Considerable overlap

There is this joke among park rangers in the US that it will never be possible to construct a food container that is perfectly bear-proof, because ‘there is considerable overlap between the dumbest humans and the smartest bears’.

Well, thanks to Lorenzo Ciccione and his colleagues we now know that there may also be considerable overlap between humans and baboons.

Part and parcel of the skill set of economists is to perform linear regression to examine if there is a correlation between two variables. The researchers knew that some primates could do simple probability assessments when trying to get food, so they decided to up the ante and ask whether baboons could do linear regressions.

To test this, they recruited 23 baboon volunteers (16 females, average age 14.5 years) and showed them pictures of scatterplots with different slopes and dot densities. The chart below shows some sample scatterplots with the t-statistic of the correlation added for clarity (the baboons didn’t get to see the t-stats, only the scatter plot).

Sample scatter plots evaluated by baboons

Source: Ciccione et al. (2024)

After an initial training period, during which the baboons learned to signal whether a scatterplot had a positive correlation (rising slope) or negative correlation (declining slope), they gave the baboons a series of tests and compared their assessments with those of 3,943 human volunteers (2,409 females, average age 28.8 years).

In some regards, the baboons weren’t as good as humans in assessing the relationship between two variables in a scatter plot. If the number of data points was very low (six dots), humans could still identify the scatter plot as one curve, while baboons could not. The researchers speculate that this might be a learned trait since people tend to learn about scatterplots in school while baboon schools tend not to have introductory statistics courses.

But apart from that outlier, the results were pretty similar. Baboons could identify trends in scatterplots with smart baboons outperforming dumb humans. In general the higher the t-statistic (i.e. the closer the relationship) the easier it was for humans and baboons alike to identify the slope of the curve. But in general, we can once again conclude that there is considerable overlap between the smartest baboon and the dumbest economist human.