Bryan Caplan  

The Flynn Effect vs. the Greatest Minds in History

Misconceptions about taxes... What does it mean to say "The ...
Measured intelligence (IQ) rose markedly over the 20th century, a robust result widely dubbed the "the Flynn effect."  But what do these rising IQs mean?  The straightforward conclusion is unbridled optimism: just as human beings are taller than they used to be, they're also smarter than they used to be.  Flynn himself is reluctant to draw this inference, but perhaps he's too cautious.

How could IQ go up while genuine intelligence stagnates?  The simplest story is that modern societies somehow "teach to the test," leading to "hollow gains."  People score higher today because they've been prepped better, even though they're no smarter than before.  Flynn's moderately optimistic account is that today's humans are genuinely better at a narrow but important range of cognitive tasks.  We're no smarter than we used to be, but we're much better at abstract and hypothetical reasoning. 

What do rising IQs really show?  I remain undecided, but here's an argument that strongly inclines me to pessimism.  To wit: When I read the smartest thinkers from the 17th, 18th, and 19th centuries, they seem roughly as smart as the smartest thinkers from the 20th century.  In fact, the same goes for the smartest Greeks from the 4th and 5th centuries B.C.  What 20th-century thinkers credibly exceed the sheer intellectual firepower of Plato, Aristotle, Hume, Kant, Descartes, or Gauss?  Note: I'm not naively comparing the best living thinkers to the totality of earlier thinkers.*  I'm comparing the best in the last century to the best in individual earlier centuries.

The high relative quality of the top minds of the past is especially glaring when you consider two weighty factors that misleadingly tip the scales in favor of modernity:

1. Population is much larger than in earlier eras.  So you'd expect the extreme tail of today's intelligence distribution to outshine the extreme tail of the past even if mean intelligence stayed the same.

2. Modern thinkers build on the shoulders of past giants, making it easy for them to exceed the knowledge and avoid the errors of earlier generations.  This in turn fosters the illusion that moderns are genuinely smarter, rather than advantageously positioned in time.

3. Women in earlier centuries would have had little opportunity to impress the world with their intellects.  So the past is playing with only half its team.

My claim: Even if you don't bother correcting for these three confounds, the top thinkers of the past still seem every bit as brilliant as the most brilliant modern minds.  And if you did run suitable corrections, you might well conclude that the smartest people since 1900 are decidedly inferior to their predecessors.

In light of international adoption research, the most promising way to resolve my puzzle is to attribute the Flynn effect to the gradual elimination of absolute poverty.  Nutrition is the most obvious mechanism: Since history's greatest minds generally came from well-fed privileged classes, their intellectual development was barely stunted, leaving them mentally in the same league as today's top minds.  The rest of mankind, however, has enjoyed massive cognitive gains due to massive growth in food availability. As Flynn explains the story
The major argument for nutrition as a post-1950 factor rests not on dietary trends, but on the pattern of IQ gains. It is assumed that the more affluent had an adequate diet in 1950 and that dietary deficiencies were concentrated mainly in the bottom half of the population. This has been stated as a hypothesis about class: Over the last say 60 years, the nutritional gap between the upper and lower classes has diminished; therefore, the IQ gap between the classes should have diminished as well; therefore gains should be larger in the bottom than in the top half of the IQ curve.
But Flynn goes on to present strong evidence against not only the nutrition story, but any story that appeals to the decline of absolute poverty (references omitted):

There are seven nations for which we have the whole IQ distribution from top to bottom: France from 1949 to 1974; The Netherlands from 1952 to 1982; Denmark from 1958 to 1987; the US from 1948 to 1989; Spain from 1970 to 1999; Norway from 1957 to 2002; Britain from 1938 to 2008. Denmark, Spain, and Norway show gains either larger or almost wholly in the bottom half of the curve, but France, the Netherlands, and the US show uniform gains over the whole curve. Britain is a special case, which I will save for detailed analysis.

Where we do not have the full distribution, a sign that gains might be concentrated in the lower half would be that the range or variance (the S.D.) of IQ scores has lessened over time. If the lower half has gained, and the upper half has not, clearly the bottom scores will come closer to the top scores. A survey of the better data sets shows that Belgium, Argentina, Sweden, Canada, New Zealand, and Estonia have no pattern of declining variance. In Israel, males show no decline but females do; however, the female data are inferior in quality and it is hardly plausible that the latter had a worse diet than the former.

Therefore, as far as we know, nutrition is viable as a causal factor in only three nations post-1950. Even in those nations, it has merely escaped falsification.
You could maintain that pre-1950 Flynn gains were driven by nutrition; the data's probably not good enough to rule it out.  But then I could just restate my puzzle by comparing earlier half-centuries to the 1950-2000 era.

My whole case admittedly rests on my personal impression of the intelligence of top thinkers, past and present.  But my stance here is hardly eccentric.  You're free to flatly insist that the sages of the past are, by modern standards, mediocre minds.  But can you really bite that bullet in good conscience? 

* Even that comparison makes the past look quite good, if you remember that, due to illiteracy, the vast majority of pre-modern human beings would have been unable to leave any lasting proof of their intellectual prowess.

Comments and Sharing

COMMENTS (21 to date)
Nathan Taylor writes:

One possible answer to this puzzle is people like John Von Neumann are/were in fact smarter than Gauss.

But now that we are standing on the shoulders of giants, so the incremental amount of new discovery to climb higher possible for a human mind during a single lifetime is less because the low hanging intellectual fruit has already been picked.

Put another way, Gauss would be a genius in today's era, but the amount of discovery a reborn Gauss would make today would not turn him into a household name. Living greatest mathematicians like Terrence Tao, Andrew Wiles and Grigori Perelman are doing more than reborn Gauss could today, and are smarter than Gauss was.

I'm unsure if this is true.

But I would say it more likely than any other explanation. An analogy here is sports. This far easier to compare across eras, and nobody would doubt today's greatest athletes would crush people from past centuries. The analogy from sports to cognition is reasonable, if not proven.

Josh writes:

Broadly agree with this analysis but do you think we should make any correction for what i’ll call the “low hanging fruit effect” i.e supply and demand, the marginal revolution, the normal distribution etc can only be discovered once and therefore it’s harder for modern humans to be considered great original thinkers? I think we see something like this effect with 1960’s music where they were first to do so many things and modern bands look quite derivative by comparison.

Henry writes:

You haven't considered the possibility that the top minds of today seem less impressive because there's less low-hanging intellectual fruit about. Perhaps there are thousands of people today who could have discovered calculus, relativity or the theory of comparative advantage had they been born earlier. But today, they're working in areas that are relatively less impressive or are a lot harder for one person to make a substantial impact in.

Henry writes:

(Ironically, my comment is now a minor example of this, as I was beaten to the punch by Josh).

Nathan Taylor writes:

One example supporting the "low hanging fruit's been picked" theory is Srinivasa Ramanujan. I would argue he was quite a bit smarter than Gauss or Newton. But impossible to know for sure.

Matt writes:

I would be curious about the impact of how the presence of lead in the environment (which has a measurable impact on IQ) contributes here.

Anon. writes:

In his book, Flynn argues that the gains are in a way "fake" (i.e. increases in IQ do not represent increases in g), but that they are still important because they represent an increase in the particular types of cognitive tasks involved in IQ tests. He writes:

"At one time, I was blind to the real-world significance Of IQ gains because I was under the spell of g. I kept looking for general intelligence gains and could not find them. I could not see the trees because I was looking for a forest.

Eventually, I came to see that piecemeal gains do not lose their real-world significance simply because there are not gains everywhere."

Matthew Moore writes:

The major IQ puzzle isn't the relatively modest increase over time per country, but the much larger variation between countries now. When I first saw the data, I literally could not believe it.

Occam's razor pushes me to look for the same cause for both. I don't necessarily rule out nutritional causes based on Flynn's analysis. Sure, simple calorie deficiency probably isn't to blame, but vitamin and mineral deficiencies were, I think, common in the upper classes of Europe. As Matt says, lead and other toxic exposures play a role.

Abe writes:

Which substantial writing of Gauss did you read? Which substantial writing of a great mathematician from the last 50 years did you read to compare it with?

Josh S writes:

To build on the "low hanging fruit" argument, I would also consider the general complexity of today's world. There is more information and previous results to build on, but building on it also requires an increasing amount of time and effort to understand what came before. I would expect fewer and fewer people over time to expend the necessary energy to get to the "frontier".

I also suspect that because of this effect, the impressive, boundary-pushing examples of brilliance have come increasingly from teams rather than individuals. One way to combat complexity with a relatively fixed brain size is to network them together. This too has diminishing returns due to transaction costs, but in the the short term at least will continue to bear fruit, especially when assisted by technology. A side effect of the team-based approach is that there often isn't a single individual "hero" who is raised to popular attention.

Floccina writes:

Couldn't it be that there is some maximum IQ which humans approach and more people in modern times are approaching it now rather than being more brilliant there is a higher % of the population way up there.

HH writes:

Ooh, lots of interesting possibilities here. Thoughts presented without endorsement:

1. Why higher average scores? The world is increasing in complexity (through growth and technology), so everyday life serves as a sort of IQ test; this gives us much more practice at IQ tests, causing the average score to rise, since practice leads to improvement. (Perhaps complexity requires practice at exactly the cognitive tasks the test measures).

2. Poverty/nutrition: I hesitate to trust the data here. Most likely the absolutely poor (and likely undernourished) were also never tested. That's still a problem with tests like PSA, skewing tests upward. That said, I do think the largest benefits of reduced poverty are already behind us.

3. Why would today's smartest people not be smarter than the past's? Might be a function of natural limits. Physically, there is only so strong we can get even with crazy nutrition & exercise (and steroids), and at those extremes ligaments tear and bones break. Perhaps the brains that would be more intelligent by IQ measures eventually trade off against other skills (social, conscientious, etc) such that even if they are more intelligent, they can't contribute as much to human knowledge (much like a super strong body with a torn ligament can't help its sports team much.

Daniel Klein writes:

Great post!

Matt Clancy writes:

I would add to those who make the low-hanging fruit arguments (and the greater complexity argument made by Josh S) that getting to the frontier requires a lot more specialization than in the past. Breakthroughs will be highly specialized and difficult for people outside the niche to sufficiently appreciate.

Andrew_FL writes:

There is no paradox between comparing an increasing average over time and an a flat extreme tail over time, as you do here. Consider life expectancy. The life expectancy of the whole population of countries like the US has risen by *decades* in the last hundred or so years. But the more you select for narrower longest lived subsets, the smaller that increase is. Someone who lived to 90 in 1900 wouldn't have much higher expectation of making it to 100 than someone who lived to be 90 in 2010.

DeservingPorcupine writes:
But I would say it more likely than any other explanation. An analogy here is sports. This far easier to compare across eras, and nobody would doubt today's greatest athletes would crush people from past centuries. The analogy from sports to cognition is reasonable, if not proven.

This is tempting reasoning, but I'm not so sure. I think the athletes from the past would be competitive today if they had access to the nutrition, pharmaceuticals, and training techniques from today. Jesse Owens plus modern training would probably be competitive with Usain Bolt.

The thing is we don't yet have "steroids for the brain". Known partial cognitive boosters like amphetamines aren't nearly as effective as steroids.

John Thacker writes:

Wouldn't a good way to test the nutrition idea be to link it to the similar phenomena of increasing height during the 20th century, especially since lack of nutrition during the first two decades is plausibly linked to stunting both height and IQ below what genes would dictate?

john hare writes:

Rephrasing what several people have pointed out about the high end. There was far less clutter historically. A 6'4" man is tall today when the average is near 6'. 6'4" was a giant when people averaged a bit over 5'. Highly intelligent people are similar in that there are far more smart people around that make them less remarkable. A 10' wave in a calm sea is remarkable. A 10' wave among 9' waves may not even be noticed.

TZ writes:

Genius is not purely intelligence, but intelligence + hard work + a relevant field + an active field + creativity.
It is possible that an increase in intelligence has occurred but the other factors have decreased:
- because of generalized prosperity there is less need to work so hard (and because the low fruits have been picked working hard for many years is necessary just to reach the active research fields)
- very smart people who want money go into finance or computer programming rather than science or philosophy. There were probably fewer employment options for the analytical minded before.
- not all fields move at the same pace, for example these days genetics is in a boom thanks to improvements in genome sequencing while theoretical physics is pretty much stagnating as no new data is coming unless new expensive equipment is created so budgetary constraints play a much bigger role now
- creativity may be correlated with unpleasant traits that get better treated now (or are being pathologized if you want to adopt a contrarian view). I don't put much stock in the tortured genius cliche, but I also don't doubt that Van Gogh would have benefited from access to psychologists and medication.

Brian writes:

"When I read the smartest thinkers from the 17th, 18th, and 19th centuries, they seem roughly as smart as the smartest thinkers from the 20th century...What 20th-century thinkers credibly exceed the sheer intellectual firepower of Plato, Aristotle, Hume, Kant, Descartes, or Gauss?"

This is an interesting comment that needs some analysis. On the one hand, I agree that famous ancient thinkers and writers come across as being very smart indeed. It's not hard to marvel at their brilliance. From that perspective, one might think they are just as smart as the smartest people today.

On the other hand, I distinctly remember having a different reaction when I was first took a philosophy course in high school and got a chance to read the writings (in translation) of famous philosophers, including the first 5 on the list. I was surprised not only at how easy it was to read their works, but also how easy it was to identify faulty assumptions and logical errors. Even as a high school senior, I was constantly thinking "this is not right because...." By contrast, reading well-known philosophers from the 20th century often leaves me scratching my head. The work being done seems much more complex, more abstract, and much more difficult to wrap one's mind around.

This phenomenon is not limited to philosophy. Compare calculus and Newton's laws of motion to string theory. The former are frequently learned by the average smart high school student; the latter is beyond many highly trained physicists and mathematicians.

This may be partly due to the low-hanging fruit issue raised by other commenters, but it also seems to be a result of the brightest thinkers being able to handle more complex manipulations of number and language than ever before. That looks to me like an example of modern thinkers being smarter.

Art writes:
3. Women in earlier centuries would have had little opportunity to impress the world with their intellects. So the past is playing with only half its team

Half its team?
Hardly matters at all at the extreme tail.

Comments for this entry have been closed
Return to top