What is recency bias and why should you care?

A Q&A with Carnegie Mellon Professor and behavioral economist Manasvini Singh

Happy new year! If your New Year’s resolution is to build a healthy new habit, then you might enjoy some back issues of this newsletter focused on goal-setting and habit formation. And don’t miss my recommended listens and reads below, featuring a flurry of interviews I did right before and after New Year’s about harnessing the fresh start effect.

Today I want to talk about a bias that can be a real pain in the neck – whether you’re choosing the right goal to tackle or the best restaurant to recommend to a friend. It turns out we have a suboptimal tendency to fixate on what’s top of mind, and one way things get to the top of mind is by being recently encountered. What happened yesterday or last week is a lot more likely to pop into your thoughts (and influence them) than what happened last month, last year, or last century – even if an event from the more distant past has greater relevance to your next decision. Unfortunately, this can lead to mistakes.

Consider a homebuyer contemplating a move to Miami or Los Angeles. Both cities face major climate risks, but might the recent fires in Los Angeles loom larger this month, tipping the balance towards Miami? In this month’s Q&A, we’ll explore recency bias and its impact not just on everyday decisions, but also on life-and-death choices—such as the procedures doctors decide to perform on their patients.

But before we dive in, here are a few listens and reads I think you might enjoy…

This Month’s Recommended Listens and Reads

• Pretty Woman Meets Behavioral Science: I had a blast visiting the podcast Love Factually (hosted by two fellow academics) to talk about what this classic 90s movie gets right and wrong about human nature, according to behavioral science research.

• Beware the Dangers of Data: This thoughtful Economist piece discusses implications of a recent paper I co-authored (led by Drs. Linda Chang and Erika Kirgios) on quantification fixation – or our tendency to attend more to quantified attributes of products, job candidates, charities, and so on when making decisions.

• New Year’s Roundup: I did a slew of interviews over the last month about the science behind why we set New Year’s resolutions and how to achieve them. My favorite was probably this interview with NPR’s “Here & Now”, but you can also find my advice on WHYY, in The Washington Post, in National Geographic, on CNN, in U.S. News & World Report, and on NPR’s “All Things Considered”.

• January was a big month for behavioral science books! I strongly encourage you to check out these terrific new titles from fellow researchers: (1) MINDMASTERS: The Data-Driven Science of Predicting and Changing Human Behavior by Sandra Matz; (2) DEFY: The Power of No in a World that Demands Yes by Sunita Sah; (3) PING: The Secrets of Successful Virtual Communication by Andrew Brodsky; (4) MAKE WORK FAIR: Data-Driven Design for Real Results by Iris Bohnet and Siri Chilazi; (5) NEGOTIATION: The Game Has Changed by Max Bazerman; (6) INSPIRE: The Universal Path for Leading Yourself and Others by Adam Galinsky; and (7) OUTRAGED: Why We Fight About Morality and Politics and How to Find Common Ground by Kurt Gray.

Q&A: Why What’s Top of Mind Matters

In this Q&A from Choiceology, Carnegie Mellon Social and Decision Sciences Professor Manasvini Singh discusses her research showing that we overreact to recent events, even in incredibly high stakes situations.

Me: I'm excited to talk about recency bias. Can you start off by defining it?

Manasvini: When we’re exhibiting recency bias, we’re making decisions that overweight things that happened closer to us in time. Things that just happened as opposed to things that happened long ago.

Me: When I think of recency bias, I imagine something like a friend asking me for a restaurant recommendation in my home city. And instead of thinking of the best restaurant in Philadelphia, I mention the place I went last week because it's what comes to mind first.

Manasvini: Exactly right. My go-to example is that I almost get into a car crash, and then I’m a little bit more careful the next day, a little less careful the day after, and by the third day, I'm basically back to normal.

Me: Oh, I love that example. You have this fascinating paper about the decisions doctors make about how to treat their patients. I interpret it as a great example of recency bias. Can you describe that work?

Manasvini: Of course. I look at whether physicians, specifically obstetricians, overreact to recent negative patient outcomes. If you have a physician, Dr. Mary, for example—Dr. Mary is going about her day, and then she performs a vaginal delivery of a baby, and there’s a bad patient outcome. On the next patient, is she more likely to perform a cesarean? I actually show that it happens in both directions. Patient 1's complication in a vaginal delivery makes it more likely that the physician switches to a cesarean on the next patient. And if there's a complication in patient 1's cesarean, the doctor is also more likely to switch to a vaginal delivery on the next patient.

Me: So, whether patient 1 is cesarean or vaginal delivery, if it doesn't go well, the next patient gets the opposite kind of treatment with a higher probability.

Manasvini: Exactly. And I'm not saying it's conscious. It's not like the physician says, "Oh, now I have to switch to this other delivery mode." If you think of the physician's threshold for performing cesarean, maybe the threshold is lowered, and they're more likely to do a cesarean on the next patient, or now it's higher, so they might be more reluctant to do a cesarean on the next patient.

Me: Going back to the driving example. If you have a near miss or an accident when you're driving, now it's going to be top of mind to be careful. So that's one example of recency. And this seems like another where if you're a doctor, you're trying to do the right thing for every patient. Am I thinking about that correctly?

Manasvini: Yes, and I think the medical setting is especially where such biases are likely to flourish because there's so much uncertainty. And when there's uncertainty, there's a greater likelihood of relying on biases.

Me: What do we know about the root cause of this type of behavior? Why would our minds be designed to swing drastically like this, overreacting to our most recent experiences?

Manasvini: There seems to be some sort of biological basis to this type of learning. There's evidence in how bees make formations and ants and foraging behaviors amongst rats and also how children learn language—this sort of incorporating the outcome of the last decision into your current decision, what I call "win, stay; lose, shift." And maybe in simpler environments, relying on your last decision is good.

Me: When things are changing a lot, I suppose I see how that could be smart. Could you talk more about when this kind of behavioral pattern is suboptimal?

Manasvini: That’s a very important question because it differentiates learning from a bias, and it's very, very hard to differentiate the two. The way I try to get at it is that I show that the patient for whom the physician used this heuristic has worse health outcomes. The mother and child have a slightly higher likelihood of dying if they follow a prior complication and there’s a switch performed on them.

Me: And this is using massive amounts of data from multiple hospital systems that you're able to see these patterns over, was it decades?

Manasvini: Over 20 years, two hospitals, because I needed electronic medical record data. And then the other thing that makes it inconsistent with learning is that experienced physicians are more likely to use this heuristic, which goes in the face of the most basic learning models. And that the physicians who use this more often are more likely to have bad outcomes over time.

Me: We've been talking a lot about this in the context of your fantastic study of medical decision-making. But I want to zoom out from that for now. If you could give one piece of advice about how people can improve their everyday decisions and avoid recency bias, what would you suggest they think about or do differently?

Manasvini: When making decisions, pause for a second and try to prime yourself with a scenario that has happened to you in the past, so you can take a more holistic view of the decision you're making currently. Make other things more recent. Even though they might not be recent in time, they will always be readily accessible in your memory. For example, if you’re going to take a big exam, and you didn't perform that well on your last one. Maybe you're nervous, and you're overweighting that past exam, but just remember that in your life you have done well several times, and you’re doing better than you think you are. And maybe that will boost your confidence and make you do better on this test.

Me: That's really nice. Thank you for taking the time to talk about recency bias and your amazing research.

This interview has been edited for clarity and length.

To learn more about Manasvini’s work, listen to the episode of Choiceology where we dig into recency bias or check out her fantastic paper in Science magazine on how this bias can contort life-and-death medical decisions.

That’s all for this month’s newsletter. See you in February!

Katy Milkman, PhD

Professor at Wharton, Host of Choiceology, an original podcast from Charles Schwab, and Bestselling Author of How to Change

P.S. Join my community of ~100,000 followers on social media, where I shares ideas, research, and more: LinkedIn / Twitter / Instagram / BlueSky / Threads