(Note: This post appeared on Scientia Salon on October 24, under the title Identity, a neurobiological perspective, with numerous comments. I am placing a copy here for convenience. The discussion has closed on Scientia Salon, but I would be happy to continue it here.)
The philosophical problem of identity is epitomized by the paradox known as the “Ship of Theseus”. Suppose a ship is rebuilt by removing one plank at a time, and replacing it with a new plank of the same shape and material. Is it still the same ship? Most people would say so. But suppose all the planks that were removed are brought together and used to construct a new ship of identical form. Wouldn’t it be more appropriate to say that is the same ship as the original, and the one with new planks is a duplicate? There is no easy answer. Every possible reply seems to lead into a morass.
The Ship of Theseus and several related paradoxes have been tangling philosophers in knots for thousands of years, dating back to the ancient Greeks and continuing with Locke, Hume, Kant, etc. It is easy to get a sense that no real progress has been made: after all the modern philosophers have spoken, the paradoxes seem just as paradoxical as they did to the ancient Greeks. It seems to me, though, that there is a route to a clearer understanding, by looking at identity in a different way: from a cognitive neuroscience rather than philosophical perspective. I propose to examine identity as a mechanism that the brain uses to organize the world of perception.
The basic premise is that at a cognitive level, an identity is a label attached by the mind or brain to a portion of the world, usually one that is continuous in space, time, and form. To motivate that premise, it will be useful to briefly examine the way that identity comes into play in modern physics and in computer science.
Identity in physics
The concept of identity is so deeply built into our thought processes that it is hard to examine. Cognitively we live in a world of things, each of which is a distinct individual thing and remains the same thing over time, as long as it continues to exist. It seems impossible to imagine a world that is not divided into things. However, the world of modern physics is just such a world. In modern physics, objects are merely clusters of matter and energy that cohere for some period of time—they have no deeper reality. In modern physics, even subatomic particles do not have individual identities.
It might help to understand this point by considering a hurricane. We think of a hurricane as a distinct entity, but really it is just a temporary arrangement of air and wind. The air that makes up a hurricane on one day has nothing in common with the air that makes it up on the following day. Its edges are indefinite in space, and its birth and death are indefinite in time. As an entity, it is purely dynamic. It shows rough continuity in space, time, and form, but no continuity in substance.
In modern physics, there is no level of identity higher than a hurricane. Some arrangements of matter may last much longer, and have borders that seem less arbitrary, but these are really just matters of degree. For a hurricane, it is reasonably easy to see that its identification as a distinct entity is a process performed by our brains. It is not quite so easy to see that the same fact applies to objects such as rocks and bodies, but it is just as true.
Once it is accepted that identity is structure our brains impose upon the world, two basic questions follow immediately. (1) Why do our brains organize the world as an array of individual things? (2) How do our brains do this? That is, what is the neural mechanism for assigning unique identities to parts of the world and tracking them over time?
Identity in computer science
The value of identity might seem intuitively obvious, but it is useful nevertheless to see how it comes into play in computer science, specifically in database programming.
A database, to a programmer, is a collection of data organized into “records” and “fields”. A “record” represents the data that comes from one individual entity—perhaps a person, place, website, purchase, data point in a scientific experiment—really any sort of thing. A “field” represents some property of that entity. For example in a database used to hold a record of transactions for a corporation that sells things on the web, each record might represent one specific transaction, and might have fields for the customer name, address, credit card number, item purchased, price paid, date and time of purchase, delivery tracking number, etc.
In most databases there is some field that serves as a “unique identifier”—in other words, a field that serves to uniquely identify each specific record and distinguish it from all other records. Sometimes unique identifiers are taken from pre-existing information. For example, in a database whose records represent people, it is common to use the social security number as a unique identifier. Everybody (within certain populations) has one, and no two people share the same one. Another frequently used unique identifier is a credit card number: no two people should ever have the same one. In many cases, though, unique identifiers are generated essentially at random. For example in a database of transactions, there might be a field for the “transaction number”, which is obtained by adding 1 to the transaction number for the previous record.
The function of unique identifiers is to avoid confusion. If the database is searched using other fields, such as name or date and time of purchase, it is hard to guarantee that the wrong record won’t be retrieved. But any record that has the correct value for the unique identifier is sure to be the correct record. Thus unique identifiers work to keep information in records that are similar in some way from getting mixed up with each other.
The only really essential property of a unique identifier is uniqueness. There must never be two records that have the same identifier value. This property is so useful, though, that it justifies using database space to store fields that have no other meaning.
A consequence of uniqueness is that identifiers are usually arbitrary and are usually generated by some central authority. If identifiers have information coded into them, or if they can be generated by multiple agents, it is hard—not impossible, but hard—to avoid the possibility of two records ending up with the same identifier. Think about social security numbers: they don’t carry any information about a person (actually they do carry a little bit, but by accident), and don’t serve any purpose except to distinguish a person from everybody else in databases.
Identity in the brain
My premise is that the human brain contains a database consisting of memory-records of clusters of matter (“things”) that are continuous in time, space, and form, and labels these memory records with unique identifiers that we call “identities”. Often we refer to identified entities using names or certain types of descriptions—”John Q. Jones”, “the planet Jupiter”, “my car”, “this house”—but those are not unique identifiers. The unique identifiers are cognitively opaque.
Why does the brain do this? Because individual things—continuous clusters of matter—often have properties that remain valid over time, but are not easily discerned. Individual people, for example, have personality features that remain generally valid, but can only be recognized on certain occasions. Labeling things with identities allows their covert features to be tracked across time.
Identification—the term I will use for this process—is so deeply built into our cognition that it is hard to think about clearly. It is hard to think about the world at all without thinking about it as a world of things that remain the same things over time.
The process of identification obeys two basic rules. The first is continuity. In order to be identified as the same thing, a cluster of matter must change location continuously and change form continuously, without any time gaps. Thus if I make a fist, it remains the same fist as long as I keep my hand clenched, but if I open my hand and then clench it again, it becomes a different fist. This is however a soft rule, and we are open to the possibility of it being violated to some degree, as in the Star Trek Transporter and the Ship of Theseus.
The second rule is uniqueness, and this one is inviolable. The rule that an identity can only have one exemplar is so deeply built into our thinking that violations seem almost like logical contradictions. But it is really no more contradictory for two bodies to be the same person—to have the same identity—than it would be for two people to have the same social security number.
Why is the uniqueness rule inviolable? Because the way our minds work depends on it. We remember information about things by linking that information to their identities. If two clusters of matter are labeled with the same identity, then it is impossible to keep any distinct properties they might have from becoming confused in our minds.
Solving the paradoxes
If these premises are accepted, then the explanation of paradoxes such as the Star Trek Transporter and Ship of Theseus is straightforward. They work by setting up scenarios that force us to violate the rule that an identity can only have one exemplar, and we simply can’t do that. Consider the Star Trek Transporter: the scenario is that the Transporter fails to work properly and leaves an exemplar of Kirk on both ends. Which one is the real Kirk? The obvious answer is that they are both the real Kirk, but our minds can’t process that. We can’t think about the two exemplars distinctly unless we assign distinct identities to them. But there is no way to do so without making a decision that seems arbitrary—hence the paradox.
If we accept that an identity is a label attached by the brain to a cluster of matter, then we immediately get several questions about brain mechanisms. How are identities generated and stored by the brain? The answer is unclear, perhaps largely because nobody has been thinking in these terms, but surely must involve the inferior temporal lobe (the terminus of the so-called “what” stream).
And if Identification is a brain process, which animals other than humans have it? The answer is not clear at all. It seems possible to get highly sophisticated behavior without Identification—by responding only to the directly perceptible features of the environment. It does not seem likely that Identification is ubiquitous throughout the animal kingdom. On the other hand, it seems likely that Identification is a necessary precursor for the processes known by the term “Theory of Mind”. It’s hard to see how knowledge and beliefs can be attributed to another animal without first representing that animal as a uniquely identifiable individual.
There is a lot more to say about these issues, but I am out of space.