Psychological Criteria of Personal Identity I Memory
The reason why the brain was considered as pertinent to personal identity in the first place is that we (correctly, in all probability) assume that it embodies the psychic functions we are looking for: cognitive abilities and character traits. Accordingly, our investigation so far shows that it is rather these abilities and traits themselves we have to consider, not their contingent physical implementation.
We therefore continue by addressing ourselves to psychological criteria of personal identity. Again, Locke was the first to embrace this approach by suggesting "sameness of consciousness" as the criterion for personal identity over time (see e.g. Locke, 1975:II, 27, §23). Although consciousness encompasses much more, episodic memory is certainly the key to understanding this criterion insofar as consciousness according to Locke involves knowledge of one's thoughts and actions. "Sameness of consciousness" is constituted by the knowledge of thoughts and actions shared by earlier and later stages of a person via episodic memory. Therefore, even though Locke does not explicitly use the term in presenting his position, he is correctly regarded as trying to give a criterion for personal identity over time in terms of episodic memory. The difficulty is, however, to state the criterion in such a way that certain more or less tricky problems can be avoided. In order to gain insight into the problems to be avoided by an adequate memory-based criterion of personal identity, it is helpful to mention and discuss briefly the main strands of critique with which Locke's approach was historically confronted.
First, there is the so-called "circularity objection". It goes back to bishop Joseph Butler and can be stated like this92: A person Pi's memory of a past experience cannot be used as a criterion for Pi being the person P2 who originally had the experience in question because to establish that Pi really remembers having had the experience, it would already have to be verified independently that P1=P2. The point is that, according to common usage,
91 Or at least not according to any non-trivialising criterion for "identity of the brain".
92 In the original writings of Butler, the objection itself is not presented in a very clear way. A contemporary reconstruction can be found in Shoemaker (1970).
"to remember" is a "success-verb", whose application is correct only if the remembered event really took place. This in turn engenders that, if what is remembered is an experience, (i.e. an episode from the first-person perspective), that experience must have been had by oneself.
It is quite obvious that the problem encountered here is a mere technicality. To avoid it we need a concept which in all save one aspect behaves like the concept of (episodic) memory: For its correct application it must not presuppose that the episode to be remembered actually happened to the person remembering it. Sidney Shoemaker (1970) coined the term "quasi-memory" for this concept. So, if we could "transplant" episodic memories (or if, in Lamarckian fashion, some episodic memories of parents could be inherited by their offspring), then the persons receiving the memories would quasi-memorise these episodes.93
A second problem with memory-based criteria for personal identity was pointed out by Thomas Reid (1983 [1785]). It draws on a strict reading of Locke's "sameness of consciousness" criterion, where "sameness" as constituted through knowledge shared by earlier and later stages is taken to imply that for various person-stages to belong to the same person the later stages have to faithfully retain the experiential contents of the earlier ones. If succeeding person stages x, y, z are such that x and y share some experiential content E, but z and x do not share that content as well, then - according to Locke's criterion in its strict interpretation - they do not constitute the same consciousness.
With his now famous "paradox of the 'brave officer'" Reid illustrates that this has quite unwelcome consequences94: Imagine a brave officer, who had been flogged for some mischief when he was still a boy, has taken a standard from the enemy during his first campaign as a young officer, and was eventually, at advanced age, promoted to rank of general. Further assume that he still could remember the flogging by the time he took the standard, and that he could remember taking the standard when he was made general, but by then had lost all memory of the flogging. According to Reid, it follows from Locke's account of personal identity that the boy and the young officer are the same person, as are the young officer and the general, whereas the boy and the general are not - and this would obviously be at odds with the transitivity of the relation of identity.
Reid's argument has to be disentangled a bit as his presentation is marred by the fact that he confuses continuants with stages. Only because of this the conflict with the transitivity of identity arises. We saw in Section 5.2.4 that different stages cannot be identical anyway, so that questions regarding the identity of continuants encountered at different times have to be put this
93 In fact, if wholesale Lamarckian inheritance of parental episodic memory would be the standard, our understanding of the life of persons would probably be very different (for the result would be a very complex network of psychological fusions and fissions pervading different organisms).
way: "Are O1 at t1 and 02 at t2 both stages of the same continuant?" Hence, what Reid should have argued is that according to Locke's theory (in its strict interpretation) the boy and the young officer are both stages of some person P,and the young officer and the general are both stages of a person P'. But as - again according to Locke's criterion for sameness of consciousness - P and P' do not share all their stages (the boy-stage belongs to person P,but not to P'), it follows that P^P'. And even though this is not logically inconsistent, Reid has indeed a point in claiming that this result shows Locke's criterion of personal identity to be inadequate. For, of course, all three stages - boy, young officer and general - should belong to one person.
What this means is that an adequate (quasi-)memory based criterion for personal identity over time has to take into account that persons may forget. This is achieved by defining a transitive relation of (quasi-)memory continuity.95 As this relation is logically complex, the definition has to be broken down into several steps96:
(R*-Def.) Let R*xy stand in for the following relation: y can quasi-memo-rise at least one episode of experience of x from the x-perspec-tive (i.e. the first person perspective).
The range of x and y are person-stages.97 As mentioned above, as is the case for memory proper, quasi-memory has to be veridical in the sense that the episodes recalled by person-stage y really happened to person-stage x, but unlike memory proper, it does not presuppose that x and y are stages of the same person. It is an indispensable proviso that the episodes are recallable from the first person-perspective of x. Else, if I can recall episodes I saw happening to my father, and my father at that time was able to recall episodes he saw happening to his father, the intended definition of "memory continuity" building on R* would yield continuity between the respective stages of my grandfather, my father and me. And as "memory continuity" is supposed to provide a criterion for personal identity, this would eventually lead to the absurd result that there is a person P of which these three stages are temporal parts. Note that for the relation R* to hold between x and y, y does not actually have to recall an episode happening to x, but just has to be able to do
95 Using Locke's terminology, we could say that "sameness of consciousness" is constituted by continuity of knowledge about one's past experiences rather than by actually shared knowledge.
96 Usually this is done by employing the apparatus of formal logic to make fully transparent the pertinent structural and inferential properties of the concepts to be defined. However, for the benefit of readers who don't have a background in formal logic, we just introduce a few symbols standing in for the concepts we are talking about (for the purpose of convenient cross-referencing) and otherwise give the account in plain English.
97 As such they have to be embodied. However, having abandoned physical persistence criteria, we do not postulate here and in the following that these stages must share the same body or brain.
so.98 Note also that, for convenience's sake, we did not attach time-indices to the stages x and y in the above definition. But it is implicit in the definition of R* that x and y can only stand in the relation R*xy if it is true for times t(x) and t(y) that t(x)<t(y).99 For if t(x)>t(y), then we would not speak of "memory", but rather of something like "precognition". The case t(x)=t(y) is ruled out by the following considerations: Either xt=yt or xf£yt. If xf£yt, then these person-stages count as embodiments of different persons anyway (and y's capacity to "recall" something experienced by x would be considered as some kind of "clairvoyance"). If xt=y, then the person-stage in question cannot, strictly speaking, experience something and at the same time recall it already.
R* is irreflexive and asymmetrical.100 Above all, it is not transitive. We now (recursively) define a transitive relation R using R* as follows:
98 This too must not be read in too strict a sense. For example, while someone is asleep at night he may be considered to be "able" to memorise episodes of the day in the sense that if we wake him up and ask him, he will tell us about it. (Generally, this dispositional reading should go for all capacities deemed constitutive for person-hood - cf. Quante 2002:20.)
99 It is important to note that the times thave to be taken to be (conveniently small) time spans, not points in time. The reason for this is that stages have to be spatiotemporal parts of continuants. If one successively divides a spatial object into ever smaller parts, the parts will still remain extended. The same has to hold for the division of a continuant into temporal parts. For convenience's sake in what follows we just consider a division of time into "small" stretches of like duration leading to what could be called "standard stages" as smallest units. But of course any temporal part of a person may count as a "stage". This does not at all affect the validity of the arguments in the main text, as long as we disregard stage overlaps or stage inclusions - where the terminology of "later"/ "earlier"/"simultaneous" becomes muddled. The ensuing problems could be solved by providing a mereological theory, but more easily they can be avoided by observing the convention that in theoretical contexts the breakdown of continuants into stages should always be carried out such that the resultant stages do not overlap (that is, instead of two overlapping stages we would consider three continuous stages, etc.).
100 For readers not familiar with the terminology regarding structural properties of relations, we explain at least those types of relations used by us in the text: A two-place relation Rxy is reflexive if on the range of objects that can stand in the relation R every object must bear the relation R to itself. For example, every object that at all has a length has the same length as itself. R is non-reflexive if there could possibly be objects that do not bear R to itself. E.g. perhaps not everyone loves himself. R is irreflexive if no object can stand in the relation R to itself. For example no object can be longer than itself. R is symmetric in case that if x stands in the relation R to y, then so will y to x. For example if Fritz is of equal height as Hans, then Hans got to be of equal height as Fritz. R is non-symmetric if there can be pairs of objects x,y with Rxy, but not Ryx. For example, Fritz may like Hans, and yet Hans may not like Fritz. R is asymmetric in case that if Rxy, then it can never hold the other way round, too. For example, if x is taller than y, then y can't be taller than x. Finally, R is transitive just in case that if Rxy and Ryz, then we must also have Rxz. For example if Bill is taller than Jacques, and Jacques is taller than Fritz, then Bill got to be taller than Fritz. R is non-transitive if that doesn't have to be the case. For example, if Anna is befriended to Lisa, and Lisa befriended to Wendy, then Anna may or may not be befriended to Wendy. R is intransitive if this can't be the case. For example, if x is father of y and y is father of z, then x can't be father of z. The way we explained these structural properties, they have to be understood as determined by the meaning of the very concepts R in question.
(R-Def.) Let Rxy stand in for the following relation: The person-stage y is indirectly linked via (quasi-)memory to the person-stage x. R is to hold between x and y if and only if there is a finite chain of person stages beginning with x and ending with y where each link is connected to the next through the relation of quasi-memorisability R*.
We can now define the relation of (quasi-)memory continuity C:
(C-Def.) Let Cxy stand in for the following relation: Two person-stages x and y stand in the relation of (quasi-)memory continuity. C is to hold between x and y if and only if y is indirectly linked through (quasi-)memory to x (i.e. Rxy), or x is indirectly linked through (quasi-)memory to y (i.e. Ryx), or x and y are the same stage (i.e. x=y).
C is a so-called "similarity relation", characterised by being reflexive and symmetric, yet not transitive. Using the relation C of (quasi-)memory continuity, we can now finally proceed to define what it means (within the framework of theories drawing on psychological continuity criteria) for a conglomerate of person-stages to constitute a person:
(P-Def.) Let Px stand in for the predicate "x is a person". P is to hold with respect to a conglomerate p of person-stages if and only if
1) all person-stages belonging to p as temporal parts are continuous (C) to one another and
2) p is maximal in the sense that if some person-stage is continuous to all stages belonging to p as temporal parts, then it is itself a temporal part of p.
From this it follows as a necessary criterion of personal identity over time that if two person-stages encountered at different times both belong to some (the same) person, then they must be continuous; and more generally:
(MemCrit1) If the members of a set of person-stages M all belong as temporal parts to some (the same) person p, then they must all be continuous to one another.
Regarding the relation between personal identity and persistence, based on the definition (P-Def.) given above, the question of whether a certain person p persists at a certain time t is rendered trivial: it boils down to the question of whether there is a person stage at t belonging to p.101
101 In line with what we said in Section 2, this definition does not rule out the possibility of "reincarnation", i.e. that a person persists during a certain time span, then ceases to persist, then persists again later. To exclude this possibility, one would have to incorporate into the definition that for any two stages x and y belonging to p with t(x)<t(y), t(y) is either the direct successor of t(x) (with respect to our division of time into standard durations) or there is a stage zalso belonging to p and t(x)<t(z)<t(y).
The memory criterion (MemCrit1) avoids Reid's paradox: the stages of the boy, the officer and the general are not ruled out as belonging to the same person as they indeed are all continuous to one another. Together with certain assumptions102, (P-Def.) also yields directly a sufficient criterion for personal identity over time in the sense of:
(MemCrit2) If the members of a set of person-stages M are all continuous to one another, then there is a person p which they all belong to as temporal parts.
Note that a special case of this is that every person stage belongs to at least one person.
So much about theory. Let's now see whether memory continuity is really a necessary criterion for personal identity. There are two classic thought experiments - one by Locke and one by Leibniz - which make it extremely plausible that this is indeed the case.
Locke's thought experiment was originally designed to refute the so called "simple view" of personal identity, which says that personal identity is constituted through sameness of soul, where "soul" is understood to be a simple, indivisible and immaterial substance.103 Apart from that, the thought experiment makes a good case for memory-continuity as a necessary criterion for personal identity. Locke (1975:II, 27, §14) invites us to assume that some present-day person is the reincarnation of Nestor at the siege of Troy in the sense that the present-day person shares with Nestor the same "immaterial spirit" (thinking substance), but has no consciousness whatsoever of any of
102 The first assumption we need is trivial, namely that for all sets of person-stages M the corresponding conglomerate p of person-stages exists. This can be secured by the simple decision to regard every possible collection of stages as a "conglomerate". The second assumption is more demanding as it says that all sets of stages where all members are continuous to one another are finite. This amounts to saying that no one persists forever or has persisted ever since. Both cases cannot (or should not) be ruled out on conceptual grounds alone. However, the second case at least can be excluded for obvious empirical reasons, and there is actually no need to worry about the first case, considering that - since we do not know the future - the constitution of continuants is de facto always only done "up until now". While the definition (P-Def.) provides a test regarding the question of whether a given conglomerate p satisfies the definiens, it does not give any hint concerning the question of how to construe such a conglomerate. This can be achieved the following way: Start with one stage a. Then pick another stage b that is continuous to a. Then pick a third stage c continuous to both a and b, and so on, until there is no stage left that is continuous to all stages already picked (the second assumption secures that this result can always be achieved in a finite number of steps). What we get is a conglomerate p satisfying definition (P-Def.). Fortunately, this way to construe persons from the set of stages is not something we do (or should do) in real life. We only use it as an easy way to see that (MemCrit2) indeed follows from (P-Def.).
103 Obviously, this is the orthodox view within Christian theology. It is still very influential today and, in recent times, was stoutly defended by philosopher Richard Swinburne (see e.g. Shoemaker and Swinburne 1984).
Nestor's actions. Would we deem the present-day person the same person as Nestor? No more, Locke says, than if some (or even all) particles of matter that once were part of Nestor accidentally were now part of the present-day person: Sameness of immaterial substance without "sameness of consciousness" will no more make for the same person than sameness of material substance without sameness of consciousness would. And indeed, if mere sameness of material substance will not suffice to constitute personal identity, then it is not intelligible why resorting to immaterial substances should make any difference. What counts in personal identity is never sameness of substance, be it material or immaterial, but "sameness of consciousness" (which we have to conceive as "memory continuity").104
Leibniz's thought experiment - which he devises in "Discourse de Métaphysique", Section xxxiv - is similar to Locke's, but probably even more convincing as it is placed in the context of a choice everyone can imagine to be presented with. Suppose someone could instantly become King of China (which Leibniz obviously assumed to be a most pleasant thing), but on the condition that he would forget everything he had ever been and experienced: as if he had been born again. Leibniz then considers the question whether it would be rational for a person presented with this option to take it - and unhesitatingly answers in the negative. For this would, he argues, be indistinguishable from that person being annihilated and a King of China being created at the same instant in her place. That is, taking the option would amount to committing suicide - not in the sense of ending one's life as a human being, but as the person one had been.105
So far as the assessments of Locke and Leibniz are in line with the intuitions of most people, their thought experiments do show that the criterion of (episodic) memory continuity states a necessary condition for the attribution of personal identity. Still, the criterion also allows for a good deal of "robustness" of personal identity with respect to memory loss: The case of Alzheimer-patients provides evidence that in the course of a degenerative
104 If this is so, it can easily be seen that sameness of immaterial substance cannot even lay claim to being a necessary condition for personal identity. For why shouldn't immaterial substances successively pass through a person during the time of her persistence just as material particles pass through an organism during its life? In both cases, as long as the right "structure" is retained, everything should be just fine. Hence, as the assumption of immaterial substances obviously does not help at all in accounting for personal identity, we encourage the reader to regard this fact as a sufficient refutation of the "simple view". Even though it may be consistent with all empirical data, the assumption of there being immaterial substances should be dropped in accordance with "Occam's razor": "Entia non sunt multiplicanda praeter necessitatem."
105 Like Locke, Leibniz also adopts the classical position that "person" is a moral category marking individuals who can be held responsible for their actions. In this vein he says (in the same section of the "Discours de Métaphysique") that "it is the memory or knowledge of this 'I' which renders it open to punishment and reward".
process persons can persist in spite of considerable loss of episodic memory. In this, it is an important stabilising factor that their character-traits remain suitably stable for their social contacts to "recognise" and anticipate their "typical" actions and behaviour. However, this "robustness" of personal identity does not extend as far as to cover also the final stages of the disease where episodic memory is eventually too attenuated and disintegrated to still provide for suitable interaction - think of the stages when the patients don't recognise their relatives and friends anymore.
Post a comment