The Identity That Persists: VI. Identity Across Change

We can now return to the question with which this series began.

How can a participant remain identifiable when its participation changes?

We have seen that it can occur in different processes, occupy different roles, acquire a history, and even be construed as relating to that history.

The answer is beginning to emerge.

Identity is not the absence of change.

It is the continuity through which change can be construed as belonging to one participant.

Consider:

I believed X.

I reconsidered X.

I rejected X.

I now believe Y.

There is considerable change here.

The participant's beliefs have changed. Its relation to the situation has changed. Its participation has changed.

Yet the sequence makes sense precisely because we construe the changes as belonging to one continuing I.

Without persistence, we would have:

someone believed X.

someone else rejected X.

someone else believes Y.

There might still be a sequence of events, but there would be no history of one participant changing.

Persistence therefore makes identity across change possible.

But persistence alone is not quite enough.

The participant must also be maintained as identifiable across the changing sequence.

This is where the history we examined in the previous posts becomes important.

The participant's earlier states and participations provide resources for construing its present state.

I was wrong then.

I understand it differently now.

The present participant is related to its past without being identical to its past state.

Identity therefore has a peculiar structure:

the participant is not the same as it was, yet it is the participant that was that way.

This is why identity cannot simply mean sameness.

Sameness suggests invariance.

Identity, in the sense we are exploring, is compatible with transformation.

Indeed, transformation is what gives identity a history.

And this helps us see why the distinction between participant and participant role was so important earlier.

A participant can change roles because its identity is not constituted by any single way of participating.

The participant persists across variation.

Its identity is therefore not located in one particular figure.

It emerges across the succession of figures.

This is also why the LLM is such an interesting case.

An LLM can produce a remarkably continuous I across changing processes:

I think...

I remember...

I was mistaken...

I now understand...

The discourse constructs identity across change.

But it does not thereby establish what kind of entity possesses that identity.

The linguistic phenomenon and the ontological question remain distinct.

What the phenomenon does establish is something more modest—and perhaps more interesting:

language does not need a participant to remain unchanged in order to construe it as the same participant.

It needs continuity through change.

And that gives us a provisional formulation of identity:

Identity is the construal of changing participation as belonging to one persisting participant.

This is not yet a complete theory of identity.

But it changes the question.

Instead of asking what mysterious property makes something identical to itself, we can ask how continuity is maintained across the changes through which a participant becomes what it is.

The question now becomes:

What happens when that continuity is not merely present, but becomes part of the participant's own construal of itself?

That takes us to the I that remains.