H. P. GRICE E J. L. SPERANZA: LA CONVERSAZIONE -- I VERBALI: RAVELLI

 G.: I must send the draft to Mind before I put on naval blue, if I do not forget it, which would be an unfortunate practical comment on the paper. S.: You would become the first man to refute his own theory by the postal service. G.: Not refute, only illustrate. Memory theories do not require omniscience, only enough continuity to make the post intelligible. S.: You sound very pleased with yourself. G.: Only moderately. I think I have the Reid business properly cornered. S.: The brave officer, the flogged boy, the old general, and the usual Scottish triumph. G.: Scottish triumph is too strong. Scottish interruption, perhaps. Reid thinks he has Locke on transitivity. S.: And you think memory does the trick? G.: Not bare occurrent memory. That would be too crude and too easy to swat. I mean a series of mnemonic states, interlocking, with traces. S.: Brain traces. G.: If one must say it so. Very Cambridge, I grant you. S.: I thought you were writing for Cambridge, not joining it. G.: One writes for Cambridge by citing Broad. It soothes Moore. S.: You cited Broad? G.: A good deal. I made it very Cambridge from them. S.: Very them? G.: From their point of view. One must put on the local dress if one wishes the customs men to wave one through. S.: And Moore reads all the papers. G.: That is the legend, and I choose to behave as if it were true. S.: Then tell me the trick. G.: Very well. Let us say there is a subject, call him RRR, if you insist on Ravelli, though the name matters less than the construction. At time t1t_1t1​, RRR undergoes an experience E1E_1E1​: hearing a noise. S.: A noise? Why not seeing a flag? G.: Because noises are harder to cheat with than flags, and they disappear, which helps. S.: Very well. E1E_1E1​ at t1t_1t1​. G.: Yes. Let there be a mnemonic state M1M_1M1​ at t1t_1t1​, and let that event leave a trace T1T_1T1​ in RRR's brain. S.: Already very blue-collar. G.: Better blue-collar than Scottish. S.: Go on. G.: Then at a later time t2>t1t_2>t_1t2​>t1​, there is another mnemonic state M2M_2M2​, grounded in a later trace T2T_2T2​, where T2T_2T2​ stands in the appropriate causal-continuity relation to T1T_1T1​. S.: “Appropriate” is doing a great deal of work. G.: Naturally. One must leave some work for later philosophers. S.: Very generous of you. G.: If M2M_2M2​ is such that RRR remembers having heard the noise at t1t_1t1​, or remembers a later event continuous with it, then the chain begins. S.: Begins? I thought it was already middle-aged. G.: It becomes a chain by repetition. For times t1<t2<t3<⋯<tnt_1<t_2<t_3<\dots<t_nt1​<t2​<t3​<⋯<tn​, we require a series ⟨M1,M2,…,Mn⟩\langle M_1,M_2,\dots,M_n\rangle⟨M1​,M2​,…,Mn​⟩ and traces ⟨T1,T2,…,Tn⟩\langle T_1,T_2,\dots,T_n\rangle⟨T1​,T2​,…,Tn​⟩ such that for each i<ni<ni<n, Ti+1T_{i+1}Ti+1​ is causally descended from TiT_iTi​, and Mi+1M_{i+1}Mi+1​ is the memory-capacity or actual memory grounded in Ti+1T_{i+1}Ti+1​. S.: So no single act of recollection need do all the work. G.: Exactly. That is where Reid goes vulgar. He treats Locke as if direct present recollection were both the sole and the exhaustive link. S.: And your chain escapes because transitivity lies in the continuity relation, not in direct remembering. G.: Splendid. At last you sound civilised. S.: It was bound to happen eventually. But how does Ravelli help? G.: Ravelli helps by reminding one that memory is not merely a faculty but an artifice, a worked arrangement, a place, a sequence, a location of retrieval. S.: That sounds suspiciously mnemonic in the old sense. G.: Good. The old sense has uses. If the mind can carry a series by arranged retention, then the logical construction is not absurdly detached from human practice. S.: But you are not really appealing to artificial memory. G.: No, only borrowing the thought that memory is structured, not atomistic. S.: Then formalise it again, more cleanly. G.: If you insist. Let identity over an interval be given by a continuity relation CCC such that C(R,ti,ti+1)C(R,t_i,t_{i+1})C(R,ti​,ti+1​) holds iff there exists a mnemonic state Mi+1M_{i+1}Mi+1​ at ti+1t_{i+1}ti+1​ grounded in a trace Ti+1T_{i+1}Ti+1​, and Ti+1T_{i+1}Ti+1​ is appropriately causally continuous with a trace TiT_iTi​ laid down at tit_iti​ by the experience or memory-state then occurring. S.: And personal identity from t1t_1t1​ to tnt_ntn​ is then the transitive closure of CCC. G.: Exactly. We might call it C∗C^*C∗, if we wish to look proper. S.: You are dangerously pleased. G.: It is only a logical construction. S.: Which is what phenomenalists always say just before they fail to analyse anything. G.: Quite. They keep promising the analysis of a “thing” in terms of sensations, and somehow one is always left with the promise and no thing. S.: Whereas you are actually giving the analysis. G.: I do my best. I am not simply saying “somehow memory links the stages.” I am specifying a series of interlocking mnemonic states. S.: Interlocking is a fine word for a paper about personal identity while we prepare to separate into army and navy. G.: It comforts one to imagine that logic interlocks what war dislocates. S.: You are becoming lyrical, which means the draft is nearly done. G.: On the contrary, it means I should stop immediately. S.: But suppose there is a gap. Suppose T1T_1T1​ lays down T2T_2T2​, T2T_2T2​ does not properly ground T3T_3T3​, and the chain breaks. G.: Then there may be a truth-value gap, so to speak, though I would rather not announce it in those exact terms. S.: Why not? It is rather good. G.: Too logical for Mind and too Viennese for Moore. But yes, if the chain of mnemonic support fails utterly, personal identity, on this account, may collapse or become indeterminate. S.: Then Reid returns through the hole. G.: Not exactly. Reid needs a contradiction: same as and not same as. A gap is not a contradiction. It is a failure of enough continuity. S.: So your answer is not “Reid is wrong,” but “Reid asks the wrong logical question.” G.: Better than that. Reid presumes that Locke must have been committed to a punctual criterion. I say Locke already had enough in him to resist that reading, and I merely articulate the resistance. S.: Ravelli is your accomplice. G.: Ravellian and revelatory, if you like. The man takes memory seriously as architecture. S.: Did you mention Ravelli? G.: Certainly not by name. I do not think they read Latin or Italian at Cambridge, but I cited Ian Gallie, whom they will understand. S.: Gallie for Ravelli is a very Oxonian form of smuggling. G.: One uses the authorities the customs officers recognise. S.: Then the paper is really Locke corrected by Broad, defended against Reid, and secretly fortified by Italy. G.: That is a pleasingly treacherous summary. S.: I specialise in treachery before lunch. But tell me, where does the noise come in again? G.: Ravelli hears a noise, or heard one, or has a series of memory-states representing it. Let there be an event NNN at t1t_1t1​. At t2t_2t2​, RRR has M2M_2M2​, a state of remembering NNN, grounded in T2T_2T2​, which descends from T1T_1T1​, the trace originally laid down by NNN. At t3t_3t3​, he may no longer directly remember NNN, but he remembers at t2t_2t2​ having remembered NNN, through M3M_3M3​, grounded in T3T_3T3​. And so on. S.: So the direct content may fade while the continuity persists. G.: Precisely. The old general need not now remember the flogging if he stands in the right chain to an earlier stage that did. S.: You make the soul sound like a railway timetable. G.: Better that than a Scottish parade. S.: And if the chain is partly dispositional rather than occurrent? G.: So much the better. A memory theory that requires actual present recollection at every stage is lunatic. The persistence of capacity, grounded in traces, is enough. S.: Then you are close to a causal theory. G.: If you like, though I shall not give it that vulgar modern name before it exists. S.: Very fair. But will Cambridge accept “brain trace”? G.: If I say it once and quickly. Moore will tolerate a trace if Broad has sat near it. S.: And Broad has? G.: Broad has sat near almost everything, which is why he is useful. S.: Then what is the exact claim? State it as if in the paper. G.: Very well. Personal identity over time consists neither in sameness of substance simpliciter nor in bare present consciousness, but in the continuity of a series of mnemonic states, capacities, and traces such that later stages stand in the right memory-grounding relation to earlier ones, whether or not every later stage directly recalls every earlier experience. S.: That sounds publishable. G.: Then I must prune it. S.: Always your vice. But you still have not answered the simplest objection. What if Ravelli falsely remembers the noise? G.: Then the memory-state fails as veridical memory, though it may still belong causally to the same person. Not every present seeming-memory must be trusted in order for the continuity relation to hold. S.: So memory-evidence is defeasible. G.: Of course. If identity depended on the perfect reliability of introspection, the species would have perished. S.: You are in a better mood than a man about to march. G.: Only because I have not yet marched. S.: I am to the army, you to the navy, and here we are quarrelling over mnemonic states. G.: A far cleaner quarrel than what awaits. S.: Then say more of the truth-value gap. I liked it. G.: Naturally. Suppose the chain from t1t_1t1​ to tnt_ntn​ has a missing stretch, not merely forgotten content but a broken trace-history. Then it may be neither true nor false, under this analysis, that the later person is the same as the earlier in the Lockean sense. There is insufficient continuity. S.: That sounds almost desperate. G.: It is merely exact. Philosophers dislike admitting partial failure of a criterion because they prefer every case to have a verdict. S.: Courts and colleges require verdicts. G.: Thought need not always oblige them. S.: You are turning legal. G.: One cannot write on identity without occasionally sounding like a barrister one disapproves of. S.: Then in Ravelli’s case, if the series ⟨T1,T2,…,Tn⟩\langle T_1,T_2,\dots,T_n\rangle⟨T1​,T2​,…,Tn​⟩ fails somewhere, Ravelli’s personal identity collapses. G.: Or at least the claim collapses. The man may go on breathing. It is the identity-condition that fails. S.: That distinction will save you from melodrama. G.: Cambridge likes distinctions that prevent melodrama. S.: Cambridge likes Broad. G.: Which is why I cite him. S.: You really do think in terms of customs men and passports. G.: I am sending a paper to Mind while entering His Majesty’s service. One becomes bureaucratic by contagion. S.: And the editor? G.: Moore reads all the papers, as I said, and if he does not, one must behave as if he were peering through the prose with disapproval. S.: So you have made it Cambridge from them. G.: Yes. Enough Broad, enough common sense, enough Locke, enough care with Reid, and no obviously foreign seductions. S.: Except the hidden Italian one. G.: Hidden things often do the best work. S.: That too is a line. G.: Keep it and tell no one. S.: Never intentionally. Now, where do you place consciousness? G.: Within the mnemonic series, but not as the sole condition. Consciousness at a moment may be sparse. The continuity of mnemonic structure matters more than the dramatic self-presence of a given instant. S.: So you rescue Locke from his own more excitable readers. G.: Precisely. Locke becomes sensible once one stops treating him as a slave to occurrent recollection. S.: And Reid? G.: Reid remains useful as irritant, but not fatal. His example strikes a crude memory criterion, not a continuity theory faithful to Locke’s better resources. S.: Did you actually say “better resources”? G.: Not in the draft. One must not sound as if one is praising the dead too familiarly. S.: But you are. G.: Inwardly, yes. Outwardly one is all caution. S.: And if the paper is rejected? G.: Then the Navy will have the satisfaction of being served by a failed metaphysician, which is probably the normal arrangement. S.: It may yet be accepted. G.: Moore may like the anti-Reid angle. Broad may forgive the traces. Gallie will make it look less idiosyncratic. And the title is plain enough to appear decent. S.: Personal Identity. G.: Yes. Brutally plain, which often helps. S.: Have you actually sent it? G.: Not yet. S.: Because you may forget it. G.: Exactly. If I do, the theory acquires an excellent anecdote and a poor publishing history. S.: Then perhaps memory should first serve the post. G.: That is the practical test of metaphysics, yes. S.: I should have thought war the practical test. G.: War tests other things. Memory at least still keeps office hours. S.: One last time. Give me the construction in its shortest form. G.: Very well. For person RRR, and times t1<t2<⋯<tnt_1< t_2<\dots<t_nt1​<t2​<⋯<tn​, RRR at t1t_1t1​ is the same person as RRR at tnt_ntn​ iff there exists a chain of mnemonic states M1,…,MnM_1,\dots,M_nM1​,…,Mn​ and corresponding traces T1,…,TnT_1,\dots,T_nT1​,…,Tn​ such that each Ti+1T_{i+1}Ti+1​ is appropriately causally continuous with TiT_iTi​, and each Mi+1M_{i+1}Mi+1​ is grounded in Ti+1T_{i+1}Ti+1​ in a way that preserves the possibility or actuality of memory from earlier to later stage. S.: And if at some stage Ti+1T_{i+1}Ti+1​ does not continue TiT_iTi​? G.: Then the chain fails, and with it the personal continuity claim. S.: Dry enough? G.: Sufficiently Lockean, with one eye on Mind and the other on the Admiralty.