Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Age, fluid intelligence, and intelligent posts

9 Post author: InquilineKea 20 February 2011 09:06AM

As this diagram shows, fluid intelligence generally declines with age after age 20 (this is an effect that is confirmed across multiple studies - you can find more if you google fluid intelligence + age online.

Yet, on the other hand, I've noticed that the most intelligent+aware posts almost always come from older people (there are exceptions, of course). Of course, intelligent posts depend more on crystallized intelligence rather than on fluid intelligence, and crystallized intelligence only grows with age (this is *especially* true for LessWrong users, since they are far more mentally engaged (in mid-life) than the groups that are probably tested on these metrics. But yet, it still takes a significant amount of mental effort to actually write intellectual posts, so I'm pretty sure that there's still some dependence on fluid intelligence.

In any case, I'd like to propose this question: Have you found it harder or easier to write thoughtful+intelligent posts as you've grown older? And how have changes in your brain and in knowledge affected the thoughtfulness of these posts. What do you think is your biggest constraint in making these posts? While I'm still very young compared to most people here, I've still noticed that the number of examples I can think of is the primary constraint to the amount I can write, and so I can expect myself to write better with time (since I'll learn more examples+have better ways of finding them), assuming no decline in fluid IQ.

Additionally, do old authors write any less well than younger authors? Especially authors who are very old? Jacques Barzun, in particular, wrote his last book at age 99. But it seems that most people have stopped producing by that age. It is said that Hans Bethe was the only nonagenarian physicist who produced a top-calibre publication in his 90s, for example. Of course, many of these effects are average, and beating the average is pretty easy if we're motivated enough to do it (simply because many others aren't motivated, and also because intellectual stimulation helps stave off brain shrinkage). But even in the end, beating the average may not be enough to prevent decline.

There's some more interesting literature on this in Dean Simonton's books (Psychology of Science in particular). Basically, they show that scientific productivity increase with age, but only up to a point, and then they start to decline. Some of these effects may be related to decreased motivation or decreased time, but it's also possible that some of them are related to decreased fluid intelligence (especially given that the declines come earlier for physical scientists than for social scientists)

Comments (50)

Comment author: gwern 21 February 2011 04:36:45AM *  6 points [-]

Have you found it harder or easier to write thoughtful+intelligent posts as you've grown older?...

As a very young person myself compared to some others (just 23), I can't say I have much insight into this issue. But I'll shed what light I may.

I can say that over the last 10 years (I've always been a heavy reader & writer), my technical skills haven't improved very much from my perspective. My prose style used to be more show-offy and complex than it is now; I'm not sure what lesson to draw from that. (It is not as if my WM or IQ plummeted when I turned 19; if anything, with n-backing, it is probably higher.)

What has improved is my ability to think up real-world examples, and ability to add citations. My essays used to be full of holes and occasional non sequiturs, but as I read more and more and the Internet became more useful, they started getting better and I began to be able to offer useful comments.

For example, I first emailed SL4 in October 2004 with a useless contribution, and most of my early emails are marginal comments I would neither up nor downvote if I saw them as comments here. And in another example, I thought up a really good insurance idea, wrote a little essay on it, and years later read of something identical; I had reinvented a standard form of insurance. These days, I think I am doing better; some things on gwern.net I am not even ashamed of.

Additionally, do old authors write any less well than younger authors? Especially authors who are very old? Jacques Barzun, in particular, wrote his last book at age 99. But it seems that most people have stopped producing by that age. It is said that Hans Bethe was the only nonagenarian physicist who produced a top-calibre publication in his 90s, for example. Of course, many of these effects are average, and beating the average is pretty easy if we're motivated enough to do it (simply because many others aren't motivated, and also because intellectual stimulation helps stave off brain shrinkage). But even in the end, beating the average may not be enough to prevent decline.

Need to break out what you mean by writing. Technically, I'd say writers usually only improve. Gene Wolfe's prose seems to be getting smoother and subtler by the decade.

Idea-wise, they can often go downhill. Take Gene Wolfe again. He's pretty prolific and only has grown in critical acclaim. But never the less, the books the fans and critics keep coming back to are some he wrote 30 or 40 years ago, The Book of the New Sun. The Book of the Long Sun and The Book of the Short Sun are its stylistic peers, but they get only a small fraction of the attention & love of New Sun. There's no obvious reason why they aren't more popular. They just aren't. Some magic is missing. (And on some respects, they are distinctly inferior: his world-building is growing outdated, his politics have grown increasingly obtrusive & shrilly conservative, and his puzzles obtuse and difficult to solve. On urth.net, the consensus is that we don't understand An Evil Guest or Pirate Freedom at all, and with The Sorcerer's House, the puzzles seem simplistic enough that we're not sure we're missing the forest for the trees.)

Or more obviously, there's Frank Herbert. Dune and its universe marches on, apparently untouched by time. But when was the last time you even saw any of his other novels like The Dosadi Experiment or Dragon in the Sea discussed? Only fans read those, and I can attest that even we don't read them often or deeply; as well, there seems to be a real trend toward them getting worse.

Comment author: InquilineKea 21 February 2011 05:25:31AM 1 point [-]

Oh, interesting points. Yeah, it's harder to create new ideas as you get older. So maybe it's possible that social scientists can still write very good papers in their 90s, because they can still get their ideas from other people.

Comment author: Vladimir_M 20 February 2011 07:46:54PM *  4 points [-]

One interesting paper about this topic I once skimmed is T.A. Salthouse, "When does age-related cognitive decline begin?" (Ungated link here.) The paper presents some figures similar to the graph in this post, with roughly similar magnitudes. (Maybe it is in fact among the sources of data?)

This is assuming the z-score in the graph is reported relative to the distribution of the scores over the entire (all-ages) sample. Which is pretty scary, when you think about it.

Comment author: Perplexed 20 February 2011 08:26:51PM *  4 points [-]

The Salthouse paper states that the Z-scores there are relative to the entire sample (ages 20 - 60). But I'm pretty sure that the OP's graph uses Z-scores relative to the variation within an age cohort. Otherwise, they would be too steep to be believable - almost mathematically impossible considering that the sample runs out to age 80 and shows a drop of about 2.0 Z units

ETA: The Flynn effect may account for about a third of the apparent drop in cognitive ability - that is, today's old-folks started off less cognitively skilled than today's youngsters, so their decline with age is not as steep as it looks.

Comment author: Vladimir_M 20 February 2011 09:01:51PM *  2 points [-]

Perplexed:

The Salthouse paper states that the Z-scores there are relative to the entire sample (ages 20 - 60). But I'm pretty sure that the OP's graph uses Z-scores relative to the variation within an age cohort.

The poster should definitely clarify that. The current presentation is unclear.

Otherwise, they would be too steep to be believable - almost mathematically impossible considering that the sample runs out to age 80 and shows a drop of about 2.0 Z units

I don't find it that implausible. Anecdotally, I've met people whose cognitive skills appeared horribly diminished already in their seventies. (Don't forget that the median age of death in the developed world is roughly in the mid-seventies for men and late seventies for women.)

Even if cognitive decline can be greatly ameliorated by persistent intellectual effort, most of the population clearly won't benefit from this, which is good to have in mind if one has seen an anecdotal sample of exceptionally intellectual people.

ETA: The Flynn effect may account for about a third of the apparent drop in cognitive ability - that is, today's old-folks started off less cognitively skilled than today's youngsters, so their decline with age is not as steep as it looks.

As in all other things related to measuring cognitive ability, the Flynn effect tends to make a confused mess of the whole situation just when one starts to come up with neat and plausible theories. It's very hard to tell how and to what extent it has bearing on these results.

Comment author: jsalvatier 20 February 2011 04:59:06PM 4 points [-]

This is terrific graph. Do you have a source? I had no idea the relationship was so strong.

Comment author: RolfAndreassen 20 February 2011 06:00:14PM 7 points [-]

And you still don't, unless you are well read and know what a difference of 2.4 in Z-score means. Are the Z-scores similar to IQ points, in which case 2.4 is not very much, or are they standard deviations, in which case 2.4 is a devastating and horrifying loss? Or something else entirely?

Comment author: Unnamed 20 February 2011 07:44:08PM 4 points [-]

z-scores count standard deviations. So if your fluid intelligence is one standard deviation better than the typical person your age, then when you're in your 50s you'll be about as good as a typical person in their 20s.

Comment author: Vladimir_M 20 February 2011 07:49:50PM *  3 points [-]

It seems like the horrifying interpretation is correct. (See my other comment where I cite a paper providing more data.)

Comment author: Perplexed 20 February 2011 04:41:29PM 4 points [-]

Have you found it harder or easier to write thoughtful+intelligent posts as you've grown older?

I'm 63. I think I'm a better writer now - mostly because I have learned to make my writing less dense. As a thinker, I'm pretty sure I have lost some ability to think deeply - to understand new and complex ideas. But in compensation, I have a better ability to evaluate broadly - to quickly see the various implications of simple ideas in a complex context.

I'm not sure I was ever very good at producing long works - more than 5 paragraphs, say. But I do notice more now that I don't really have the patience for finishing big projects. Or for suffering fools gladly.

Comment author: NancyLebovitz 21 February 2011 01:47:35AM 1 point [-]

Could you expand on how you've made your writing less dense?

I could probably use more redundancy/roughage myself.

Comment author: Perplexed 21 February 2011 02:10:10AM 6 points [-]

I use short sentences sometimes. I'm not afraid to repeat myself. And sometimes I deliberately write in a country-boy/Will Rodgers conversational voice.

When watching movies, watch for scenes where someone is being interrogated, and they find it necessary to step back from providing a direct answer to a question and instead insert a short anecdote to provide the context for the pithy answer to the original question. Notice how the transition from direct response to anecdote and then back to response is signaled by a change in body language, tone, and even formality of language.

You need to provide similar signals when writing. And since you can't do your signaling using body language, you need to do it by wasting some words. Go ahead and "waste" them. Words are cheap.

Comment author: JenniferRM 21 February 2011 03:31:50AM 5 points [-]

direct response to anecdote and then back to response

That was the structure your post used! ...which was an awesome example of a coherent meshing of form and content, where the claims were consistent with the exhibited behavior.

From my mid-twenties to my early thirties while writing in online discussion forums I've noticed something like content-behavior coherence increasing but I've never really thought about it explicitly or hypothesized mechanisms.

My impression is that my writing has increased in quality but decreased in flexibility? If I've determined for myself the correct answer to something based on previously existing evidence I'm likely to be acting in a manner consistent with my beliefs and can simply re-verify the state of evidence on the subject and its connection to my conclusions and then (assuming the answer hasn't changed) repeat the earlier sentiment with more attention to communication. The content-behavior coherence might fall out of this, but I still struggle with verbosity...

Hey! This suggests an improvement strategy: find a forum for writers talking about the craft of writing, and experiment with asking questions and later with giving advice about writing until I can offer good advice that has "behavior-content coherence" :-)

Comment author: NancyLebovitz 20 February 2011 12:52:29PM 4 points [-]

First whack at the question: I think my ability to write comments-- I've never done a lot of posts-- is at least as good as it was 30 years ago. I'm 57.

I get approximately the same amount and kind of respect.

The only change I see is that I think I've got a more emotional range in what I write.

Comment author: FAWS 20 February 2011 01:19:04PM 2 points [-]

I'm curious. Were you on Usenet in the really early days, or did you often write letters to the editor, or participate in fanzines, or what are you comparing with?

Comment author: NancyLebovitz 20 February 2011 01:25:11PM 3 points [-]

Usenet in the early nineties (not the really early days) and apas for at least a decade before that.

Comment author: FAWS 20 February 2011 01:38:11PM 1 point [-]

Ah, interesting. I was aware of fanzines, but not that they were just part of a more general phenomenon.

Comment author: JenniferRM 21 February 2011 04:01:23AM 3 points [-]

This made me thing of two topically related claims:

  • I've heard (for example here) that verbal prodigies are exceptionally rare compared to mathematical and musical prodigies. A novel proof can be delivered at 16, but to write something really amazing you have to have something to say first and that generally requires experience and reflection.

  • In Human performance, psychometry, and baseball statistics Craig Heldreth talked about the age of peak performance for different sports and professions, and noted (without a citation I could track easily) that business executives peak at about age 60. This suggested to me that instrumental rationality takes a long time to peak and is likely to be more related to crystallized intelligence than fluid intelligence.

Comment author: khafra 23 February 2011 12:06:48AM 4 points [-]

A cynical mind might take this as evidence for the proposition that “it’s not what you know, it’s who you know” applies strongly to occupants of the C-suite, as it takes a long time to build solid relationships with lots of important people.

Comment author: CarlShulman 20 February 2011 05:12:41PM 3 points [-]

I've seen unpublished work showing fluid intelligence also declines more rapidly for those with higher initial scores.

Comment author: Oligopsony 22 February 2011 01:49:50PM 3 points [-]

Did it address whether that might be due to testing error?

Comment author: CarlShulman 22 February 2011 03:32:46PM 3 points [-]

I didn't see a control for regression to the mean as such, but it was sustained over time. Regression to the mean on a population selected for high initial scores should take place primarily at the second testing.

Comment author: Oligopsony 22 February 2011 06:31:34PM 1 point [-]

This makes sense.

I wonder how much might be explained by "error-like" environmental factors that persist for medium-range amounts of time - working in a cognitively demanding workplace, for instance, or a living situation amenable to regular sleep and exercise. The archetypal error term, after all, is (beyond the finite sample size of the questions) a product of environmental effects that last a day or so - a headcold, eating a healthy or unhealthy or no breakfast, receiving a compliment or insult earlier in the day, and so on. (We'd expect to see this same regression to the mean on the left side of the distribution, although not necessariy at the same level if test-retest covariance itself varies with score.)

Or it could indicate something about the "true distribution" of "absolute intelligence;" the scale we have, after all, is normal by definition, such that we have no reason to believe that differences between 100 and 115 and between 115 and 130 correspond to the same difference in absolute ability (however one might imagine that.) If we presuppose that everyone experiences the same absolute level of cognitive decline (plus an error term), then a greater fall in sigmas among the highest testers might imply a diminishing returns to numerical IQ. (The same effect might or might not occur on the left side of the distribution.)

Comment author: gwern 03 March 2011 01:46:00AM *  2 points [-]

If anyone is curious about the source of the image, http://gizmodo.com/#!5495086/this-is-your-faulty-brain-on-a-microchip says:

"Lead image adapted from Park, DC; Lautenschlager, G; Hedden, T; Davidson, NS; Smith, AD; Smith, PK. (2002). Models of visuospatial and verbal memory across the adult life span. Psychology and Aging, 17: 299-320"

Comment author: gwern 22 March 2012 06:18:56PM 2 points [-]

If anyone is still interested in the topic, I've been compiling citations of interest to me in http://www.gwern.net/DNB%20FAQ#aging

Comment author: Desrtopa 24 February 2011 04:25:13PM 2 points [-]

There's a hypothesis that some thinkers start out with a high level of productivity and skill at a young age, and gradually decline from there, and some start out at a lower level of skill and productivity, and gradually improve over time, before hitting a decline at very old age. I believe Wikipedia has an article on it, but I can no longer remember what it's called. If anyone can remember, I believe the page discusses the body of evidence for it.

Comment author: gwern 26 February 2011 09:30:39PM *  1 point [-]

I finally remembered where I read about that kind of dichotomy - it was a Malcolm Gladwell article which does indeed mention Monet but only in the context of using Cezanne as an example of the gradually improving type: http://www.gladwell.com/2008/2008_10_20_a_latebloomers.html

Comment author: gwern 25 February 2011 02:18:41AM 1 point [-]

For some reason, that idea triggers an association with the Impressionist painter Monet.

Comment author: komponisto 21 February 2011 05:47:37AM 2 points [-]

I haven't found an exactly appropriate segue in the post or comments (which are too specific about domains and don't mention this one), but this post nevertheless seems like an appropriate context to mention that composers appear to be a fairly strong counterexample to the conventional wisdom about creativity declining with age: they not only tend to be most productive, but also usually do their best (and even sometimes most innovative and radical) work, in late age, whether "late age" means 30, 50, or 90.

That phenomenon so dreaded in mathematics, namely doing some brilliant piece of work in one's twenties and never doing anything significant again despite trying, just doesn't seem to exist in music.

Currently I know not the causes nor the implications of this difference.

Comment author: Nornagest 21 February 2011 06:31:21AM 1 point [-]

That's a bit of a surprise, considering the oft-cited links between mathematics and music and the well-publicized math prodigies out there. One possible explanation might be that an apparent prodigy in math is something different from a prodigy in music; that fluid intelligence lets you pick up the technical skills to be writing symphonies as a teenager, but that writing engaging and expressive music requires an additional, different skillset that's more reliant on crystallized intelligence.

I'm not a classical music buff, but I've heard that Mozart's childhood work, for example, wasn't very good. Certainly not in comparison with his later stuff.

Comment author: TheOtherDave 20 February 2011 02:31:14PM 2 points [-]

At 42, I keep learning new things about how to communicate clearly. I seem to be better at it than I was at 20, though I haven't systematically tested.

Comment author: Alicorn 20 February 2011 04:03:55PM 3 points [-]

I think 42 is probably a special case ;)

Comment author: David_Gerard 20 February 2011 09:38:47AM *  2 points [-]

I've observed and now experience the phenomenon of discovering practical politics in middle age - in your forties, you've learnt something from twenty years' chimpanzee tribal interaction and your brain still works a bit, so you can more usefully understand and apply just how much everything is politics.

Whether it's improved my writing is another matter. I'm better at sentences, but I have less motivation to write something that's worth reading in the first place.

Comment author: knb 21 February 2011 06:14:07AM 1 point [-]

Bob Dylan has talked about the decline in his creative skills as he aged. Some years ago, I saw an interview with him, where he said that he can't understand how he was ever able to write songs as good as his early work. To him, it was like a different person wrote the music.

Comment author: CronoDAS 20 February 2011 07:00:26PM 1 point [-]

I haven't seen too many writers get worse as they get older. Piers Anthony's writing has gotten consistently worse over time, in my opinion, but I can't think of any other specific examples.

Comment author: Eliezer_Yudkowsky 21 February 2011 10:16:50PM 9 points [-]

I think of this as Senior Author Syndrome. Chief exhibits are Anne McCaffrey, Mercedes Lackey and Orson Scott Card. The first symptom is forgetting how to hurt your characters. Just pick up a late book and an early book and compare how much jeopardy the characters get into.

Comment author: orthonormal 27 February 2011 03:26:34AM 1 point [-]

The bizarre exception to this is Douglas Adams' Mostly Harmless.

Comment author: JGWeissman 27 February 2011 04:36:16AM 2 points [-]

I recall reading that Douglas Adams was depressed when he wrote Mostly Harmless, and intended to fix the disasters in the next book (which is totally possible in the HGTTG insane joke logic), but he died before he could write the next book.

Comment author: Risto_Saarelma 21 February 2011 03:49:39PM *  4 points [-]

It seems like most hard SF writers get hit by this. Frederik Pohl's stuff from the late 80s onward has been pretty dismal, Larry Niven still writes stuff and is still the guy who wrote those good books in the 70s, Isaac Asimov's and Arthur C. Clarke's late stuff were weird messes. Robert L. Forward went from Dragon's Egg to something called Saturn Rukh which sounds like a book length deadpan parody of terrible hard SF. Based on the reactions to his recent books, Greg Egan might also be on the way to becoming a cranky old man who wrote those good books in the 1990s.

This might be tied to how hard SF is about fresh, new perspectives and fluid thinking upending obsolete tradition. The successful young SF writers are ones with lots of fluid intelligence and little care for crystallized tradition. These aren't necessarily the sort of people who develop extraordinary insight in the domain of crystallized nuance when their fluid intelligence starts to wane.

Writers who never really hit it big might need to keep cranking out the same sort of books they got famous with decades ago, even if they are tired with them and just going through the motions, since they're stuck in the SF ghetto and have no name recognition outside it. Clarke and Asimov probably could have written anything they wanted and gotten it published, but still wrote tired sequels to their old hits. I understand Heinlein did start writing anything he wanted, and the results weren't quite high literature.

Comment author: CronoDAS 23 February 2011 01:20:36AM 4 points [-]

I have to object here: Asimov did indeed write "anything he wanted" and got it published. He's got a grand total of 515 books to his name. In addition to the science fiction novels and anthologies he wrote and edited, his books include include such things as two non-SF mystery novels, lots and lots of "popular science" nonfiction, five books of limericks, and the two-volume Asimov's Guide to Shakespeare.

Additionally, according to his autobiography, the driving force behind Asimov returning to his old science fiction themes was his publishers' insistence. So blame them. ;)

Comment author: NancyLebovitz 23 February 2011 05:52:59AM 1 point [-]

As I recall, Asimov said in an essay written near the end of his life that he regretted not writing historical fiction.

Comment author: Risto_Saarelma 23 February 2011 05:49:15AM 0 points [-]

Fair enough. Still, as far as I know, no literary fiction of note. Does he have any fiction as well-received as the Foundation trilogy or The Caves of Steel from the mid-60s onward?

Comment author: NancyLebovitz 20 February 2011 08:28:17PM 2 points [-]

Heinlein's later novels are generally considered to be inferior to his earlier work, but he had brain-related health problems.

CS Lewis suddenly lost the vivid visual images which he used as the starting points for fiction.

Comment author: Alicorn 20 February 2011 07:25:19PM 3 points [-]

I think Orson Scott Card has deteriorated.

Comment author: Sniffnoy 20 February 2011 09:21:31PM 1 point [-]

Niven, too.

Comment author: James_Miller 20 February 2011 09:41:40PM 1 point [-]

And Tom Clancy

Comment author: LucasSloan 23 February 2011 10:44:54PM 0 points [-]

Lois McMaster Bujold. Harry Turtledove.

Comment author: wedrifid 02 March 2011 09:32:54PM *  1 point [-]

I haven't seen too many writers get worse as they get older

David Eddings.

Isobelle Carmody (although this case may be limited to the Obernewtyn world in particular - I haven't read other recent works).

A notable non-exception is Terry Pratchett. His most recent works are still superior to his earliest despite Alzheimer's.

Comment author: InquilineKea 21 February 2011 02:02:47AM 0 points [-]

Thanks for all the interesting responses, everyone!

Here's another question: Do you find that academic books and papers take longer to read as you grow older? Or not at all?

Comment author: ThomasR 22 February 2011 04:16:56PM 1 point [-]

That is hard to estimate, but I think I need the same or less time for studying them. But of course the issue is how one reads them and how much one spends into extracting the hidden ideas and translating them into one's own mental structures (instead of turning one's mind into an emulation of the author's one) . One can study and understand very advanced papers fast and well without a productive "translation", and the more one knows, the easier it is to restrict reading that way.