Skip to main content

Verified by Psychology Today

Changing Your Mind Based on New Evidence Is Natural

Successful animals know that learning is Bayesian.

Politicians are often accused of "flip-flopping" when they change their minds about a policy that they once supported. Is this such a bad thing?

As we learn more and more about COVID-19, public health recommendations change quickly. This has caused some commentators to discount the value of public health experts who "can"t make up their minds." Yet, nature shows us that wise animals modify their behavior based on new information and I suggest that this is a general strategy worth emulating. To understand why, we have to go back over 200 years.

The 18th-century minister, philosopher, and statistician Thomas Bayes developed a logic of decision-making based on accumulating evidence—learning. Differing from traditional statistical logic, Bayesian logic assumes that we have some prior knowledge about the likelihood of an event, and with experience, we update our estimate.

For example, a traditional statistical approach would assume that there is a 50:50 chance of being struck by lightning when venturing outside—which essentially means that being struck by lightning is random. By contrast, Bayesian logic assumes that on a sunny day, there is much less than a 50% chance of getting struck by lightning. By contrast, if you are on a golf course or alpine peak during a thunderstorm, there is a much greater chance. And, with additional information—for instance, the distance between you and each successive lightning strike—it"s possible to get a better estimate of the risk of being struck by lightning.

Bayesian logic learns from prior experience. Formerly, one begins with what's called a prior probability distribution and updates the "prior" based on accumulated evidence. This new posterior probability distribution is, as Bayesian's would assert, the best estimate of an event occurring—whether it's a lightning strike, or an attack by an eagle, lion, or terrorist.

Updating risk assessments based on new evidence should be ubiquitous. Some, but not all, animals have been shown to behave as though they are using Bayesian logic. For those that use Bayesian logic, we assume that natural selection has selected for reasonable priors. This is particularly true when we think about predation risk.

Rodents or primates living with snakes, for instance, might be more prone to respond alarmingly to a sudden encounter with a long thin object than those living without snakes. Or dense, vegetated habitats may be avoided because obstructed visibility increases predation risk.

When animals find themselves in novel environments—which occurs with increasing frequency by intentionally and unintentionally modifying the habitat, and shifting the species—we often see sub-optimal outcomes that could be based on a now faulty Bayesian prior.

As an example, consider animals recently relocated to snake-free environments. If there are no snakes around, jumping back in response to a slightly curved stick that may look like a snake is an unnecessarily time-consuming response. Or, consider the benefits of not foraging near shrubs that may contain a sit-and-wait predator like a snake. If snakes are no longer present, not foraging near shrubs unnecessarily eliminates potentially valuable food. We expect that, on average, animals making suboptimal foraging decisions will leave fewer descendants, and, more generally, natural selection will select for ways to eliminate costly responses if those costly responses are no longer needed.

So that brings us back to the criticisms of flip-flopping politicians and of varying COVID-19 recommendations. Accumulating evidence and basing your decisions on new evidence is a time-tested strategy for success.

We should support politicians who are amenable to changing their minds when new information arises. And, we should be prepared for changing public health recommendations as we learn more and more about a disease that we knew virtually nothing about a year ago. Our health and safety and health of our economy depend upon an idea formalized in the 18th century by a thoughtful minister.


Portions of this essay are adapted from my forthcoming book "The Nature of Fear: Survival Lessons from the Wild" which will be published by Harvard University Press in September 2020.