Rational Polarization (finally!)

It was a bench.

“Why was this outside your window?”, my parents asked pointedly.

I wasn’t quick on my feet. I had nothing. So I told them everything: “I’ve been sneaking out”, I began...

That was when it stopped. That was one in a long line of lucky breaks—for me.

But not for them. My friends were quicker on their feet; their parents slower to see the problem; their luck sooner to run out.

So we went our separate ways. While I went off to universities in liberal cities, many of them were stuck in our conservative hometown. While I was having my eyes opened by higher-education, some of them were fighting for their lives.

I was a near thing, but they made it. Obviously so did I.

Yet this isn’t a story about how a bench changed my life. It’s a story about how a bench changed my beliefs. So let me ask: what do you think happened to our politics? Who now worries about far-right militias, and who about Black Lives Matter? Who believes gun rights should be restricted, and who owns handguns for their own protection? Who voted for Biden, and who thinks Trump shook things up in a needed way?

I think you can guess. 

There’s nothing surprising about that. Most societies display local conformity with global disunity: people’s attitudes are predictable given their social group, despite varying widely across such groups. Failed institutions create libertarians, universities produce progressives, and the underside of Columbia, Missouri shapes you in a very different way than the academic centers of Cambridge, Massachusetts do.

The result is polarization: people who start out with similar beliefs but set out on different trajectories—like me and my old friends—often come to have disagreements that are profound, persistent, and predictable.

My question is why.

...

Well, I've finally got a full statement of my answer: I just posted a draft of my paper, "Rational Polarization". It shows how ambiguous evidence can lead rational people into predictable, profound, and persistent polarization—and argues that the it's empirically plausible that this mechanism plays a major role in real-world polarization.

The formal result: There can be a series of updates, each of which is expected to make your beliefs more accurate, which nevertheless collectively lead you to become radicalized in a predictable direction.

The experimental result: There is a (common) process of "cognitive search"—like searching for a word, or a memory, or an objection—that gives rise to such ambiguous evidence, and can be used to polarize real people.

The empirical result: Models and simulations suggest that the ambiguous evidence generated by this process plausibly helps to explain some of the core psychological drivers of polarization—including confirmation bias, motivated reasoning, and the group-polarization effect.


Comments and feedback welcome!! It would be incredibly helpful both to hear about detailed issues about the paper, or big-picture questions—my next project is to expand these thoughts into a book. Comment, or email me at kevindorst[at sign here]pitt.edu.

This project has been a long-time coming—thanks especially to everyone who commented on last year's blog series, where I first worked out these ideas in public. I've done my best to remember as many names as possible in footnote 64 of the paper; sorry if I've missed you!