Bayesian Thinking for Toddlers

Bayesian thinking is easy, and toddlers do it all the time. The fundamental principle is learning from experience: hypotheses that predict the data well receive a boost in plausibility, whereas hypotheses that predict the data poorly suffer a decline. For example, hypothesis A could state “At 6 am my parents will generally be asleep” and hypothesis B could state “At 6 am my parents will generally be awake”. When a toddler then wakes up at 6 am and notices that both parents are still sound asleep, this observation increases the plausibility of hypothesis A and decreases that of hypothesis B. Easy! Knowledgeable readers will discover that the dinosaur cover story hints at concepts such as Ockham’s razor, coherent knowledge updating, and probability as degree of reasonable belief. Statisticians may recognize Phil Dawid's prequential principle in action. Toddlers may just want to look at the dinosaurs.