Parfit Argued That Personal Identity Is Not What Matters
Derek Parfit’s Reasons and Persons was published by Oxford University Press under the Clarendon Press imprint in 1984. Part Three of the book is titled “Personal Identity” and it contains the sharpest argument in the philosophy of copying. Parfit’s conclusion is that personal identity is not what matters. What matters is a relation he calls R, psychological connectedness and psychological continuity, and that relation can hold to more than one later person at once. Once you accept that, the fear of being copied loses its object.
Two cases and one relation
Parfit builds the argument on two cases that look similar and turn out to divide cleanly.
The teletransporter case scans a person on Earth, destroys the body in the scanning, transmits the information to Mars, and reconstructs a person there. The Mars person wakes with the Earth person’s memories and character. Parfit asks whether the person has travelled or died.
The branch-line case has a scanner that fails to destroy the original. Two people wake, one on Earth and one on Mars, with the same past and separate futures. Parfit asks which one is the person.
His answer is that the cases are not distinguished by identity. In the first the person survives without being identical to anyone, because the relation that matters holds and there is only one later person to hold it to. In the second identity fails without the loss of what matters, because the relation holds to two people and neither is worse off for that. He states the conclusion as a preference. A rational person should not fear the first and should not prefer the second.
| Case | What happens | Parfit’s verdict |
|---|---|---|
| Teletransporter | Scan, destroy, reconstruct on Mars | Survival without identity |
| Branch line | Scan, fail to destroy, reconstruct on Mars | Identity lost, nothing that matters lost |
| Relation R | Psychological connectedness and continuity | Holds to one or to many |
| The fear | Dying in the scanner | Not rationally required |
Why this is the hardest case for a copy
Every other argument about copies runs into the objection that a copy is a different person and the original is dead. Parfit does not answer the objection. He dissolves it. He grants that the copy is a different person and denies that this is the fact that should determine your attitude.
That move is what makes his position difficult to use and valuable to hold. It does not tell you that a copy is you. It tells you that the question “is it me” is not the one that decides the case. A defender of copying who wants to comfort someone facing a scanner needs more than Parfit gives, because he refuses the consolation that they will continue. A critic of copying who wants to prove that the person dies also gets nothing, because he grants that fact and says it does not settle the matter.
Bernal reaches a different conclusion by a different route and the contrast is instructive. In the 1929 case for removing the body he describes a compound mind in which component brains are replaced as they die and the memories transfer to a common stock, and asserts that continuity of the self is preserved. Parfit would say Bernal is asserting the wrong thing. On Parfit’s account the self is exactly what does not need to be preserved.
Comparison to The Consciousness AI
Parfit’s position makes one design question sharper and leaves another untouched. The sharper one is the branching problem. If this project ever produced a second instance of a running system and both instances reported being the same subject, Parfit’s account says there is no further fact about which one is. The design should not be built to resolve that, because there is nothing there to resolve.
The untouched one is whether either instance is conscious at all. Parfit assumes throughout that there are subjects to survive or fail to survive. He brackets the question of what makes a system a subject. The project’s Global Workspace layer and Affective Core are meant to carry functional organisation and that is the same assumption Parfit makes. Neither his book nor the project’s architecture demonstrates that organisation is enough.
This is the honest statement of the gap and it should be kept visible. Personal identity and consciousness are different questions and both need answers.
What the argument does not establish
Parfit does not show that a copy is conscious. He shows that if a copy is conscious and has your psychology, then the fact that it is a different person is not the fact that should worry you. That conditional is real work and it is not a proof of the antecedent.
He also does not treat the source question. In the branch-line case both people survive. In the analysis of the coin toss in SOMA the harder case is a copy created while the source is left in place and left worse off. Parfit’s relation holds to the copy and to the source and he does not ask whether the procedure is good for the source. That is an ethics question and it is a separate one.
One further limit is worth naming. Parfit writes as a reductionist and he says the self is not a deep further fact. Readers who hold otherwise get from him a well stated challenge and not a refutation.
The standing of the book
Reasons and Persons is the standard reference in the analytic literature on personal identity and it is the text that later work on copying argues with. Its Part Four on future generations is separate and is not treated here. Part Three is the part the mind uploading question turns on and it is the reason the book belongs on the shelf next to the engineering literature rather than only in the philosophy stacks.
Reasons and Persons by Derek Parfit was published by Oxford University Press, Clarendon Press, in 1984, and reissued with a new preface in 1987. Part Three, “Personal Identity”, contains the teletransporter case, the branch-line case, and the definition of Relation R as psychological connectedness and psychological continuity. Parfit was a Fellow of All Souls College, Oxford, and lived from 1942 to 2017.