Comments on: Richard Feynman’s “Low” IQ

By: Jose

Jose — Fri, 28 Nov 2014 10:53:12 +0000

Feynman always tried to give unusually smart answers on these tests on purpose, and these answers were usually disregarded.

By: Jordan

Jordan — Sun, 07 Sep 2014 22:02:31 +0000

He was an Empiricist. Empiricists rely on heuristics, not upfront type problem solving. Some people suggest that the iq test he took had a low ceiling (though even back then a ceiling of at least 140 was likely), and thus could not reflect his true score, but from listening to him talk, I can’t see his “IQ” being more than 120-130. The relationship between Heuristics and ‘intelligence’ (note, not iq), is something that are only recently exploring.

By: William Newman

William Newman — Wed, 03 Jul 2013 15:53:41 +0000

Dr. Bilbo, could you give some more detail on how to proceed from your link to the 1939 exam? I see a hyperlink on your linked page which leads to “problems, solutions, winners, and scores from 1985-present” but it’s not obvious to me how to get to older exams.

By: Dr. Bilbo

Dr. Bilbo — Wed, 03 Jul 2013 05:08:05 +0000

Actually, the Putnam exam was far from “notoriously difficult” in 1939 (the year that Feynman won); it was a straightforward computational exam which can be proved by obtaining the actual exam (link). By the way, this was only the 2nd Putnam exam held and the team results were as follows:

1. Brooklyn College
2. MIT
3. Mississippi Women’s College

Pretty heady company, eh?

By the way, where did you get the info that Feynman won this exam by a large margin?

By: Romeo

Romeo — Wed, 10 Apr 2013 22:40:34 +0000

Feynman sucked at humanities. He admitted it himself. He was a one-sided scientist.

By: Bruce Charlton

Bruce Charlton — Tue, 10 Jan 2012 06:02:26 +0000

I have never found it at all hard to understand why people get low scores in specific exams. Surely there are 101 reasons why an individual on a specific day will under-perform (illness — e.g. headache, hay fever, sleepiness — being the most obvious). The longer and more fatiguing the exam, the greater the under-performance. But — while random factors could make a significant difference to a short test with few items — it would be very unusual to over-perform on a long and multi-item test (like a WISC or WAIS), so the situation is not symmetrical. This also means that it is not valid to average IQ test scores. The most valid way to deal with variation in tests would be to accept the highest score achieved in the longest and most rigorous valuation — but to use only these scores in developing population norms. (I.e. a random population sample would do repeated long, multi-item IQ tests — and the highest value each individual attained would be used to set the population norms.) The big problem of IQ testing is the population norms. It is extraordinarily difficult to get these precisely correct. So, while testing can be used to rank a group very effectively — but not at the individual and specific level, for reasons above — hair-splitting about specific IQ scores (especially distant from the mean) in terms of population percentages is nonsense. In sum, differential IQ measurements are valid but intrinsically not precise.