Story of I:
1. In her thorough and comprehensive study Persons and Bodies, Lynne Rudder Baker argues for what she calls the Constitution View of persons. This view comprises two strands. According to one strand, persons are constituted by, but not identical with, their bodies. According to the other strand, persons are essentially self-conscious beings with a distinctively first-person perspective on the world. The book is basically an elucidation and defence of these two strands.
I am pretty much in agreement with Baker on both strands. I agree, contra the Animalists, that persons are not identical to their bodies/brains, but reject Dualism; so I agree that the relation between a person and his body is constitution-without-identity. And I agree that persons are distinctively and uniquely self-conscious beings, with a first-person perspective on themselves and the world. (My own views on these matters, for what they’re worth, are set out in my short book Personal Identity and Self-Consciousness (Routledge, 1998).)
However, I found it a bit odd that both these strands were described as comprising the Constitution View, as if the two strands formed a unified view. I would have thought that ‘Constitution View’ is really only a name for the first, metaphysical, strand; after all, many different metaphysical conceptions of the relation between a person and his body could agree that there is an intimate link between personhood and self-consciousness. But this may just be a book-keeping point. What of more substantial issues? I have no intention of discussing every issues Baker raises, and will confine my comments largely to claims made in Chapters 3, 5 and 9.
2. Baker begins her discussion of self-consciousness by distinguishing “two grades of first-person phenomena: weak and strong” (p. 61) Only subjects of strong first-person phenomena have a first-person perspective (and are self-conscious). Why draw this distinction? Because many creatures (animals, infants, etc) have an egocentric perspective on the world, and much of their behaviour can only be explained from that perspective. (As Baker puts it: if a “dog could speak, he might say … ‘There’s a bone buried there in front of me, and I want it.’” (p. 61)) Yet they nonetheless lack a first-person concept and are not self-conscious (not persons). Creatures who exhibit weak first-person phenomena can distinguish between first and third person standpoints, but cannot conceptualise that distinction (cannot “conceive of oneself as oneself” (p. 64, my italics).)
The distinction between weak and strong first-person phenomena can also be characterised linguistically: subjects of weak first-person phenomena can make first-person attributions (‘I am tall’), whereas subjects of strong first-person phenomena can also attribute first-person reference to themselves (‘I believe that I am tall’). The latter kind of sentence Baker, drawing on ideas of Castañeda, calls an ‘I*’ sentence. Thus thinking ‘I think that I* am F’ “is an indication that one is entertaining an I* thought” (p. 65) and hence has a first-person perspective. On this view, ‘I’ and ‘I*’ express different concepts – a toddler, for example, may possess the former but not the latter. “For a being without a concept of itself as itself*, ‘I’ is just a marker of perspective. Acquisition of a first-person perspective brings with it a genuine self-concept … “ (p. 69).
3. Now I‘m not totally unsympathetic to the spirit of this, but let me make a couple of comments. First, even on her own terms, I don’t see the need for Baker to invoke the ‘*’ terminology. Nor am I sure I understand what ‘I think that I* am F’ is supposed to mean if it means anything other than ‘I think that I am F’. Why didn’t Baker just put her point this way: beings with just a weak first-person perspective can think basic or non-iterated ‘I’-thoughts (‘I am F’), whereas beings with a strong first-person perspective (self-conscious beings, persons) can also think iterated ‘I’-thoughts (‘I think that I am F’). So being able to think iterated ‘I’-thoughts is the hallmark of self-consciousness, and there is no need to make any Castañeda-style comment on, or supplement to, the second occurrence of ‘I’ in such iterations.
Perhaps I can fill out this point. When Castañeda introduced the ‘*’ terminology he did so with respect to the pronoun ‘he’ (not ‘I’). And for good reason. A sentence like ‘John believes that he is bald’ is at least three-ways ambiguous. It could mean that John believes that Bill is bald (because ‘he’ refers to Bill in that context); or it could mean that John believes that John is bald (without realising that he is the John in question); or finally it could be that John is having a fully self-conscious thought about himself, that we would express perspiculously by saying: John believes that he himself is bald. Castañeda’s ‘*’ just marks the ‘himself’ spot, indicating a fully self-conscious thought. Now here the ‘*’ terminology has a point and marks a contrast. But there simply is no such contrast to be drawn in the case of ‘I’: ‘I’ could not but be ‘I*’, so to speak. Could Baker re-phrase her point in terms of ‘he*’? What would her concession to animals and toddlers be then: the dog believes that he* is about to be attacked? But then how would the dog contrast with us? She could say: no, all we can say is that the dog believes that he is about to be attacked. But how do we now characterise the low-grade first-person perspective allegedly had by the dog? (That is, what now makes it worth calling a first-person phenomenon?)
Second, I want anyway to disagree with Baker’s starting point here. I don’t think beings who have perspectives on the world, and can modify their behaviour in response to input from their environment, should thereby be attributed ‘I’-thoughts. Rather, my view would be: there is a simple connection between having ‘I’-thoughts and being self-conscious, viz., a being is self-conscious (has a first-person perspective) iff that being has, or is capable of having, ‘I’-thoughts. Animals and toddlers don’t have ‘I’-thoughts (even though they have some sort of egocentric perspective on the world), hence they’re not self-conscious.
Why prefer my account to Baker’s? First, I would not want to attribute ‘I’-thoughts to any old thing that has a perspective on the world and is capable of adjusting its behaviour to fits its goals. Doubtless certain kinds of missile have in-built maps with themselves as origin and can modify their speed and direction in sophisticated ways. But missiles don’t have ‘I’-thoughts. This point is hardly knock-down, but the onus is now on Baker to say how she can treat missiles differently, in this respect, from animals and toddlers. Second, it’s not obvious to me why a being who can have non-iterated ‘I’-thoughts is not also capable of having iterated ‘I’-thoughts. If a thinker can think ‘I think that p’ why can’t he also think ‘I think that I think that p’? Baker would have to insist that iteration involves a massive and qualitative conceptual leap. The trouble is that it doesn’t seem to at all. Third, it seems to make sense to attribute a being present-tense ‘I’-thoughts (‘I am F’) only if it also makes sense to attribute to that being past and future (and maybe counterfactual) ‘I’-thoughts (‘I was F’, ‘I will be F’, ‘I might have been F’). But it’s surely implausible to attribute the latter types of thought to animals and toddlers; so it’s also implausible to attribute to them simple present-tense ‘I’-thoughts.
In sum, I don’t think that the difference between subjective perspective and self-consciousness is usefully understood in terms of a distinction between basic and iterated ‘I’-thoughts.
4. Baker then makes some comments about some features of the first-person perspective (pp. 69 – 76). I agree with her that all uses of ‘I’ are immune to reference-failure. However, I don’t think that Wittgenstein meant to imply that ‘as object’ uses of ‘I’ are not so immune (if that’s what Baker thinks). The distinction between ‘as subject’ and ‘as object’ uses was, I think, meant to be an epistemic distinction, corresponding to features of different ways of knowing truths about oneself. (Again see Chapters 7 & 8 of my book, and references therein to Evans’ work.)
Baker then defends an argument to show that the first-person perspective is relational in a certain sense (pp. 72 – 76). I would have liked more discussion of premise (2) of that argument (“x can think of herself as herself* only if x has concepts that can apply to things different from x”). This is presumably supposed to be a necessary truth, but is it obvious that it is? The fact that developmental psychologists accept it is hardly relevant to the question of its necessity. Is it metaphysically or logically impossible for there to be a Solitary Egotistical God? (Issues to do with private language bear here too.)
5. In pp. 76 – 79 Baker argues for the indispensability of the first-person perspective. The first sort of ineliminability concerns language: “First-person reference is not eliminable from I* sentences, whether it is eliminable from simple, direct-discourse ‘I’ sentences or not.” (p. 76) Leaving to one side my earlier qualms about the ‘*’ notation, I agree with this remark. But Baker is willing to give some credence to the Russell-Geach view that first-person reference is eliminable in direct discourse. I am not (and our disagreement here doubtless leads back to Baker’s willingness to attribute basic ‘I’-thoughts to animals and infants). There are (at least) two ways in which the ‘I’ in ‘I am F’ might be eliminated: ‘I’ might be replaced by (e.g.,) a name OR ‘I’ might be replaced by a quantifier, yielding ‘there is F’ (the Russell-Geach proposal).
The first way strikes me as implausible. If ever there was a reason for believing in referential opacity, it’s in the case of indexicals. I can surely think ‘I am F’ but not think ‘Garrett is F’ (because I’m amnesiac, say). But then, with Frege, we should think that ‘I am F’ (uttered by me) and ‘Garrett is F’ (uttered by anyone) do not have the same content; so the former cannot be eliminated in favour of the latter. The Russell-Geach proposal is open to a different worry. Suppose I am thinking of Paris and you are thinking of Vienna. It would not be enough to re-write these indexical truths as ‘there is thinking of Paris’ and ‘there is thinking of Vienna’, for this would leave out the fact that I was thinking of Paris and you were thinking of Vienna. So the quantified sentences would need to be relativised to persons. So my thought ‘I am thinking of Paris’ would be rendered either ‘there is thinking of Paris (me)’ or ‘there is thinking of Paris (Garrett)’. If the first, we seem to have simply another way of writing ‘I am thinking of Paris’ (so ‘I’ has not been eliminated). If the second, the earlier opacity objection re-surfaces: ‘there is thinking of Paris (Garrett)’ simply does not capture what I think when I think ‘I am thinking of Paris’.
So all ‘I’-sentences are ineliminable, in my view. This makes them indispensable in one sense: viz., if someone is thinking an ‘I’-thought, and we want to give a complete inventory of their mind, we must include that very ‘I’-thought in our inventory. But equally, if someone thinks ‘Tully is F’ and ‘Cicero is G’, we must cite those very thoughts in our inventory (at least for those of us, like myself, who believe in opacity quite generally). So there’s nothing special about ‘I’-thoughts in this regard. Given all this, I obviously agree with Baker about her second way in which first-person reference is indispensable: for psychological explanation.
6. I had a general worry about Baker’s account of self-consciousness, but I’m not sure it’s much of an objection. There is something unsatisfying about it, as if, at the end of it, we still don’t know what self-consciousness really amounts to. She drew some reasonable and familiar distinctions, and laid down some plausible constraints that any account of the concept of self-consciousness must respect, but I didn’t feel that the concept of self-consciousness had been mined particularly deeply. The reason this may not be an objection is that it’s not clear that there is much to be said. It’s not that I think that only a reductive analysis can be satisfying (I don’t). Indeed, it’s pretty clear that the concept of self-consciousness cannot be given any reductive analysis. That is, we cannot analysis ‘self-consciousness’ as ‘consciousness + X’, where X can be characterised independently of the concept of self-consciousness. I take this to be one moral of Anscombe’s example of the ‘A’-users in her thoughtful article ‘The First Person’. (I assume that in the present discussion we’re talking only of the concept of self-consciousness, not the property it’s a concept of. That property presumably is some kind of neural property (in us). It’s only the concept that’s irreducible.)
It’s something to be told that self-consciousness consists in having a first-person perspective which, in turn, consists in being able to entertain ‘I’-thoughts. Some light is cast. But then the question arises: what is it to have ‘I’-thoughts?, and merely distinguishing ‘I’ thoughts from other kinds of thought, thus showing that ‘I’ thoughts are unique and ineliminable, somehow does not provide much of an answer. Still, this is not an objection to Baker if no one can do any better. I think a bit more can be said about the concept of self-consciousness and its link with the ‘as subject’ use of ‘I’ (see Chapters 7 & 8 of my book), but I agree I don’t say much.
7. In Chapter 5, Baker discusses the traditional question of personal identity over time. Her own view is that personal identity over time is not analysable in terms of anything more basic. She motivates this view by considering attempts at such analyses, and showing how they fail. She considers a number of familiar criteria of personal identity — sameness of body, sameness of living organism, sameness of brain, psychological continuity, and sameness of soul — and finds them all wanting.
I agree with many of the points she makes against these criteria. One additional point I would make is that the sameness of living organism criterion really has no advantange over the sameness of body criterion. One objection to the sameness of body criterion is that, according to it and absurdly, I am identical to my corpse. This absurdity is supposed to be avoided if we identify the person with the living organism, rather than the body. But is it? The following question now comes to the fore: is the organism when alive identical to the organism when dead? If ‘yes’, there is no advantage over the body criterion. If ‘no’ (on the grounds that there is no organism after death), then my body now is not identical to my human organism (since my body does exist after death). But that seems an absurd duplication. Either way then, no advantage is gained by moving to the living organism criterion.
Baker objects to the psychological criterion that it cannot handle cases of branching in a satisfactory way. I have argued that it can (see my book, Chapter 4), but will not press the point here. One point worth making is that the criteria of personal identity that Baker criticises do not exhaust the field. She looks at purely physical criteria and at purely psychological criteria, without looking at mixed criteria of the sort advocated by myself and (I think) Parfit. The view I endorse, for example, says that A at t1 is the same person as B at t2 iff A and B stand to each other in the relation of non-branching psychological continuity, where the cause of the continuity is either normal or physically continuous with the normal cause. Baker will not like this way of avoiding the duplication objection, but it’s at least worth considering mixed views in general.
8. Another point. Baker seems to think (see especially p. 131) that the various criteria she criticises are intended as reductive accounts of personal identity. I don’t see them that way, at least not in the first instance. In my criterion above, for example, ‘A’ and ‘B’ occur in the RHS, and these names are introduced via the sortal person. There is nothing reductive about the criterion as stated. Of course, if Parfit is right, it may be possible to eliminate reference to persons in the RHS, but that is a further, and controversial, claim.
Are non-reductive criteria not open to the charge of circularity? Yes, but this is not necessarily an objection. There are circles, and there are circles. A circular analysis can still be illuminating, provided it makes vivid the connections between the target concept and some other range of concepts. For example, it’s circular and useless to be told, as an analysis of the concept red: x is red iff x is red. But it is circular and illuminating (if true) to be told: x is red iff x looks red to normal human observers in normal lighting conditions. This is circular because ‘red’ appears in the RHS, but illuminating because it links red with other concepts (normal human observer, normal lighting conditions). (Wiggins is good on the general point. See Sameness and Substance, pp. 49 – 55.) Anyway, Baker’s objection to the criteria she criticises is not that they are circular, but subject to more straightforward faults.
9. Now Baker concludes from her critique of standard views that the concept of personal identity is unanalysable or primitive. One cannot “give informative sufficient conditions for sameness of person over time without presupposing sameness of person.” (p. 119) However, and following on from her discussion of the first-person perspective, Baker does feel she can say: P1 at t1 is the same person as P2 at t2 iff P1 and P2 have the same first-person perspective, though she concedes that this condition is not terribly informative.
However, it soon becomes clear that Baker’s view of personal identity is very radical indeed. (She insists on calling it the Constitution View. Fair enough, it’s a free country. But nothing she says in this chapter (Chapter 5) draws on the metaphysical account of the constitution relation she outlines elsewhere. So it just seems misleading to me.) She seems to think there are no a priori constraints on personal identity over time (i.e., on identity of first-person perspective). In a case of fission, for example, Baker thinks that I could be identical to one of the off-shoots, or to the other, or to neither (even though all other facts remain the same in these three possible scenarios).
She also says (p. 137) that if either off-shoot had my first-person perspective, I would know it. But how? Both off-shoots would think they were me, and neither would have information the other lacked. How could one know and the other not? Both would believe they were me, and one would be wrong. (I found a number of Baker’s comments on pp. 136 – 8 puzzling, appearing to confuse the synchronic triviality that for any person P at t1, P knows at t1 that he is himself, with some non-trivial diachronic identity claim. There is transparency in the synchronic case, but surely not in the diachronic case. See below.)
10. Now Baker’s no-constraints view is very radical (though she has an ally in Colin McGinn, The Character of Mind, Chapter 6), and does not follow from her rejection of the standard criteria of personal identity. Let’s take the second point first. Consider an analogy with knowledge. Let’s agree that there is no neat set of necessary and sufficient conditions for ‘S knows that P’. All the standard accounts are open to counterexamples. Does it follow that there is no sufficient condition (however weak) for ‘S knows that P’ specifiable without using the concept of knowledge? Well, no, or not obviously. Isn’t it plausible that (necessarily) if S has a justified true belief, reliably caused in a certain way, counterfactually sensitive to P’s truth-value in other worlds (add in your favourite condition …) then S knows that P? That seems plausible to me; or at least, its falsity does not follow simply from the concession that there is no set of necessary and sufficient conditions for ‘S knows that P’. And I would make exactly the same point against Baker: from the failure of attempts to give informative conditions necessary and sufficient for ‘A at t1 is the same person as B at t2’ it doesn’t follow that there is no set of sufficient conditions.
And now the first point. Baker’s view (like McGinn’s) is so radical as to be unacceptable. To start with it just seems pretty obvious that there is a (weak) sufficient condition for my identity over time: if my brain, body, and psychological stream continue as normal, and there’s no duplication, fission, fusion, teletransportation, etc., then I will occupy this body tomorrow. But Baker has to deny this, otherwise she would be embracing an informative sufficient condition for personal identity over time. She must think that a different first-person perspective (different person) could occupy my body tomorrow, even though everything else just continues as normal. This strikes me as bizarre.
The bizarreness shows up epistemically. On Baker’s view, in a case of fission, I could be B or C or neither. Suppose I’m B. Since, ex hypothesi, everything is the same from the perspectives of B and C, neither would know that they were me. Moreover, the fact that I am B would be completely unknowable (by anyone): no one could ever know it. (B would know that he is himself, but that is a triviality.) You might say: so what, maybe some truths are unknowable? (See p. 134.) But the trouble is that the scepticism generalises. Even in a normal everyday case, how is a person supposed to know at t2 that he existed in the same body at t1? He can know that if he knows that the same first-person perspective persisted in the same body from t1 to t2. But how is he supposed to know that, given that he has no priviliged access to identity of first-person perspective over time, and given that identity of first-person perspective is not fixed by the facts he does have access to (sameness of body/brain and psychological continuity)? This is an objection to Baker’s view since we do not feel any sceptical qualms when we regularly re-identify our friends and acquaintances. (Remember this discussion is not taking place in the context of Cartesian scepticism.)
To re-iterate an earlier point: Baker makes it seem that there is no epistemological problem here, but only, I think, by projecting features of synchronic self-identification (criterionless, no identification of a subject, etc) onto diachronic self-identifications (which do not have these features). For example, she writes: “The following seem to me to be incontrovertible facts, easily discernible from a first-person perspective: Every morning when I wake up, I know that I am still existing – without consulting my mirror, my memory, or anything else. I can tell.” (p. 136). But, given Baker’s no-constraints view of the first-person perspective, how can she tell? Remember we’re not concerned with the synchronic, indexical triviality that any person can say the sentence “I am me” and thereby speak the truth (cf. ‘I am here’ and ‘the time is now’). So Baker must be saying that she can tell now (in the morning) that the same first-person perspective has persisted through the night. But how can she tell, on her view? On that view, all the psychological facts (memory, etc.) and physical facts (identity of brain/body) do not fix identity of first-person perspective. Nothing fixes the latter, for Baker, so how can it’s persistence be known (immediately or otherwise)?
11. A different point. Baker thinks it’s an advantage of her view that it does not allow for indeterminacy in personal identity over time. But it’s unclear to me both why this would be an advantage and why her view doesn’t allow for indeterminacy. Take the second issue first. What one would expect is that the necessity of such determinacy would flow from the nature of the first-person perspective. But, as noted earlier, Baker doesn’t really say much about the first-person perspective, so the requirement of determinacy just looks like a stipulation on her part. If we think of a case in the middle of Parfit’s Combined Spectrum, where I have half my cells removed and replaced, and have my many of my psychological states altered radically, it’s far from obvious to me that it would be wrong to say: it’s vague or indeterminate whether one and the same first-person perspective has survived. Why would this be an absurd thing to say?
As to the first issue, I don’t see why it’s a defect in a criterion of personal identity that it allows for indeterminate cases. A belief, quite generally, in the possibility of indeterminate cases arises from the most benign of motives. One simply judges that certain small changes would not destroy the identity of an F, whereas other, bigger, changes would, and so concludes that in cases in the middle range it is indeterminate whether the same F persists. (The only alternative description is that there is a sharp cut-off point. But, like many, I find the idea that a small change could make the difference between identity and non-identity implausible.) For example: I know that changing one plank will not destroy a ship, but that changing all but one plank will (I’ll then have dismantled one ship, and rebuilt another). So in cases in the middle, the sentence ‘the earlier ship is the later ship’ is indeterminate in truth-value. (It’s controversial what the source of this indeterminacy is – language or the world – but that’s a different issue.)
I think it’s unfair of Baker to characterise the indeterminacy theorist as saying: the resulting ship (person) is partly one ship (person), partly another. A belief in indeterminacy does not involve that kind of incoherence. (Even in the case of colours it would be wrong to describe grey as ‘partly white, partly black’. Grey is a new colour which results from combining white and black. Something is ‘partly white, partly black’ iff it has a white part and a black part.) So I don’t see that it’s a point against a theory of personal identity that it allows for indeterminate cases
12. Although it may not be worth mentioning, I really didn’t follow pp. 138 – 40, and couldn’t see what the point of condition (T) was supposed to be.
Perhaps I can say a bit more. Baker is offering us a sufficient condition for sameness of human person over time. But since Baker (and me) think that ‘person’ and ‘human’ are sortals associated with different criteria of identity, the phrase ‘same human person’ sounds a bit odd. The only way I can understand ‘x is the same human person as y’ is with one or other of the sortals as dominant. That is, either ‘x is the same human person as y’ is true iff x is the same person as y and both are human or ‘x is the same human person as y’ is true iff x is the same human being as y and both are persons. Now we’ve been assuming that the truth-condition of ‘x is same human as y’ is not problematic, and Baker has been telling us what she thinks can be said about the truth-condition of ‘x is the same person as y’, so what new issue is being raised in these pages?
13. On p. 214, Baker says that “although human persons are not essentially human (they may have [i.e., survive in] inorganic bodies), anything that begins existence as a human person is essentially embodied.” But (on her view) why? We are referred to Chapter 4, presumably principle (T6) of that chapter, but I couldn’t see what the argument for this principle was supposed to be. The reason I’m mentioning this is that I would have thought that anyone with Baker’s views would reject, or at least not endorse, (T6). According to Baker, what makes for personal identity over time is the continuation of the (primitive, unanalysable) first-person perspective. This can continue in non-human bodies, or silicon bodies, or what have you. Fine. But if immaterial souls are a logical possibility, why should it not be logically possible for my first-person perspective to continue in an immaterial soul? I can’t see why Baker would deny this possibility, given her views. Of course, if she thinks that immaterial substances in general are logically impossible (I don’t get the impression she does), that’s obviously a reason to deny that my first-person perspective could continue in an immaterial soul. Even then, though, this denial wouldn’t be a consequence of her theory of personal identity, but of more general views about immaterial substances. Anyway, if immaterial substances are a logical possibility (as even many materialists think), then Baker should not be endorsing (T6).
14. Finally, can I just observe that I find Baker’s conditions for failing “to take persons seriously” (p. 218 – 222) somewhat idiosyncratic. To take persons seriously, I would have thought, is to agree that persons are valuable beings, whose lives are typically to be valued much more than those of other animals or material possessions. Someone who believes that we are not essentially persons can still take persons seriously, in this sense of the phrase. And if Baker also wants to understand “taking persons seriously” as “regarding persons as irreducible, ineliminable elements of the universe”, then Reductionists, like Parfit, should just turn round and denounce that idiosyncratic sense too. out, that would hardly be surprising.