Introduction

Posted on Edited on Disqus:

The Simple Solution to Personal Identity

The question of what defines a person, as a unique entity separate from all other people, has been a topic in philosophy for centuries. In recent years it has cropped up with more urgency, largely due to the growing possibility of technology to “upload” a person’s mind into a digital substrate. This mind uploading has been the topic of both serious research (e.g. Whole Brain Emulation: A Roadmap) and popular media (e.g. Upload). As a result, discussions on the internet have frequently spun around in the same circles, with some observers claiming that an upload is obviously “just a copy” (and not the same person as the original), while others feeling it no less obvious that a perfect copy is the same person as the original. The two groups come to exactly opposite conclusions about whether mind uploading would be a useful technology for survival. (One could make the exact same arguments about the science-fictional transporter, though writers frequently hand-wave away the issue and dodge the hard questions in this case, making it less commonly the inspiration for serious discussion.)

While most participants in such discussion are reacting purely based on their own “common sense” rather than any deep philosophical considerations, the core of the discussion is exactly the philosophical topic of personal identity. Exactly what is it that makes one person the same over time and through change? What is the nature of “self”? How does it relate to your body, your memories, and your personality? These all seem like esoteric, academic questions, until the day comes when you have to choose between dying forever and stepping into the uploading (or transporter) machine — and then they suddenly become literally a matter of life and death.

Fortunately, despite all the pitfalls of common-sense reasoning and all the back-and-forth arguments among philosophers over the centuries, there is a simple solution which provides sensible, practical, and coherent answers to these questions, and all the tricky scenarios we can imagine. It’s based in part on fuzzy logic, which was first formalized in 1965, and so not available to most of the philosophers of yore. It also relies in part on our modern understanding of the brain and how it creates the mind. As a bonus, the solution applies not just to people but also objects, such as the famous “Ship of Theseus,” or George Washington’s axe. The solution can be stated as simply as this:

Two things share the same identity to the extent that their relevant information content is the same.

Of course, while it’s simple to state, it will need some elaboration to be fully understood. The rest of this essay will do exactly that, in brief form; and then further essays will dive deeper into various aspects and implications.

So let’s start by noting (and dispensing with) a possibly troublesome limitation of our words. The solution starts with “two things” — which could be taken to imply that we have already decided that whatever two we are thinking of are two separate, non-identical things, i.e. not the same thing — not sharing the same identity, making the rest of the sentence moot. That is obviously not the intent here. What we really mean is two thing references, i.e. I point to something, and you point to something, and we’re trying to decide if what we’re pointing at is the same thing (even if they are at two different positions in time or space). A more precise statement of the solution might be Two thing references refer to the same thing to the extent that their relevant information content is the same — but that seems linguistically awkward, so we’ll stick with the formulation above. Hopefully we are all now clear on the meaning of the first part.

Next, let’s ensure we are just as clear on what it means to “share the same identity.” This is just a slightly more precise way of saying “are the same thing.” When I say that I am the same person I was yesterday, I mean that me-now and me-yesterday share the same personal identity. If I assert that this axe on my shelf is in fact George Washington’s axe, then I am asserting that this axe and George’s axe share the same identity.

Now we get to a key bit in the next four words: “to the extent that.” This tiny phrase is so crucial! It declares that identity is not a Boolean property. Boolean properties have only two possible values: true and false. For thousands of years, philosophy was based entirely on Boolean logic (even before it was called such); a proposition could be either true or not true, and there were no other alternatives. In the real world, however, Boolean properties are surprisingly hard to find. Day and night? No, there are also twilight and dawn, which smoothly transition between the two. Tall and short? Again no; I may be a little tall or moderately tall or very tall (and same for “short”). The same applies to statements like “it is warm today” or even “it is raining”; it may be a little warm or very warm; it may be just barely raining or a torrential downpour, and anything in between.

While people have always dealt comfortably with the non-Boolean nature of the world in everyday life, it wasn’t until the mid-20th century that logical tools — known as “fuzzy logic” — were developed to deal with it. As a result, we have a long (and misguided) tradition of thinking of identity as a Boolean property, and this has directly resulted in philoosphers, both armchair and professional, getting tied up in logical knots. Virtually any theory of personal identity that assumes that two entities must be either “same” or “not same” quickly breaks down when confronted with easily-imagined edge cases, and we’ll demonstrate exactly that in later essays.

But we live in the 21st century, and need no longer accept such limitations. Identity, like so many other things in the real world, is a “fuzzy” property. Note that fuzzy does not mean ill-defined; it means merely that it admits degrees of truth. The two things you and I are pointing at may be somewhat the same thing, or mostly the same thing, or almost completely the same thing. They might also be entirely the same thing, or not at all the same thing; fuzzy logic is a superset of Boolean logic. Applied to personal identity, we will see that me-now is almost entirely the same person as me-yesterday, but somewhat less the same person as the me of ten years ago. And this refinement alone causes so many arguments to simply melt away.

The last bit of the solution explains how to judge how much two things are the same: by comparing their “relevant information content.” The word “relevant” is relevant here. It admits that “sameness” is relative to some specific purpose we must have in mind. Let’s consider two bowls of paint: are they the same paint? To an artist, the relevant properties are probably color and consistency. If they look the same and behave the same on brush and canvas, then they are the same paint. To a chemist, however, the relevant information content may be their chemical formulas; it could be that they look the same, but will react quite differently to certain reagents.

Some readers may object to this relativism. “Hold on!” I can almost hear them cry, “if the paints differ by any conceivable test, then they are not the same paint, regardless of what the artist thinks.” But this absolutism — particularly if approached in a Boolean way — produces a theory of identity that is simply impractical, i.e., has no practical application. If you get a haircut, are you no longer the same person? And if so, can I have all your stuff, since the person who owned that stuff is no longer around? Moreover, even the haircut is more extreme than it needs to be; our bodies are constantly recycling proteins, lipids, and entire cells; it is never exactly the same from moment to moment. But we don’t care, because those tiny physical details simply aren’t relevant for our purposes.

Let’s take another example: did you ever read the book Harry Potter and the Chamber of Secrets? I did; if you did as well, then I would say we read the same book. But surely your copy of the book was physically different from mine; maybe you read the paperback edition while mine was hardbound. Or I could have used an e-reader; you might even have listened to it as an audiobook. For establishing whether we’ve read the same book, those details don’t matter; what’s relevant is the information content of the book: the plot, the characters, the dialog, the twists and turns of the story.

Now imagine a detective at a crime scene, where the only clue is a tattered copy of Chamber of Secrets found under the victim’s duffel bag. A key witness saw somebody reading Chamber of Secrets at the cafe the day before. Was that the same book? In this case “same” (i.e., the question of the book’s identity) has a completely different meaning. The detective needs to know if the two books are actually the same copy, i.e., are made of the same material, carry the same fingerprints, are tattered in exactly the same way, etc., because these attributes can be used to connect a suspect to a crime. The twists and turns of the story in this case are not relevant; had all the pages been entirely blank, it would have been just as useful for the detective’s purposes.

As a shorthand, we will sometimes use the word “instance” to refer to a particular copy of some easily-duplicated thing, like a book or a computer file. You and I read different instances of the same book, we might say. But if we’re being careful, we’ll note that calling something “the same instance” is just making another identity claim. It is one where the relevant information content is the particular atoms of the object, as if we could go in and somehow tag them, to know that this particular copy of the book has almost entirely the same set of specific atoms as the one we say yesterday. This is a fictitious concept — since all atoms of a particular isotope are identical and can’t be tagged — but it works well enough in the current world, since (A) we assign ownership to specific instances of things, and (B) we don’t have the technology to duplicate an object at the atomic level, so there are usually observable physical differences good enough to stand in for the atomic labels. (Note that with digital objects like computer files, the situation gets considerably more subtle; in this case we usually distinguish “my copy” from “your copy” of the file not by the file itself, but from the physical medium it’s stored on — unless it’s stored in the cloud, in which case it comes down to accounts and passwords!)

So, while we can distinguish instances, for most purposes this is not the most useful sort of identity. But things may get sticky in cases where, due to coincidental technological limitations, instances happen to serve as a convenient shorthand for the sorts of identity we do care about, because the entities involved are not copyable. People obviously fall into this category. But we can imagine less emotionally laden scenarios with the same property: for example, imagine a world where each book is hand-written by its author, and the idea of copying someone else’s book is abhorrent. In such a world, there is exactly one instance of each book, and we would speak loosely about whether you and I read the “same book” by simply pointing to it; there it is, that’s the one we both read. In such a world, “common sense” would identify a book entirely by its physical instance; the idea of the same book existing in two places at once (let alone millions!) would be completely foreign. Heated arguments would erupt on the internet (if something like the internet could exist in such a world) about whether, if we did copy a book, that would have any relevance at all if the original book were consumed in a fire.

Exactly how and why common sense can lead us astray in this way will be a topic of our next essay. And in fact, we’ve just about covered the simple solution to personal (and other) identity well enough for now. Before moving on to preview future essays, let’s just take a moment to explore the implications of this solution. These too will be elaborated more later, but in brief:

  • The relevant information content of a person is the content of their minds — their memories, personality traits, hopes, fears, habits, and everything else that goes into their psychological makeup. (This was first advocated by John Locke in the 1600s, though he didn’t have access to fuzzy logic, so later philosophers poked holes in his version of the theory.)

  • You are mostly the same person over time because most of the information content of your mind is the same over time. Of course, as time passes, you become gradually less the same person as your younger self.

  • If we could duplicate a person, via a Star Trek-style transporter or by any other method, then the same person could exist in two places at once. This is no stranger than two copies of a book existing at the same time. It only feels strange to us because, like the folks in the world with only one copy of each book, we are simply not used to it.

  • Two instances of the same person would gradually become less the same person as each other, just as each of them becomes less the same person as their common younger self. This is no more mysterious than how we all become somewhat different from our younger selves over time.

  • If mind uploading technology is ever developed, we can feel comfortable using it for our own survival, because the information content of the post-upload mind will be the same as that of the pre-upload mind (this being the whole point of the technology). You are still you, even if you now exist in some man-made digital form rather than in a complex lump of proteins and lipids. Moreover, once uploaded, you can make regular backups and be assured that if you ever need to be restored from backup, that restored person is also still you.

Let’s close this essay now by previewing future installments, which will dive deeply into particular topics that might still be troubling some readers. (I will update this section as those future essays are written, and expectations gradually turn into reality.)

  • The uses and limitations of common sense and intuition
  • How to recognize a sound philosophy
  • An explanation (with brief history) of fuzzy logic
  • A brief historical recap of personal identity in philosophy
  • Relevant information content
  • The (fictional) spirit of things (and, things owned by famous people)
  • Scales of complexity (how people differ from rocks, books, and computer programs)
  • Duplicate identity and instances (contrasted with twins/clones)
  • Books and backups (and software and source code)
  • Other theories of personal identity, and their flaws
  • Review/summary

My hope is that time spent researching and writing these essays will be rewarded with less time spent countering the same tired old arguments on the internet — but far more importantly, I hope that it may prevent at least some people from throwing their lives away because of some fatal philosophical misunderstanding. Thank you for taking the time to read this solution, and I hope the essays to come clear up any confusion that still remains.