Reason Evolved Not to Help Us Make Better Decisions, But to Help Us Win Arguments
The case for the argumentative theory of reason, as made by The Office
Quick personal note: I’m moving over to Substack to try on writing full time!
Thanks in advance for your support!
“Mmm…kind of an oaky afterbirth.”
— Michael Scott, referring to a bottle of red wine in “Dinner Party”
“Dinner Party” — one of the best, albeit cringiest, episodes of The Office — finds Jim and Pam tricked into a dinner party at Michael and Jan’s condo. The whole episode is worth watching, but for the sake of this essay, I’m going to focus on one scene.
(Note that prior knowledge of The Office is not necessary to read this essay, though ardent fans will be more familiar with the scene in question. That said, if you want to watch it, I’ve linked to it below. The scene in question starts at 2:10.)
An hour or so into the party, after an unimaginably awkward tour of the apartment, the cameras capture Jim in the bathroom.
“Michael and Jan seem to be playing their own separate game,” he begins, “And it’s called ‘Let’s See How Uncomfortable We Can Make Our Guests.’ And they’re both winning. So I am going to make a run for it.” The next time we see him, he’s emerging from the bathroom.
“So you’ll never believe this,” Jim begins, speaking to the room. “My landlord just called. Apparently my apartment flooded. Something with the sprinkler. Pam, we should probably get going and see the damage.” Pam’s look as she realizes what Jim is up to is classic — this slow smile, followed by a measured, concessive shrug, which she executes flawlessly as she eases off the couch to join him.
“Okay…” she says, trying — and failing — not to betray her eagerness to leave. Michael, though, isn’t giving up so easily.
“You don’t need two of you to do that!” He exclaims.
“That’s…true…” Jim says, trailing off. Because Jim is now thought to have a flooded apartment, though, he continues towards the door.
“Dinner sounded delicious,” Jim says, feigning regret. “Pam — I’ll see you at home?” Pam, realizing she’s about to be left alone at the party, takes action.
“Oh, Jim, I don’t think you’re going to abandon this party here all by itself,” she says.
“I don’t know, because everything I own is there,” Jim says.
“You can buy new stuff,” Pam begins, “but you can’t buy a new party.”
“That is a great point!” Michael says to Jim, who we catch looking at Pam, almost impressed, as Michael ushers him back into the living room.
Of course, this is a TV show — not real life. It’s hard to imagine any host pushing back on their guests’ leaving to address a flooded apartment,¹ not to mention that Jim — who worships Pam — would ever actually leave her at a dinner party with Michael and Jan.
And yet, the back and forth in the scene — however unrealistic it might seem — illustrates an important truth about the human condition.
That is, how we use reason, and why.
For most of history, psychologists have analyzed reason as a tool we use to make good decisions. Given how we use the word, this makes sense. We “reason” our way out of tricky situations. We refer to good decisions as “reasonable.” We beg people who are disagreeing with us to “see reason.” Even the Enlightenment is associated with the “sovereignty of reason.”²
And yet, analyzing reason as a tool we use to make good decisions reveals all sorts of problems with it. The main one is confirmation bias — the cognitive “error” of seeking out information, or “reasons,” that confirm what you already believe.
Perhaps nothing has shown our susceptibility to confirmation bias like the pandemic.³ Those against wearing masks seek out evidence that says they don’t work; those for wearing masks find evidence saying they do. This is not the use of reason to arrive at the right decision — although you can be sure that both parties believe they’ve arrived at it — rather, it’s the use of reason to confirm what you already believe. In the case of masks, this bias can be deadly.
The existence of confirmation bias is more than enough to make us question that reason evolved to help us make good decisions — the hypothesis that, up until recently, dominated academic thought.
So, we’re left to wonder — what hypothesis does its existence support?
In a 2011 paper titled: “Why Do Humans Reason? Arguments for an Argumentative Theory,” Hugo Mercier and Dan Sperber argue that humans use reason not to make decisions, but to win arguments.
Evolutionary biology supports this idea. Tens of thousands of years ago, our ancestors would’ve collaborated in all sorts of ways, to make all sorts of decisions. This collaboration would’ve required communication. But there is a problem here, noted by Mercier — that is, communication opens the door to manipulation.
If you and I are having a conversation, and I accept everything you tell me, you might realize that you can manipulate me into doing things that benefit you, and more importantly, might not benefit me. This could explain why we evolved some bit of skepticism — Mercier uses the technical term “epistemic vigilance” — in order to evaluate what we’re told. The tricky part emerges when you tell me something that contradicts something I already believe. As we’ve seen with social media, people don’t always like having their beliefs challenged. It’s uncomfortable because it makes us feel vulnerable. Naturally, our reaction to being told something that contradicts something we already believe is to reject it.
Now, this is all well and good on social media — or at least “well and good” from the perspective of actual, physical risk. Parties arguing over opposing views on Facebook are not risking life and limb if they fail to reach an agreement.⁴ Our ancestors, though, weren’t so lucky. Say you tell me that an opposing clan is coming to murder us. You know you’re right, but I don’t believe you — say, because I think you have some ulterior motive.⁵ Suddenly, the incentive for you to convince me is incredibly powerful. This, Sperber proposes, is why reason evolved.
Of course, our world is different than our ancestors’. Arguments with life and death consequences still happen, but they’re more likely to be relegated to places most of us aren’t. The arguments I’m used to are less about trying to convince someone that someone’s coming to murder them, and more about trying to figure what golf course to play this weekend, or what to eat for dinner.⁶
We all know how these arguments tend to go. Someone asks where people want to eat, and you throw out an option.
“Chipotle,” you say.
“Eh, why Chipotle?” Someone says back.
“Yeah, I feel like we always go there,” someone else says.
So you start you pulling evidence out of thin air to explain why you want to eat at Chipotle — “It’s close,” or “It’s cheap,” or “It’ll be quick,” etc.
Your friend might either refute what you’re saying — “It’s not that close,” or “There’s a pork shortage, so the prices are up,” or “It’s Halloween, so everyone and their mother will be there in costume trying to get a free burrito” — or start giving reasons why another place might be better.
“Panera is healthier and closer, and I have a gift card,” they might say. And this might go on for a while. The next thing you know, it’s devolved into an actual fight, and you’re going back and forth about which chain’s carbon emissions are more deplorable, and renouncing each other’s politics over something as simple as where to eat for dinner.
This is a bit of a cartoon version, but it should be at least vaguely familiar to anyone who has ever found themselves in an argument — especially one that gets heated, as arguments often do. In those situations, it’s easy to find yourself unconsciously pulling evidence from wherever you can to support your argument.
It’s here that the existence of confirmation bias begins to look a lot more like a feature, and a lot less like a bug.
Think about it: If my goal is to convince you that Chipotle is where we should eat for dinner, a tool that generates arguments both for Chipotle and against alternatives would be pretty damn useful.
Of course, the stakes are low here, but in other cases, they might not be. Our ancestors’ confirmation biases would’ve made it easier for them to find evidence to support whatever case it was that they were trying to make — like, say, that everyone should turn tail and run, because an opposing clan was actually coming to murder everyone.⁷
It’s unlikely that we’d be presented with the same hypothetical today, but that doesn’t make what we’ve inherited from our ancestors any less useful.
Let’s return, for a moment, to The Office.
It’s clear that neither Pam nor Jim — nor Michael, for that matter — were in any mortal danger throughout the course of “Dinner Party.” Still, it’s fascinating to watch the interplay of their reason and confirmation biases work to help them achieve their desired ends throughout the episode — perhaps even more so because the stakes are so low.
When the scene begins, both Jim and Pam have the same end in mind: escaping the party. Jim’s confirmation bias provides him with the “flooded apartment” story. Pam’s confirmation bias then kicks in immediately, albeit implicitly, as she realizes this story aligns with her desired end, too.
Michael’s desired end, of course, is preserving the evening. His confirmation bias kicks in as he realizes he’s at risk of failing to achieve it.
“You don’t need two of you to do that!” He exclaims — “that” being leaving the party to address Jim’s flooded apartment. As Jim concedes Michael is right and says goodbye to Pam, her desired end shifts from leaving with Jim — no longer an option, given Michael’s quick thinking — to keeping Jim at the party.
“Oh, Jim, I don’t think you’re going to abandon this party here all by itself,” she says. It’s a weak reason, but a reason nonetheless — her confirmation bias buying her time, and maybe the support of others at the party.
Jim exercises the same bias, albeit to achieve his desired end, which is still leaving the party.
“I don’t know, because everything I own is there,” he says, which would be a fair reason to leave, if he weren’t lying about apartment being flooded. Add to this that Pam knows he’s lying, and suddenly, her confirmation bias kicks into overdrive. She presses him, knowing her actual reason for keeping him there trumps his reason for wanting to leave.
“You can buy new stuff,” she says to Jim. “But you can’t buy a new party.” With that, Jim gives up, allowing Michael to usher him back into the living room.
What makes Pam’s behavior in this episode such a compelling use of reason and confirmation bias is that we see her using the tools two separate times in two different ways to achieve two entirely different ends.
And though Pam is a fictional character, she is also all of us. Our motivations, like hers, change constantly and without warning. Confirmation bias, far from being a bug in our cognition, is, in fact, a feature — one of the best tools in our toolkit to help us win arguments, and thus systematically derive the most value from our circumstances.
In some cases, this might mean escaping a dinner party.
In others, it might mean making sure you’re not left alone at one.
¹ The apartment wasn’t actually flooded, but Michael didn’t know that.
² Not to mention that the “Age of Reason” Wikipedia page redirects you to the “Age of Enlightenment.”
³ There is a book waiting to be written on the use of confirmation bias during the pandemic.
⁴ This is not to say that there exist no broader, second-order effects of disagreements online. I’ve written about those here.
⁵ Maybe I think you’re trying to frighten me into deserting the village, and will then come steal my things once I’m gone.
⁶ The joke here is that it always seems to be the arguments with the least on the line that turn into shouting matches, perhaps because there’s some innate, unspoken frustration between the disagreeing parties for spending energy fighting over something so trivial. Just my armchair psychologist take.
⁷ If you think about it, it’s actually not that much of a long shot that one of us is descended from someone that either did something like this, or had something like this happen to them.