Barwise).

I write:

in σ in Ψ: x believes p

to say that, in a certain situation σ, someone in state Ψ would believe p (among, presumably, many other things).  In Putnam’s story the situations and states are such that:

In σ_e in Ψ: x believes rain is H2O.

In σ_t in Ψ: x believes rain is XYZ.

It might be tempting to gloss all this by treating psychological states as functions from situations onto attitudes, but this would be a mistake.  Functions are identical if they yield the same values for arguments but psychological states can be different in psychologically potent respects even though the differences don't show up as differences in what subjects think or think about. (see. "Now,Me")

If we were perfect epistemic engines, our psychological states would be realized by true beliefs in every situation in which we found ourselves.  Alas we know that isn’t so and the situation dependence of the attitudes reveals another source of error.  A psychological state realized by true beliefs in one situation might be realized by false beliefs in another. 

Here is Oedipus insisting, “My Mom is not my wife”.  The proposition he believes is false; indeed, necessarily false by some ways of reckoning.  But Oedipus is not irrational or unfathomable, only badly situated.  Surely things could be arranged, behind the scenes, over on Twin Earth, so that Twin Oedipus is married to a woman who is not his mother, without it making any difference to what is in Twin Oedipus’s head. 

Twin Earth, so constructed, is what Dan Dennett, would call one of Oedipus’s “notional worlds”.  That is, worlds or , better,  (since Twin Earth is in our possible world) , situations in which Oedipus’s psychological state would be realized by true beliefs.

Dennett noticed that, for the purposes of explaining behavior, what matters is not what propositions subjects actually believe, but what they would believe in their notional situations.   This makes sense from an evolutionary point of view.  Presumably, true belief has survival value.   But evolution can operate only on what is inside our heads and what we believe and whether it is true is largely determined by what’s outside.  However clever we may be, we may still, like Oedipus, end up badly situated.  On the other hand, it would be a poor psychological design that allowed us to get into psychological states that could not be realized by true beliefs in any situation. Thus we might expect that the psychological states of well adapted organisms would be realizable by true beliefs in at least some situations.  Moreover we should expect those organisms to behave and think in ways that were appropriate to those notional situations.  As I will put it, your notional situations are the sorts of situations you think are in. 

As folk psychologists we are adept at understanding one another in this way.  If we are told that someone believes the proposition that Jocasta ≠ Jocasta, we have no way of telling what he might do.  On the other hand if we know that someone thinks they are in a situation of the sort Twin Oedipus is actually in, we understand his psychological state, know how he might have acquired it and how he is likely to respond to what happens next.

What happens next in the Oedipus story is that a group of messengers arrive telling tales.  Listening to them, Oedipus and Twin Oedipus will change their common psychological state until each them comes to think he is in a situation in which it is immensely probable that his mother is his wife.  At that point, in a transition described by laws like (L), the penny drops.  Oedipus and Twin Oedipus move into a new psychological state which is realizable by true beliefs only at situations in which they have committed incest.  Oedipus’s actual situation has become his, and Twin Oedipus’s notional one.  And both behave appropriately to that situation; they go ape, as anyone would if he thought he was in a situation in which he had married their mother. 

In the course of the story, Oedipus and Twin Oedipus acquire new beliefs given what they see and hear.  Assessed in terms of propositional content, they learn very different things.  The messengers hither are telling the awful truth about Jocasta; over yon, they are telling ugly lies about Jocasta’s twin.  Still, the Oedipi’s sensory inputs will be psychologically indistinguishable and will have identical effects on their psychological states and notional situations.   For the purposes of explaining and predicting behavior, the most productive way to characterize empirical uptake is not in terms of what propositions it is evidence for or what  propositions it might lead a subject to believe, but rather in terms of its effects on the subject’s notional situations.   By this accounting we can say that in our story both Oedipi make the same inferences based upon the same information.  Of course, they end up with different beliefs.  In Twin Oedipus’s case: necessarily false beliefs;  but that is beside the psychological point. 

Likewise it is natural to want to describe Oedipus' change of mind through the course of this story as an increase in knowledge.  By the end of the story he knows something about Jocasta he did not know before.  But, as we shall see in later posts, this sort of epistemic gain cannot be understood in terms of learning new propositions, but by coming to believe propositions already believed, by way of different, more informed, psychological states.  Knowledge is a matter of information, not justification.

Of course, mistakes about identity have had a special philosophical luminance ever since Frege wrote about The Morning and The Evening Star.  In the next post on this theme, we will put two dimensional psychology to work at sorting out sense and reference.

Open Access: What is to be done?

Thanks to  all those who commented on the Open Access post. The comments lead me to a rich thread on the same theme, which I was not aware of, at the Leiter Report from 2006, which everyone should read.

Though my post has had many viewers, thanks to a link from Brian Leiter there has been no stampede of senior professors signing up to take The Pledge.  The heroic exception is Rick Grush who declares himself a "Commercial Free Philosopher" and invites others to join him and show support by posting his logo.  Grush's manifesto "Why I am a Commercial Free Philosopher" is sound and eloquent and should be read by everyone interested in this theme. 

Open Access Philosophy

Why are there still hard copy philosophical journals and books?  Why is so much on-line philosophy hidden behind subscription walls?  Why are universities, students and researchers being forced to pay for access to information authors would happily give away for free?

Who disagrees with this:

The Internet has fundamentally changed the practical and economic realities of distributing scientific knowledge and cultural heritage. For the first time ever, the Internet now offers the chance to constitute a global and interactive representation of human knowledge, including cultural heritage and the guarantee of worldwide access.

Our mission of disseminating knowledge is only half complete if the information is not made widely and readily available to society. New possibilities of knowledge dissemination not only through the classical form but also and increasingly through the open access paradigm via the Internet have to be supported. We define open access as a comprehensive source of human knowledge and cultural heritage that has been approved by the scientific community.

In order to realize the vision of a global and accessible representation of knowledge, the future Web has to be sustainable, interactive, and transparent. Content and software tools must be openly accessible and compatible.

Berlin Declaration on open access Knowledge in the Sciences and Humanities

Shouldn't philosophers be especially sensitive to the moral and intellectual imperatives of the open access movement?  Why is it that scientists have been so much more ready to embrace it than philosophers?

When I put this question to professional philosophers I hear two different kinds of arguments. One foolish and one cynical.

Philosophical Intuitions

In philosophy, your first intuitions about a problem are the opinions you hold after you've discussed it for a hour with someone who has thought about it for a lifetime.

Thrown for a loop

[I don't like blog software's treatment of comments. Reading through a comment thread is like reading a series of footnotes to footnotes, not a conversation.  So my practice will be to try to include good comments in the main body of the blog where they get the attention they deserve.]

Responding to my proposed solution to The Trolley Problem, Richard Chappell writes.

I dunno, the distinction between (directly) preventing the deaths vs. causing them to be prevented (by someone else) strikes me as a bit artificial. But in any case, I don't think it tracks the standard Trolley intuitions. Consider the loop case: (8) You can pull a switch to divert the trolley onto a loop which connects back onto the main track where 5 victims await. Normally this wouldn't help, but fortunately there's a fat man tied to the rails in the loop section, who will bring the train to a halt if it hits him. Should you pull the switch? Presumably anyone who would divert the trolley in the original case would also do so in the loop case. But you would say it's the fat man, rather than the switch-puller, who directly prevents the 5 deaths in this case. So directly preventing the deaths cannot be the grounds for pulling the switch. An alternative candidate explanation would be to appeal to whether Jones brings about the one death (saving the five) by acting on the victim (e.g. pushing him), or by indirect means (pulling a switch). Perhaps our intuitions reflect the principle that only the latter (i.e. indirectly causing death) is permissible.

Thanks for your comment!

Let’s spell out the relevant cases:

DIVERT:  A runaway trolley is coming down the tracks.  It will hit five people if it is not diverted down a different track, but there is one person who will be killed if the trolley takes that route. Seeing this, a bystander throws a switch which diverts the trolley.  The one is killed; the five survive; the trolley continues on its way.

PUSH:  A runaway trolley is coming down the tracks.  It will hit five people if it isn't stopped.  A bystander pushes a fat man onto the tracks.  As the bystander intended, the trolley hits the fat man, killing him but stopping the trolley before it hits the five.

LOOP: A runaway trolley is coming down the tracks.  It will hit five people if it isn't stopped.  A bystander throws a switch which diverts a trolley down a siding.  The siding loops back to the main line and is still headed for the five, but there is a fat man on the siding track who is struck by the trolley, killing him, but stopping the trolley before it hits the five. All as the bystander intended.

Thomson’s LOOP case is a kind of parlor trick that flummoxes everyone when we first encounter it but I think its significance is generally misunderstood...

A Solution to the Trolley Problem

In the trolley problem some people think it is permissible for a bystander to turn the trolley onto the one to save the five. But many of those same people do not think it permissible to push a fat man onto the track to stop the trolley from running over the five.

The Trolley Problem is not to say whether this opinion is correct, but to explain what non-moral difference between these cases could reasonably lead people to these different verdicts in these two cases.  This is a problem even for someone who does not think it permissible to turn from the one onto the five.

In what we may call the standard trolley case a bystander has the choice of throwing a switch which will divert the trolley or not throwing the switch. I take it that people who believe that it is permissible to throw the switch in the standard case believe the following.

It is permissible for you to do something, α, that kills one person if:
        i) by doing α you prevent the death of five people.
        ii) if you refrain from doing α the five people will die.

The Problem is why this principle does not also permit pushing the fat man in front of the trolley. 

To see why consider the following cases.

Trolley Poem

The Trolley Problem

Most people, they say, will turn from the five
onto the one.
I'm sure they're right.

Most people got no idea about railroading.
Don't know how to start one, don't know how to stop.
But they'll turn on you.
Bystanders. Most People.
Standing by to run you down.

Most people will raise your taxes.
Most people will take your house for a ball park.
Most people will send your boy to die in a foreign war just because
they think they might see something,
way down the road,
they'd rather not run into.

Most people stick to the main line.
Most people don't trust the one.
What's he doing out there,
on his own track,
outlying?

So they'll grind that one under
and won't think twice about it.

Most people can do the math.