The Marshmallow Test Revisited: Trust, Broken Promises and Cuttlefish

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

The famous marshmallow test began at Stanford's Bing Nursery School in the early 1970s with no marshmallows at all: Walter Mischel and Ebbe Ebbesen offered 32 four year olds pretzel sticks or animal cookies, with a second treat for waiting 15 minutes. The researchers expected that seeing the reward would help children wait, and found the opposite. In 1972 follow ups with Antonette Zeiss, when marshmallows did appear, the children who lasted were the ones who covered their eyes, turned their chairs to the wall, sang, kicked the desk or fell asleep, and hiding the treat under a cake tin made waiting far easier.

Later follow ups tied preschool waiting times to parent ratings in 1988, SAT scores in 1990, adult BMI in 2013 and, in a 2011 fMRI study of participants in their mid 40s, to prefrontal versus ventral striatum activity. Then a 2012 University of Rochester study showed that children who first met an unreliable adult, one who promised art supplies and never delivered, waited about three minutes, while those who met a reliable adult waited about 12. Large replications by Tyler Watts and colleagues in 2018, confirmed in 2024, found the link to later success largely disappeared once family background was accounted for.

• Children told to imagine the sweet taste of the treat gave in just as fast as those staring at it
• A 2020 University of California study where waits nearly doubled when kids knew their teacher would find out
• A 2022 comparison: Japanese children waited longer for food, American children for wrapped gifts
• Cuttlefish at the Marine Biological Laboratory holding out 50 to 130 seconds for live grass shrimp over crab
• Why grabbing the sure reward can be the logical choice in an unpredictable world