Mice treat sunk costs as real

From Erica Goode at The New York Times:

In a study published on Thursday in the journal Science, investigators at the University of Minnesota reported that mice and rats were just as likely as humans to be influenced by sunk costs.

The more time they invested in waiting for a reward — in the case of the rodents, flavored pellets; in the case of the humans, entertaining videos — the less likely they were to quit the pursuit before the delay ended.

“Whatever is going on in the humans is also going on in the nonhuman animals,” said A. David Redish, a professor of neuroscience at the University of Minnesota and an author of the study.

Via Michelle Dawson, here is another study with a differing emphasis:

We found that the sunk cost effect was lower in the ASD [autism spectrum disorder] group than in the control group.

Here are previous MR posts on sunk costs.


'The more time they invested in waiting for a reward — in the case of the rodents, flavored pellets'

So, it appears the rodents actually were reliably fed, though the interval was variable. This would not be proof of sunk costs as real, but instead of the rodents finding a reliable source of treats. It isn't as if the caged rats could do something else to get those flavored pellets, after all.

Though this is the sort of post one would expect from Prof. Tabarrok, whose less frequent appearances might also be considered an example of sunk costs being real, at least in terms of waiting around for his particular brand of insight.

"The more time they invested in waiting for a reward — in the case of the rodents, flavored pellets; in the case of the humans, entertaining videos — the less likely they were to quit the pursuit before the delay ended."

A true sunk cost has no bearing on the future. Is it really true that at any point the amount of time a rat might have to wait for its pellet has no correlation with the amount of time it has already waited?

Good point. I think the study assumes waiting is a Gaussian distribution (normal curve) when in fact many queues it is found are in fact a Poisson distribution, with "fat tails" meaning sometimes the wait is once in a while extraordinarily long (recall your favorite horror story waiting in line at the bank or supermarket). So perhaps rats, mice and humans have internalized this bit of statistical wisdom? (Not) Fooled by Randomness!

How many rats starved to death waiting for the pellet?

I feed two stray cats for my elderly parents every morning- they started it several years back, so I have continued it as they both care for the critters- they are at the door every morning waiting to be fed, even though my routine varies by up to 5-6 hours at times (between about 7 am to noon since my schedule is a bit variable). I have wondered how long they would wait if I didn't feed them- I do doubt that I would ever find them dead on the back stoop because I didn't feed them for two months.

You would expect two species that share so much DNA had similar underlying process in a lot of things. That is why we test drugs on rats.

But given the field and the replication crisis I am not sure why we should believe it except that it is in accord with our prejudices.

Just so. Given the trend of recent decades at both NYT & Science, any reports from then are suspect. Is there really a controlled experiment here, when the rats have very limited alternative ways to spend their time?

There's a person who's job is central to this and similar studies. After the studies are completed his role is to euthanize the rodents, ie. kill them with gas.

Talking to one such person they said mice are actually less lovable and more anti-social and likely to bite you than rats. Mickey Mouse is nasty.

But they were lab mice. What else do they have to do?

I spent the time reading this post,


I might as well comment.

That was a sunk cost joke if you didn't get it.

The more sunk cost research I read, the more I come to the conclusion that researchers don't understand sunk cost and that most examples used to illustrate sunk costs decision making are actually examples of uncertainty on how the marginal costs have changed since incurring the original cost.

I was thinking the same thing, particularly since this blog is based on trade-offs and expectations at the margin. This simply appears to validate Pavlov. To the larger point of sunk cost, how do opportunity and transaction costs factor in the equation? Economists seem to neglect that there these costs can and often do off-set the sunk cost effect. If anyone could enlighten me on why this isn't the case I'd greatly appreciate their insight.

I agree. As Ray Lopez pointed out above, the authors not only assumed a wait time distribution, but also that the mice used that is their model. Even if the model of the distribution of costs is correct, having the wrong model in your expectations isn't the sunk cost fallacy.

I am heartened to hear that research which shows that humans engage in the sunk cost fallacy is now shared with another animal in the kingdom, mice. I remember reading a pigeon sunk cost fallacy study.

So, its us, mice and pigeons.

And rats. Decades ago, experimental economists discovered that Giffen goods do exist -- if the consumer is a rat.

Actually, I see that some psychologists published first, by over a year:

I am certainly not autistic but I am a sort of mathematician, and I rarely fall for the sunk cost fallacy. For example, if a movie is terrible, I get up and leave (assuming I am alone) even though I paid for it.

Comments for this post are closed