It is taken to be a platitude that one person can’t be morally responsible for the actions of someone else (see, e.g., Ted Sider’s book Four Dimensionalism for a recent reaffirmation of this claim). But there seems to be a fairly simple argument against this view, an argument drawing from Derek Parfit’s work on personal identity. Suppose that Adam has committed a brutal murder, and he then undergoes fission, that is, one half of his brain is put in one identical triplet’s body, and the other half is put in the other identical triplet’s body (call the two other triplets Brett and Carl). Let us then stipulate that each of the resulting persons is qualitatively identical to Adam. What’s happened to Adam? There are the usual four options: Adam survives as both, survives as Brett, survives as Carl, or doesn’t survive. There are clearly two persons in existence after the fission, though, so because one person doesn’t equal two, Adam can’t survive as both. And there’s no non-arbitrary reason for why he would’ve survived as Brett and not Carl, and vice versa. So the best description of the case is that Adam hasn’t survived the fission. But this isn’t even remotely as bad as ordinary death; indeed, what’s occurred to Adam is just as good as ordinary survival: both Brett and Carl will (quasi-)remember Adam’s commission of the crime, they’ll fulfill his intentions (at least insofar as their duplication won’t cause pragmatic conflicts in carrying them out), and they’ll persist in his beliefs, desires, goals, and general character.
But what about responsibility for his actions? If the loss of identity in this way is unimportant with regard to ordinary patterns of prudential concern, then why should it matter with regard to moral attributions like responsiiblity? What matters instead, it seems clear, are the relations of psychological connectedness that obtain between Brett/Carl and Adam. But insofar as these relations obtain to the highest possible degree, there seems no reason to deny that both Brett and Carl are responsible (or least quasi-responsible) for Adam’s actions, which is a straightforward denial of the so-called platitude: one person can be responsible for the actions of someone else.
This may just be because I can’t seem to understand how responsibilty can obtain without relying on the identity of the individual involved in the act in question; but, could you explain what you mean by “psychological connectedness”? I’m having a difficulty understanding how that isn’t just another way of saying the perpetuation of the identity of Adam. It seems that the degree to which brett&carl bear (quasi-)responsibility is the degree to which they are in fact the continued existence of the agent Adam.
Dave,
Good to have your first post on board.
It seems to me like you’re just using one or more platitudes (one person can’t equal two, arbitrary reasons are bad reasons) to reject another (one person can’t be responsible for the actions of another).
So why couldn’t someone just modus tollens your modus ponens? That is, couldn’t someone just say that since it’s clear that responsibility attaches to the agent who committed the act in question, something must be wrong about our usual thinking about personal identity? Here’s a less polemical way of putting it. Since we’ve got incompatible platitudes, how should we decide which to keep and which to discard?
First, to waste’o’space (sounds like an insult!): psychological connectedness consists in certain direct psychological relations that obtain between two person-stages, and these include direct connections of memory, the fulfillment of intentions in action, persistence of beliefs, goals, and desires, and resemblance connections (e.g., character). Psychological connectedness cannot be the same thing as the identity relation, however, simply because psychological connectedness can hold one-many, whereas the identity relation (being a mathematical relation) must hold one-one. So Brett and Carl cannot be identical to Adam, given that transitivity would require them to be identical to each other, which they are not (in the most succinct terms, what’s true of one is not true of the other, insofar as they occupy different locations in space/time). So to explain fission cases, we have to look for another relation that obtains between the pre-fission person and the fission products that makes sense of the fact that everything that matters in ordinary survival has been preserved. The best explanation seems to be the robust psychological connectedness that obtains. So this is the relation that matters in identity, and we ought to see it as the relation that underlies our ordinary patterns of concern (the patterns we usually think track the identity relation). (Incidentally, Parfit thinks it’s psychological *continuity* [overlapping chains of connectedness] that matters here, but I’ve argued in “Selves and Moral Units” [Pacific Phil. Quarterly, Dec. 1999] that it’s really connectedness that’s doing all the work.) So if responsibility attribution is one of those patterns of concern, then it should also track connectedness, which means one person can be held (q-)responsible for the actions of another.
As for Josh’s comments, I think it’s a real stretch to hold that “one person can’t equal two” is a platitude on a par with “one person can’t be responsible for the actions of another.” The former is an instantiation of a mathematical principle about identity (“one doesn’t equal two”), whereas the latter is an assertion making reference to a fairly value-laden concept (responsibility). Once we see (via the fission case) that our values actually commit us to something quite different, we should abandon the “platitude,” given that it’s not something we really believe at the end of the day anyway. I submit we don’t have the same cognitive option when it comes to the straightforwardly mathematical claim.
Thanks for the reply, though that’s not quite what I had in mind – I can now see that my earlier post was ambiguous, so let me try to be a little more precise. First, I wasn’t taking responsibility in any value-laden sense (not that that’s a *bad* thing); I had in mind responsibility as attribution, where it makes sense to say person P did action A, regardless of whether A is good, bad or indifferent. (Here I’m not sure why you say that the claim that one person can’t be responsible for the acts of another is not something “we really believe at the end of the day,” since you yourself call it a platitude – maybe you mean we *shouldn’t* believe it once we subject it to critical scrutiny?) I also agree, of course, that the mathematical “one doesn’t equal two” is harder to abandon than the responsibility platitude. I was thinking, rather, of a personal identity platitude, that one person can’t *become* two (psychologically connected but physically distinct) people – in, say, fission cases. Maybe your original post is assuming some arguments with which I’m not familiar (very possible), and so I’m just missing something here. But, in any case, as I put it in my original comment, this kind of claim is about personal identity (not mathematics). I hope it goes without saying, I’m not sure what to think about all of this stuff, but it doesn’t seem obvious which we should disrupt – our platitudes about action or those about personal identity.
Josh: The “one person can’t become two” platitude is grounded squarely on the thought that personal identity is a mathematical relation (it’s the identity relation, after all), so it remains exactly on a par with “one doesn’t equal two.” Nevertheless, you might be saying, why can’t we at least consider prizing apart the concept of *personal* identity from the concept of identity simpliciter? Now this wouldn’t be a wholly insane move, but it would still require such a massive change in our conceptual apparatus as to render the concept of “person” nearly unrecognizable. (For example, if we allow that Adam has become Brett and Carl, then if Brett shoots Carl, is it a murder or a suicide? If Brett and Carl play tennis, is it really solitaire? If, through a bizarrely incestuous turn of events, Brett and Carl make love, is it sex or masturbation?) So abandoning this conceptual requirement would wreak serious havoc with our language in a way abandoning the platitude “one person can’t be responsible for another’s actions” wouldn’t. (So I guess I’ve just offered a consequentialist response to the question at the end of your first comments.)
Yeah, that prizing apart is what I had in mind…though your comments go a long way to showing that’s also not a very promising move in any case. Thanks.
First, two questions:
-What if Adam had taken out half of his brain & destroyed it, but remained living with the other half?
-What if Adam had kept half his brain & put half in Brett in the same manner as in the original?
Now, unrelated to those questions, some paragraphs:
I think that the kind of “identity” that matters for “personal identity” cannot be simple mathematical x = x identity. x = x just says that anything is the same as itself, but persons are always changing, so we need another definition. If we let each variable represent a thing at an instant, I would say that personal identity is only equivalent to the identity function A = A at a single instant, which is trivial and uninteresting, as it merely says that any thing at an instnat is identical to itself at that instant. But looking over time, identity is more like a mathematical relation of becoming, A–>B, meaning “A becomes B (in the right way)”. This relationship is not symmetrical: A–>B does not imply that B–>A. In fact, symmetry never holds, except in the trivial case where A = B (i.e. we’re talking about one thing at a single instant with two names). Transitivity does hold, though. A–>B & B–>C implies A–>C.
There are plenty of instances where one becomes two, like when a bacterium divides, or two clouds split, or one sovereign state secedes from another. After one item splits into two, there are some ways in which we treat each of the two like the original and some ways in which we don’t. After one item changes significantly, there are some ways in which we treat it like the old version and some ways in which we do not. Retreating into 1 = 1 mathematical certainty is not going to erase these complexities.
Suppose a personified bacterium chose to invade me & make me sick. Once in, it couldn’t have left even if it wanted to & it couldn’t help but make me sick. It split into two identical bacteria, which then proceeded to replicate into millions of identical bacteria. Could we blame any or all of these bacteria for invading me & making me sick? I say sure – one bacterium can become many, and each of those many can be identified with that one that they came from, even though they can’t be identified with each other.
I think this means denying the theorem that “if A–>B and A–>C then either B–>C or C–>B”. That is, it means accepting that A can become multiple unrelated things.
I don’t think that the one-can-become-many hypothesis or this formalization produce much havoc. As long as there’s a neat temporal cut-off where the one splits into more than one, we simply say that, with respect to any event after that splitting, the offspring are different & separate creatures. With respect to anything from before the splitting, they all inherit the status of the original in the same way that one individual maintains its properties over time. So, I’d say Brett & Carl are different people with respect to everything they’ve done since the brain splitting (so it’s sex, tennis, and murder), though they both “are” Adam in the same way that I “am” younger-me, and thus are responsible for what Adam did before the splitting (where “is” means “comes from (in the right way)”). There may be some properties that depend on both pre-split and post-split circumstances and are no longer easy to define after some cases of splitting (like the age of a bacterium). If there’s havoc to be found, it’d be in this third category. But, assuming that responsibility is a historical characteristic, it seems to carry on unchanged from the moment of the murder onwards.
I don’t think we should be too startled about finding some language & conceptual difficulties when we imagine brain transplants & person splitting. There’s no need to rush to whatever concept of identity seems certain and simple just to avoid these complexities. (In this way, at least, philosophers ought to be different from vehement “life begins at conception” believers.) Why would you expect our language & concepts to be ready to easily address issues of person splitting? Maybe if we reproduced asexually things would be different.
To Dan (Boisvert??): For your first two questions, if 1/2 of Adam’s brain had been destroyed, but the other half was functionally equivalent to the destroyed half, then Adam would unquestionably be the survivor. Nevertheless, if he keeps 1/2 and the other 1/2 is put in Brett’s skull, the best description of the case is now that Adam didn’t survive. Two people would now exist who are (psychologically) exactly similar to Adam, but both can’t be Adam (because both would have to be identical to each other, which is absurd). Some may want to say that because 1/2 the brain is with Adam’s original body that that gives us a non-arbitrary reason to favor Adam as the survivor, but that’s a kind of sentimentality about the physical that I doubt warrants such an identity-judgment.
At any rate, you want to replace the identity-relation with a sort of “becoming” relation, because a thing (X) is identical only to itself at an instant and so the identity relation is rather trivial/useless when it comes to most items in the real world. But if the idea of descriptive metaphysics is to describe the structure of our thoughts about the world (cf., Strawson’s _Individuals_), then this is to ignore a central feature about our thinking about the world. In other words, we *do* ascribe identity to objects that undergo all sorts of physical changes (and that we’re *well aware* undergo such changes). So the chair I’m sitting in now is identical to — it’s the same chair as — the chair I was sitting in yesterday. My dog sitting by my side is the same dog as the one who was doing so yesterday. And so forth. I don’t say that the chair *became* the chair I’m now sitting in, and I certainly don’t say that my dog yesterday *became* the dog that sits by me now. Identity across time is perfectly compatible with change of physical parts. But insofar as there’s nothing significantly different about persons than there is other objects about which we make identity judgments (where those identity-judgments involve the mathematical, “equals” sign), there’s no reason to withhold application of the mathematical identity relation to persons as well. Parfit’s main point (and the one I was drawing from with my comments on responsibility) is that, when it comes to persons, anyway, the identity relation just isn’t what matters with regard to our patterns of concern.
Hi David. I’m not Boisvert, so I’ll start including my last initial with my posts.
I’m not attached to the word “becoming” or any other particular way of putting my concept of personal identity into English words. I’m fine with saying that the dog *is* the same dog as yesterday, but there are certain ways in which the vague phrase “is the same” may be misleading, so I’m trying to formalize it to clarify the concept. Perhaps saying that your dog yesterday “later is” the dog sitting next to you now is more intuitive than saying “becomes”.
I think that forbidding one person from becoming two leads to unsatisfying consequences, as in the two cases that I raised and you responded to. It seems to me that whether new, half-brained Adam is the same person as old, full-brained Adam should not depend on what happens to the half-brain that has been removed from Adam during the operation. Further complications can make the difficulty more acute. For instance, does new Adam become old Adam again if Brett dies? What if the death occurs during or just after the brain transfer? Or, what if Adam’s removed half-brain is preserved for a week before being put into Brett? Does the original Adam then survive until Brett gets his new half-brain?
I would rather say that, in your original case, both Brett and Carl are Adam after the brain transfer. You reject this solution because you think that it implies that Brett is Carl. I think we can avoid this implication with a fairly minor formal change to our definition of personal identity, which I tried to introduce in my earlier post, and will try to explain more clearly here. [Here, each variable (A,B,…) represents a person at an instant.]
If we say that A and B are the same person, using your preferred criteria for personal identity, it is implicit in that statement that one of them (say A) exists before the other one (or else they exist at the same instant). I am suggesting that we supplement your concept of personal identity with this already implicit fact. This fact makes personal identity directional, so we should write something like A–>B instead of A=B (if A and B exist at the same instant, then it’s true both that A–>B and that B–>A). In English, you could say that “A later is B” or that “B is a later version of A” (remembering that they may also exist at the same time), or, speaking informally, that “A is B” or that “B is A”. Even though both “A is B” and “B is A” are correct to say, we may not write both “A–>B” and “B–>A”. Since B exists later than A (or simultaneously), B–>A is generally false when A–>B is true. “–>” is not symmetrical.
If you were to neglect the directional nature of personal identity and consider it an equivalence relation, then if you had A=B and A=C you would conclude that A, B, and C were all the same person, and that B=C. But, with our asymmetrical operation “–>”, if you have that A is B (A–>B) and that A is C (A–>C), it does not follow that all three are the same person or that B is C. You cannot make any conclusion about the relationship between B and C. Thus, it is perfectly consistent to say that Brett is not Carl, even though Brett is Adam and Carl is Adam.
This directional definition of personal identity changes two things: it allows one person to become several different people, and it allows several different people to become the same person. (This latter possibility is allowed by the formalism, but may be ruled out by the content of what it means to be the same person.) Otherwise, it works just the same as the definition that David has been using. We can still say, for instance, that Dirk on his first birthday, Dirk in second grade, Dirk three weeks ago, Dirk yesterday, and Dirk today are all the same person. This is because “–>” is transitive, and we can consider a person at several different instants to all be the same person provided that we can link them in a chain, like D(1)–>D(2)–>D(3)–>…–>D(n-1)–>D(n). When person-splitting occurs, as with Adam becoming both Brett and Carl, you can’t create such a chain.
Very interesting stuff, Dan (K.). The question is going to be just what the advantage is of introducing your non-standard conception of identity. Now as I see it, the advantage of your conception is that it preserves a kind of identity-talk in the fission case (without implying identity between Brett and Carl), and it can also avoid the weirdness of saying that, when Carl and Brett are alive, neither is identical to Adam, whereas if Carl suddenly dies, Brett is Adam again. The cost is that we lose the ordinary understanding of identity as, well, *identity* (the equals sign). The advantage of my (well, Parfit’s) view is that the ordinary understanding of identity is preserved, and the disadvantage is that in fission cases (and the collapsing Carl case), you get the weirdness that my identity can change depending on what happens to other people. Now to mitigate that “disadvantage” I follow Parfit in suggesting that the identity-relation just isn’t what matters to undergird our patterns of concern; instead, what matters are the psychological relations (connectedness and/or continuity) involved, which can take a one-many form.
So it looks like we’d both say the same things about the patterns of concern involved (e.g., Brett and Carl bear responsibility for Adam’s actions), but we differ over why (you think it’s because there’s a sense of identity/becoming that’s preserved, while I think it’s because psychological connectedness obtains). The real question, then, is whether or not we want to adopt the non-standard conception of identity you employ. I value the preservation of concepts as much as as possible, so I’d opt against doing so. But at the end of the day, given the *practical* agreements involved, it may not matter so much either way.
Well, Dave you can imagine my answer on this one, but that would make for a long, tangential post… So…
What leads to the conflict here is a conflation of terms. In Parfit’s view personal identity does not matter, so platitudes involving talk of persons are simply no good. Persons are not important concepts in the Parfitian framework. This is a further problem (in my mind anyway) of using Parfit’s notion of the R-relation when applied to ethics and morality. That is, the more general question of where responsibility is cast in the first place…
Using standard moral practices we place blame (or praise) on the person who committed the action. But this, as Dave example shows, would not cast the net wide enough to encompass all possibilities of moral blame. That is, if the R-relation (psychological connectedness and continuity) is what proves to be important to us, then that ought also to be the case in attributions of praise and blame. In Adam’s case, he died (because the Partian uniqueness clause does not hold), but is psychologically connected to these two post operation people. Since the connectedness is what really matters to us (according to Parfit), Adam , in spite of his dying, has met a fate as good as normal survival. That is all to say, attributions of praise and blame ought not to apply to people, in Parfit’s conception, but R-related sums.
Here however, is the concern… How are we supposed to know what an R-related sum is? Persons we can pick out. They do not change so drastically as to render them difficult to discern. I know that it is my boss that I see every morning, and hold him responsible for his treatment of me according. If he’s a jerk, I am less cordial the next day as a measure of moral blame, for example. But how is this to be done (on practical terms) if people are no longer the focus of moral praise and blame, but R-related sums are? Since the R-relation holds as a measure of metal connections, often times connections that are not directly accessible to the moral community, how is it that we are to determine exactly what R-related sum to hold accountable for this-and-such action?
“… so because one person doesn’t equal two, Adam can’t survive as both.”
Why not? It seems to me that in your hypothetical you have accomplished exactly that. Do you have some reason for thinking that identity is unduplicatable, and what beyond “it’s never been done yet”?
This seems like any interesting problem, and I think that David’s right to conclude that we can be responsible for the actions of another. We should be clear, though, about the notion of responsibility at work here. It seems clear that neither Brett nor Carl can be causally responsible for Adam’s actions since neither actually caused these acts in any way. I take this claim to be relying on a basic metaphysical intuition pretty much on a par with our intuitions about the identity relation. At the same time, we do want to say that Adam cannot evade responsibility for his crimes by undergoing fission. It seems that, here, we are invoking both moral and legal responsibility. For simplicity sake, let’s just focus on moral responsibility. Now my intuitions here aren’t as strong, but I would like to find a way to say that Adam doesn’t get away with murder. It seems that the only non-arbitrary way to do this is to say that both Carl and Brett are responsible for his murderous act. But I still don’t think that we’ve faced up to what a drastic overhaul in our ordinary viewpoint this is going to require. We tend to think that fission is not a responsibility removing operation. At the same time we have very strong intuition that an agent is responsible for an action if and only if she caused that action (directly or indirectly) and she could have done otherwise. It’s not clear to me why, in a competition between these apparently inconsistent intuitions, it is the latter which has to go. I suspect that David doesn’t see much problem here because he already has, what I would regard as, a non-standard moral theory which rejects this latter intuition. But with a more standard appoach to ethics (and here I would say that any ethical theory which is not committed to this latter intuition is non-standard), it is just not clear to me that the case of Adam, Brett, and Carl provides enough motivation to sacrifice this so-called platitude. If I am faced with a choice between the, admittedly, dissappointing conclusion that Brett and Carl are innocent and the wholesale rejection of my ethical theory, is it really that obvious that I should opt for the wholesale rejection?
Of course, this is probably a false dilemma, but I think it does make it clear that something more is at stake than just the simple rejection of an assumption that most of us, unreflectively, take for granted.
(read a little further)
“…whereas the identity relation (being a mathematical relation) must hold one-one. So Brett and Carl cannot be identical to Adam…”
I think here you have conflated mathematical identity with personal identity. I-now am not matematically identical with I-yesterday; atoms have changed, chemical processes have processed, new ideas have formed, etc. By extension, Brett-after-the-transplant is as much Adam-before-the-transplant (assuming as you do and we all must in this hypothetical that the transplant itself doesn’t destroy self) as Carl-after-the-transplant. And by your hypothesis, that is as much as you-now are identical to you-last-week.
You are still failing to persuade me that one can’t become two, or NOT( Brett IS Adam AND Carl IS Adam ).
“The cost is that we lose the ordinary understanding of identity as, well, *identity* …”
Dave, I think the problem is that the ordinary understanding of identity isn’t suited for and can’t handle the situation you’ve described. Much like ordinary (old) understandings of time fell down when considering objects travelling near light speed. When/if fissions like you describe become possible we’ll all update our concepts of identity to match.
Until then, some of the choices you are making are opposite what I’d make. To say that Adam with half his brain missing IS or IS NOT Adam depending on whether the missing have was destroyed or placed in Brett is slap-in-my-face wrong. It’s placing that determine identity well outside where they should be, i.e. with the person in question.
How about before you try to determine if Brett and/or Carl are responsible for the murder, you first try some more edge cases with identity. For instance, what happened in the scifi thriller “Dark City”: is Adam still Adam if all his memories are erased and replaced with some other memories? Also, check out the novel “Fire upon the Deep”: science fiction where several of the characters have alien identities.
Replying to Craig,
I think you are approaching this dilemma in a methodologically different way. Dave’s example, and the Parfit’s “Reasons and Persons” are in the business of dissolving our attraction to the concept of personal identity. As a fact, people believe they persist through time as themselves and only as themselves, despite their atom-changing from various time-slices of themselves. The mathematical identity Dave was talking about IS a part of this notion. I do not exist as more than one person, and our concept of personal identity, and the usefulness of referencing persons at all, is dependant on the fact that two people cannot be one. Adam cannot survive in two different bodies.
Think of it this way… Adam’s brain goes into two separate bodies, and he (let’s just stipulate) survives as both. These two bodies travel to distant parts of the Earth and collect a number of distinguishing characteristics and memories. Are we still to hold that after 20 years of changing that they are still the same person? Seems that we can’t. Hence Dave’s stipulation that “one person cannot equal two.” It simply destroys our notion of personal identity to make such a claim.
Parfit uses these example to show that our concern in personal identity is really a concern for something else, namely a kind of psychological relation. This, it turns out, is a one-many relation and not the one-to-one relation of personal identity. Our notions of personal identity are not going to change when this sort of fission becomes actual, it’s our concern for it that will change.
I was wondering about a more “ordinary” kind of case. Can’t I be morally responsible for the actions of my own child (in certain contexts). So, say I allow my thirteen year old boy to take the family car on a little spin, just for fun, through the neighborhood. Surely I can be held responsible for my own act of giving him the keys, and so forth. And I can be held morally responsible for some of the reasonably foreseeable upshots. I think I can also be held morally responsible for my boy’s driving the car–for various of his actions. No?
John is right. There are cases where we hold someone responsible for something that she did not do. But I don’t think such cases threaten our commonsense notion of morality in the same way that Dave’s example does. We can fairly easily account for such responsibility by qualifying the platitude in noncontroversial ways. But I can’t imagine any noncontroversial way of making Brett and Carl responsible for Adam’s action since they did not exist at that point.
I had meant to say earlier that I think Chris’s reply to Craig is right. If we’re to keep our concept of identity (as it applies to all sorts of metaphysical objects) intact, we have to view it as a one-one relation. But what the fission case brings out is that the only thing we’re losing is the *uniqueness* relation (the one-one relation), so Brett/Carl can’t be identical to Adam. But so what? Everything else that matters has been preserved, so identity can’t be what matters for those patterns of concern.
As for Jeremy, thanks for the comments (and good to hear from you). I wonder if you could say more about your occasional comments that my view requires a “non-standard” ethical theory, given (a) I haven’t really stated any particular ethical theory, and (b) the only non-standard element involved, it seems, is the denial of the platitude about responsibility.