Eminent scientist and intelligence blogger Bruce Charlton once blogged about converting reaction times into IQ score equivalents, and now that blogger alcoholicwisdom has told this blog about a really user friendly reaction time test at humanbenchmark.com, it’s time to explore this idea further.
PART 1: TECHNICAL DETAILS (SKIP TO THE PART 2)
The average reaction time of those who take that test is 253 ms. In order to equate that to IQ, we need to know the average IQ of people who take that test. Well we can estimate the average IQ to be 117 because many years ago, a company called Braintainment sold a complex series of reaction time tests called “Thinkfast” and customers on their website had an average IQ of 117 based on a five-minute IQ test. Even though Thinkfast customers are not the same as humanbenchmark.com visitors, it’s reasonable to assume that both populations have the same IQ, since they’re both self-selected internet users with an interest in reaction time.
Thus 253 ms = 117 IQ
Some might argue that you can’t just equate the two figures because there’s no reason to assume chronometric players will be equally self-selected for reaction time as they are for IQ. But what they’re really self-selected for is intellectual interests and the internet access to pursue them. Since this self-selection is g loaded, and since both reaction time and a 5 minute IQ test are probably both only moderately g loaded, it seems reasonable to equate them. In fact, this is a well known psychometric technique called equipercentile equating, and was used to norm the famous Mega test using SAT scores (see section 8.4.1 of the Prometheus MC Report for more details).
The next thing we need to do is equate the standard deviation for simple reaction time with the standard deviation for IQ. Scholar Michael Woodley and his colleagues have argued that the population standard deviation for simple reaction time in Western countries is 160.4 ms, a truly colossal figure. However scholar Arthur Jensen has shown that when you measure the mean reaction time repeatedly in college students (each one playing 20 times and taking the average simple reaction time for each student), most of the variability between individuals gets cancelled out and the standard deviation of the averaged out reaction times is only 29.23. The standard deviation for IQ in the general U.S. population is 15.
Thus, when taking your average reaction time from 20 tries:
29.23 ms = 15 IQ points
Some people might object that college students have a restricted range of reaction times because they’re selected for general intelligence. However because the correlation between reaction time and education is weak, and because even among college students, there is enormous range in education (some drop out their first year, others go on to get PhDs) it’s likely that college students are not a chronometrically range restricted population.
It should be noted that Bruce Charlton believes one’s best reaction time scores are more meaningful than one’s average scores, however we do not have stats for best performance, so average performance will have to do for now.
So now that we have these two data points, we can convert reaction time to the IQ scale.
PART 2: TAKE THE HUMANBENCHMARK.COM TEST & CONVERT SCORE TO IQ SCALE
Step 1: Make sure you are on a desktop or laptop computer with a clickable mouse (mobile devices give much slower reaction times)
Step 2: Go to humanbenchmark.com’s reaction time test
Step 3: Play 8 warm-up tries to get practice. When you reach “Trie: 5 of 5”, it will ask you if you want to save your scores. Just ignore that question and keep playing until you reach “Trie: 8 of 5” (your 8th warm-up). You will notice your average reaction time being calculated on the screen. Ignore this because you are just doing warm-ups so your scores don’t count.
Step 4: Revisit the humanbenchmark.com reaction time test on a fresh screen so that you can get a new running average. Now play 20 consecutive tries and record the average reaction time it calculates from all 20 tries. This will be your score.
Step 5: Go to the poll below to record your reaction time and observe the IQ equivalent. Don’t be depressed if your reaction time is terrible. Simple reaction time is just a crude physiological proxy for intelligence, much like brain size so I expect brilliant people to regress enormously to the mean and for many gifted people to have slow reaction times. Just as there were Geniuses like Einstein who had small brains, there will be Geniuses who have slow reaction times. Also, reaction time slows precipitously with age. If you’re older than the average person who visits humanbenchmark.com, your IQ equivalent will be further depressed (severely so).
NOTE: The IQ conversions I discuss are just my own personal conjecture and have not been endorsed by humanbenchmark.com (a website I have no connection to)
What about the issue of varying Internet speeds or CPU speeds? Or people clicking on an old mouse vs. tapping on a shiny new iPad? I think it’s meaningless to compare these results, because each user has a somewhat different computer setup.
i hope this isnt correct im african and i was teased about being black and the kids said we have small brains. at first i blew it off, but days pasts and he presist eventually i found all these pages on race, intellegence and brain size… i read up on the information to find out about the depressing iq gap… i noticed the differeces in the shapes between skull and brains (esp blacks) any the evidence points to be true.
this test didnt help i got a reaction time of 323. and my girlfriend agreed to do it (who is dark skin black) got an average of 353… meaning our IQs’s would be around 75 and (both fitting decently into “black statistical norm”…can anyone help me with more research or other possible reasons?
mike nolen said:
I wouldn’t put too much validity in this test as a good measure for IQ. Although it can correlate, I would suggest you take either a full proper online IQ test or find a center near you to get more accurate results. Also don’t let potential genetic differences bog you down too much, find something that you excel at and focus on it. Conscientiousness (work ethic and persistence) is second in line to IQ as an indicator of long term success.
Out of curiosity I just tried it, and on my iPad the reaction times were about 50ms *longer* than with a laptop & wireless mouse. I wonder if a mouse attached with a cable would also affect results.
Also, IQ scores are based on one’s age-peer group, so shouldn’t that also be the case for reaction time tests? Of course a 20-year-old will be faster than a 70-year-old, but that doesn’t mean his/her IQ is higher, and your test/survey doesn’t seem to take age into account.
Yeah my RTs are about 15 IQ points slower on an iPad & slower still on an iPhone which is why I explicitly said to use a mouse not a mobile devise. As for different computer speeds, that could create error.
As I said RT does drop precipitously with age but as long as the average person here is roughly as old as the average person who visits humanbenchmark.com, the aggregate norms should be comparable even though IQ’s for individual older people will be way too low
Smartphone is much better than mouse.
I mark on wrong category obviously because my lack of attention. First time i get 117 on iq. Second, 148. Third, 152. Fourth, 162, 7 times in this ‘last’. I will continue take this test many other times to have better average, but i can’t seem me as “the guy with iq above 130”.
I’m 26 old.
But it’s your average score out of 20 attempts that counts
Yes, i know. But never imagine i could have iq higher than 130. Only my verbal, not in english of cours. I do again two batteries, 20 times each one and take 130 and 180 ( 😐 ). But, i will continue thinking iq is part of intelligence, is important but isn’t the entire concept of intelligence.
grey enlightenment said:
interesting idea but a lot of factors , including even the height of the participant. A taller person may take a few more MS to register the click by having longer arms
Actually taller people have faster RTs though they probably do well in spite of their height, not because of it.
It depends how we define best scores. I think best average score might be an ideal measure, though norming it in a fair way would be problematic
What’s difference bettween this reaction time test and the other test, of one of the last post??
On this test i was very good but on other test my average was 0,7. Why?
What these tests are better?
In the other one you had to click in a narrow part of the screen which made it harder & more dependent on vision & hand-eye coordination. This one seems like a better measure of pure mental speed
Really have hard time believing this reaction time related to intelligence in term of complicated problem solving skill.
Trick to get reaction time fast: Use your peripheral vision for the test. It is animal nature to react instinctly to peripheral stimuli for alarm.
Also peripheral vision actually goes to different part of brain for more basic intint pathway which often involves less relays (synapses). Thus it involves less stations and get faster response.
You all know knee jerk response do not involve brain (spinal cord only). Knee jerking response is way faster than any decision involving cortex.
Reaction Time is related to complex problem solving because complex problem solving requires holding a lot of information in working memory, but that information decays rapidly so you need the speed to process it before it does so.
But don’t base too much on this one test. It only measures simple RT; complex RT is more g loaded, and RT consistency is more g loaded than RT speed.
Is your eye color blue?
No; why do you ask?
Got 350 ms with mobile phone.
Around 200-230 ms on computer using mouse. Had to motivate myself to do it. Seems rather odd to use sensory information and muscles to measure IQ.
It is proven: mobile technology makes you dumber.
I have a math degree yet my score on the RT test says I have an IQ of approximately 50.
This is pure bullshit. I want my 400ms back.
All I know is I entered this little challenge thinking the correlation with IQ be groundless, and ended up by realizing that, in my case, the expected IQ was precisely correct :))
while i do not necessarily agree with the correlations between iq and reaction time, i think it is certainly one metric by which we can measure cognitive ability, since it depends on the time it takes for people’s neurons to fire and send signals tothe eye, to the brain, and finally, to the finger.
I’m 37. With a rather bad (10 year old) mouse my fastest click was 66ms and my average was 179ms but i’m really sure that i can easily do better if i try my best. In conclusion…. I have doubts about the accuracy of this test.
That is why you should average multiple attempts. As 66ms is luck, and when you get one result under 100ms.. you average of 5 will be more than you can actually pull off consistently.
John Ortiz said:
My two best reaction times were 119 and 124. I’m 47 years old by the way.
I did this and got some supposedly impossibly low reaction times of e.g. 21, 33, 35 ms. Did anyone else get any this low? And I mean that those were three out of five of my scores on one trial, not out of several of them. Just curious.
There are a couple issues with your post. Despite the intellectual implication in its name, humanbenchmark is a site mostly frequented by gamers (popularized due to the importance of reaction time in shooters and video games in general). You can verify this by searching for forums referencing the site through Google’s search and also just by reading the comments on the site in general; the great majority are gaming related.
And even if the population that frequented it was really composed of people interested in their results for intellectual reasons, it seems fairly obvious that just by the site being free to use it would be unwise to assume that the IQ results of its users would be equal to those of a similar but paid site, as someone who is willing to pay for such test would most likely be highly interested in the subject and would care about doing it in an accurate way, while someone doing a free test is a lot less likely to be that invested in the subject and might just have stumbled upon the site through other means and chooses to give it a try as they have nothing to lose anyway.
For higher accuracy in your scale, I believe it would be better to use a lower IQ as the starting point and to use older data from the website to calculate the average for desktop computers, as nowadays there is a much higher diversity of devices around with a gigantic variance in latency, making it very difficult to use the current average in a meaningful way. For example, since this post was made the average score increased by around 30ms on HB (from 253ms in 2014 to 284ms in 2018), most likely due to the release of windows10 and touchscreen devices becoming even more common.
i.imgur.com/cI8GiIU.png this is how the average has changed over the years. Back in 2007-2010, windows XP and CRT monitors dominated the computer market, which overall lead to much lower results. I’d argue it is safe to assume that in a completely latency corrected environment, the true average of the scores on HB would be around 190-210ms, with most of the added latency back then coming from the mouse delay, as CRTs are essentially lag-free and as windows XP did not have any fancy visual effects or forced V-sync like its modern counterparts, which add a ton of lag.
A somewhat recent study which corrected for latency in a computer based reaction test somewhat similar to the one on humanbenchmark came to the average of 215ms, which is very close to HB’s initial findings from a decade ago. (frontiersin.org/articles/10.3389/fnhum.2015.00131/full). I tried the test used in their study myself and found it to be significantly harder than HB’s, with smaller and faster targets, so it seems logical to assume that the scores on HB’s site when accounted for latency would be somewhat lower than those found in this study.
I suggest using something such as 200ms = 105IQ as the starting point, with suggested values to deduct from your total latency based on your hardware. A cheap wired office mouse should add around +20ms, windows10 around +10-15ms, a modern IPS screen possibly +20-30ms, etc…a modern laptop possibly adding +60ms in total latency, if not more. I realize this complicates things a bit too much for what was meant to be a simple and user friendly way of obtaining intelligence related data, but in the other hand the lack of any correction makes it all pretty much meaningless.
As for my own results, several years ago when doing this test in an average setup without any latency minimization knowledge I recall averaging 200ms. Now while doing it with high-end hardware with the least possible amount of added delay in my setup I manage 135-145ms averages very consistently when focused, with 80% of all clicks being between 130-150ms, and 95% of all clicks probably falling between 120ms-165ms. (rough estimations). Using my proposed correction that would be the equivalent of a 135IQ score, which I have no idea if is accurate or not, but at least it is a sane value.
So what do you make of the average being 253 ms on that website?
hi. I know this is an abandoned blog yet I can’t seem to see this article transferred to the newer site.
I tried to to a more realistic approach to the test, I also think the average human reaction excluding people with disabilities mental or not is below 250 yet I’m not sure is 215ms, maybe for a healthy individual yet a broad spectrum of people aren’t so healthy, I’d try to make that like the average on an iq test and start from there in increments of standard deviation of ~20ms, that I think would have a closer correlation to the actual person’s iq. I also think that on average a smarter person would have a better reaction time and a not so smart person would genereally have a response time that’s always above 200 or above our said average response time for normal healthy people.
sadly the website can’t filter out the bad results that have to do with slow device or people being not interested in playing
My IQ is 3 🙂
About 368, no higher than 370 or lower than 360. This should correspond to an IQ of 60, maybe close to 65. But here I am composing a sentence without a lot of difficulty.
Reaction time tests measure how how well neurons communicate with each other and the body (ideally, they often fail as i detail below). That is all. However even then there are many factors to consider when interpreting results. If one is multitasking while taking the exam or thinking about another matter, their result would likely be adversely affected, as a component of their mental faculties is occupied with another task. People who are conditioned to think before reacting may also produce mediocre results, as these types of tests reward spontaneous impulse action. For example some people will identify the color/ text change before making a decision while others will be on autopilot and click to any change on the screen, even if its an ad popup. The latter may very well produce better results, because they react quicker albeit not fully aware of what they are reacting to. For what its worth I test for a decent iq around 130 but my reaction time results are very mediocre, 30% in 220s range 20% in 250s and a few 150s and the rest below 280.
Also pigeons have exponentially better reaction times than humans.
For the sake of statistical curiosity, I not only recorded my reaction times and averaged them out, but then punched them into this statistical calculator too: https://www.calculator.net/standard-deviation-calculator.html
My mean reaction time was 249.35ms, with plus or minus 9.774ms giving the 99% confidence interval. Not particularly impressive, and especially underwhelming given that I smashed the ceiling by large margins on both subtests of the WAIS-IV Processing Speed Index. Apparently, my ability to process more complex stimuli vastly outstrips my ability at responding to simpler ones. This matches with my experience at playing video games: good at strategizing in real time, less good at things like first-person shooters which require split-second reaction time succeed competitively.
My standard deviation was 16.96ms, well below Jensen’s figure of 29.23ms. My reaction time may be bad, but it’s CONSISTENTLY bad! And after removing the one high outlier of 298ms, the SD drops to about 12ms! I’ve heard before that consistency in reaction time is more g-loaded than the reaction time itself. This observation seems to correspond with that hypothesis.
242.9 averaged from 10 sessions
IDK what my iq is but this was ok. I suspect 117 is a bit inflated for the reaction time but maybe people were using hdtv type displays with slow response time / input lag.
This 20 trials reaction time test IQ matches my WAIS score almost 100%, I got 135 on WAIS and 216 in reaction time.