Short answer: YES, it is surprising, but it is true.
Based on existing answers and comments, there seems to be some confusion, so let's start by clarifying:
- This question is not about whether we were smarter/faster in Victorian times. It is not about 120 year old data.
- This question is not about how quickly people answered questions. It is how quickly they respond to a stimulus. (But, how quickly people make choices between options is a closely-related area of research.)
- This question is not about causality, it is about correlation.
- This question is not about what you expect might be the answer, based on your conjectures. It, like every question on this site, is what the literature shows based on empirical data.
So what does the literature say?
Wikipedia gives us a quick overview:
Researchers have reported medium-sized correlations between reaction time and measures of intelligence: There is thus a tendency for individuals with higher IQ to be faster on reaction time tests.
Research into this link between mental speed and general intelligence (perhaps first proposed by Charles Spearman) was re-popularised by Arthur Jensen, and the "Choice reaction Apparatus" associated with his name became a common standard tool in reaction time-IQ research.
The strength of the RT-IQ association is a subject of research. Several studies have reported association between simple reaction time and intelligence of around (r=-.31), with a tendency for larger associations between choice reaction time and intelligence (r=-.49).
Seems astonishing to you? Jensen (who was credited with repopularising the idea by Wikipedia, above) agrees:
It seems almost incredible that individual differences in reaction time (RT) in simple tasks that involve no intellectual content and are so easy as to be performed by most persons in less than 1 s should be correlated with scores on nonspeeded, complex tests of reasoning ability, vocabulary, and general knowledge - the kinds of content that compose IQ tests. Nevertheless, in recent years, the correlation between Rt and IQ has become an empirically well established fact, based on thousands of subjects in scores of studies conducted in many laboratories around the world.
In this this next article, he discusses how the idea was counter-intuitive and initially rejected:
Jensen has done some experiments:
The abstract is a bit light on (and I haven't read the paywalled article), but claims:
[Reaction Time] and [Movement Time] show reliable individual differences which are significantly correlated with intelligence as measured by Raven's Standard Progressive Matrices.
Moving away from Jensen, do we have anyone else making the claim? Here's the reference was used by Wikipedia.
The association between reaction times and psychometric intelligence test scores is a major plank of the information-processing approach to mental ability differences.
They measured the correlation amongst 900 56-year-old Scottish people between reaction time and AH4, a type of intelligence ("general mental ability") test.
Note: Quick stats lesson recap: Correlation coefficients range from 1 (total positive correlation) to 0 (no correlation) to -1 (total negative correlation). If there is a correlation between higher mental ability and shorter reaction time, we should expect the r-values to be negative.
AH4 Part I total scores correlated -.31 with simple reaction time, -.49 with four-choice reaction time, and -.26 with intraindividual variability in both reaction time procedures. [...] Separate analyses were conducted after partitioning the total group according to sex, educational level, social class grouping, and number of errors on the four-choice reaction time task. None of these factors significantly altered the effect sizes.
Two of the same co-authors from the previous paper, builds on the correlation between psychometric intelligence and reaction times, and the unexplained correlation between intelligence and longer life, and suggests, with (presumably) the same sample of people, it is actually the faster reaction times that are the important factor.
As an aside: Another related concept is Inspection Time:
the exposure duration required for a human subject to reliably identify a simple stimulus
A meta-analysis found that it also correlated with IQ.
As to the elite sports-player conjecture, it seems to contain a number of assumptions that are not necessarily true:
- Most importantly, the correlation is moderate, and not total. Plenty of intelligent people may have slower than normal reaction times. Plenty of unintelligent people may have faster than normal reaction times.
- Many sports-players do succeed in academic endeavours.
- It isn't safe to assume that elite sports-players would want to enter academia, even if their abilities allowed it.