Fictional characters with the highest IQ

0 Quick Answer

The highest score stated for a fictional character on any widely seen production is 197, and the most repeated is 187, which two separate shows assigned three years apart. Spencer Reid says his in the first episode of Criminal Minds in 2005. Sheldon Cooper says his in the eleventh episode of The Big Bang Theory in 2008. Both numbers are canon in the only sense that matters for fiction: a writer put them in a script, an actor said them, and the line is on the record.

That makes this page unlike every other famous IQ page. Nobody was assessed, so nothing needs correcting, and ACIS produces no estimate of its own here. What can be done instead is to take each figure at face value and run it through the arithmetic any real score is run through. On the standard adult scale, which has a mean of 100 and a standard deviation of 15, a score of 187 sits 5.80 standard deviations above the mean. Under the normal model that puts it at roughly one person in 302 million, or about 27 people alive in a world of 8.2 billion. Two of them would have to be an FBI profiler in Quantico and a theoretical physicist in Pasadena.

The rest of this page does that conversion for every stated figure, names the episode or the page where each one appears, and separates them from the much larger set of numbers that circulate for characters like Tony Stark and Rick Sanchez with no line behind them anywhere. Anime characters are excluded; that is a separate list with different conventions.

1 Every Stated Number and Where It Comes From

Thirteen figures across ten productions have a locatable source, and once they are laid out together the pattern in them is immediate. The rule applied here is narrow. A number qualifies if it is spoken in dialogue, printed on screen, or written in the source text, and if the episode, film, or book can be named. A number does not qualify because a wiki asserts it, because a listicle repeats it, or because the character is obviously brilliant.

The rarity column converts each score with the same arithmetic used for a real result. Subtract 100, divide by 15, and read the tail of the normal curve. That is exactly what a percentile on a report is, and it is set out step by step on the score versus percentile page. Nothing here depends on the fiction being plausible. It depends only on the score having been stated.

Two things jump out of that table. The first is the gap in the middle: there is almost nothing between 105 and 159. Writers who want a character to be smart do not write 118. The second is that the four figures inside the range a real battery can actually report are Homer at 105, Forrest Gump at 75, Charlie Gordon at 68, and Homer again at 55. Every score written to signal brilliance sits outside it.

2 What a Score of 187 Would Actually Describe

A score is a rank, not a quantity, and once you treat 187 as a rank it stops being an adjective and starts being a headcount. The adult scale is defined so that the average is 100 and the standard deviation is 15. Every score converts to a position on that curve, and the position converts to a proportion of people, which converts to a number of human beings. The conversion is fixed. It does not depend on which test is being discussed, only on the scale.

Run 187 through it. The distance from the mean is 87 points, which is 5.80 standard deviations. The area of a normal curve beyond 5.80 is about 3.3 parts in a billion. That is roughly one person in 302 million. The world population passed 8.2 billion in 2024, so the expected number of living people at or above 187 is about 27. Not 27 million and not 27 thousand. Twenty-seven.

Now do the same for 197, which is the top of the table. The distance is 6.47 standard deviations and the tail is about one in 20 billion. Multiply by 8.2 billion and the expected number of living people at that level is 0.41, which is to say fewer than one. The Population Reference Bureau estimated in November 2022, in an analysis by Toshiko Kaneda and Carl Haub, that about 117 billion members of the species have ever been born. Against that denominator, a 197 has an expected count of about six people in the entire history of humanity. Push to 200 and the expected count over all 117 billion births falls to about 1.5.

None of this is a criticism of the writing. It is what the number says when read as a number, which is the one thing nobody does with it. You can run any score through the same conversion on the rarity calculator, and the shape of the whole scale, band by band, is on the IQ score scale. The reason each extra point costs so much at the top is the geometry of the curve itself, explained on the page on the 15 point standard deviation.

3 The Coincidence of Two Separate 187s

Two American network dramas, three years apart, on two different networks, independently landed on the same figure. In September 2005, in the first episode of Criminal Minds, Spencer Reid tells a colleague that he does not believe intelligence can be accurately quantified, and then quantifies it. The line is recorded in the Wikiquote entry for the first season: an I.Q. of 187, an eidetic memory, and a reading speed of 20,000 words per minute. In March 2008, in "The Pancake Batter Anomaly," Sheldon Cooper tells Penny he has an IQ of 187 as an argument for why he would already have thought of ordering soup. The full exchange is in the fan transcript archive for that episode.

Read the two lines side by side and they are doing opposite work. Reid's is a disclaimer followed by a boast, which is the joke. Sheldon's is a boast used to win a domestic argument, which is also the joke. In both cases the number is characterization, not measurement, and both writers reached for the same magnitude because it reads as extreme without reading as invented.

Reid's other two claims are worth a sentence each because they are more testable than the score. An eidetic memory in the strict sense, meaning a persistent visual image that can be scanned after the stimulus is removed, has never been reliably demonstrated in adults, and the popular use of the term almost always describes something else. A reading rate of 20,000 words per minute is roughly 40 times the upper end of normal adult reading with comprehension. The score, oddly, is the most defensible of the three claims, because it at least names a scale.

The habit of attaching a specific figure to a person because it sounds authoritative is not confined to fiction. It is the same mechanism that produces the numbers attached to public figures, traced case by case on the celebrity audit and on the page on presidential IQ claims. The difference is that in fiction the attribution is honest, because there is an author to point at.

4 Why No Real Test Reports a Score That High

No instrument in professional use will print 187 on a report, and the reason is not caution. It is that there are neither items nor people up there. A test score is a rank against a reference sample. To place someone at the 99.9999997th percentile, the sample has to contain people whose performance defines that region, and it has to contain enough of them that the boundary is not set by one person's lucky morning.

Look at what the publishers actually print. Pearson's published WAIS-5 sample score report, dated 2024, plots the primary composite profile on an axis that runs from 40 to 160 and stops. That top value is exactly four standard deviations above the mean. The same convention holds across the major batteries, and what happens above it, along with the famous real-world numbers that appear to break it, is worked through on the highest IQ ever page.

The arithmetic makes the constraint obvious. A standardization sample for a major adult battery runs to a few thousand people. The ACIS adult reference frame contains 3,243 English speaking records across ages 16 to 90, with technical tables computed on 2,750 complete records, and those figures are documented in the technical manual. At a rate of one in 302 million, a frame of 3,243 records would be expected to contain 0.00001 people at 187. There is no observation there to anchor to. Anything printed at that point is the shape of a curve extended past the last real person, not a measured rank.

• At 130, about one person in 44, a sample of 3,000 contains roughly 68 people. The boundary is well populated.
• At 145, about one in 750, a sample of 3,000 contains about four people. The boundary is thin.
• At 160, about one in 31,600, a sample of 3,000 contains none at all in most draws.
• At 187, a sample would need to run to hundreds of millions before one such person appeared.

This is also why classification tables stop where they do. ACIS runs Average from 90 to 109, High Average from 110 to 119, Superior from 120 to 129, and then four gifted bands, and there is no genius band anywhere in it, for the reasons set out on the page on the genius label. How reference frames are built in the first place, and what breaks when the tails are thin, is on the norming page.

5 Why Writers Keep Choosing the 180s and 190s

The chosen numbers cluster in a narrow band, and the band is chosen for how it sounds, not for what it means. Of the nine above-average figures in the ledger, six fall between 159 and 197. None falls between 106 and 158. That is not how ability is distributed; it is how memorable numbers are distributed.

A figure in the 180s does three things at once. It is far enough from 100 that no reader mistakes it for merely bright. It is short, two digits past a familiar anchor, so it survives being repeated. And it stops short of 200, which sounds like a made-up round number and invites the exact scrutiny a writer is trying to avoid. Notice that the two productions that did go past 190 both chose 197 rather than 200. The odd digit is doing the work of sounding measured.

There is a second constraint. The writer needs the number to imply a specific kind of person, and the audience has to supply that meaning without help. A stated 165 for Malcolm in a family sitcom and a stated 187 for a federal profiler are not describing different quantities of the same thing so much as different genres. The score is functioning as a genre marker, which is why the same 187 can support a joke about soup and a scene about serial murder.

The pattern also explains the absence at the low end of the smart range. There is no famous fictional 122, because 122 is not a story. It is a real and meaningfully high score, better than roughly nine people in ten, and it would place a character in the Superior band described on the page on what counts as a good score. It just does not sound like anything. Fiction needs the number to do work that the actual distribution cannot do, so it reaches past the distribution.

6 The Simpsons Has Quantified More Characters Than Anyone

One animated series accounts for five of the thirteen stated figures, and it is the only franchise that has put numbers on both ends of the scale. That is worth its own section, because the show treats the score as a recurring device rather than a one-off characterization, and the device is used for different jokes each time.

In "HOMR," the ninth episode of season 12, first broadcast on January 7, 2001 and written by Al Jean, an X-ray finds a crayon lodged in Homer's brain since childhood. The record of the episode states that after it is removed his IQ goes up fifty points to 105. The premise only lands because both numbers are legible: 55 is a real position on the scale, three standard deviations below the mean, about one person in 741, and 105 is a real position too, a little above average and utterly unremarkable. The joke is that the entire arc of the episode moves Homer from the far tail to the middle of the pack.

In "Smart and Smarter," the thirteenth episode of season 15, broadcast on February 22, 2004 and written by Carolyn Omine, testing reveals that Maggie's IQ of 167 is higher than Lisa's IQ of 159. Those two numbers are eight points apart and the plot treats the gap as decisive. On the real scale, eight points near the top is a genuine rarity difference, roughly one in 252,000 against roughly one in 24,000, which is about a tenfold change. It is also well inside the measurement error of any real instrument, since the standard error on a good composite runs a couple of points and the confidence interval around a single score is several points wide in each direction. The mechanics of that interval are on the reliability and validity page.

The third case is the one with the weakest documentation. Professor John I. Q. Nerdelbaum Frink Jr. is described in the reference record for the character as having an IQ of 197, down from 199 after a concussion, with the figure attributed to "They Saved Lisa's Brain," the twenty-second episode of season 10, broadcast on May 9, 1999. The episode is the one in which Springfield's Mensa chapter briefly runs the town. Unlike the Reid and Cooper lines, this one is harder to pin to a transcript, so it belongs in a different evidence class: attributed to a named episode rather than quoted from it.

7 The Numbers That Circulate With Nothing Behind Them

The most confidently quoted fictional IQ scores are the ones with no line anywhere in the source material. This is the category that makes the search result page for this topic mostly wrong. A character is obviously brilliant, somebody assigns a number, the number gets repeated, and within a few years it is quoted as though a script contained it.

The test applied here is simple and it is the same one used for real people: name the episode, the issue, the page, or the line. If no source can be named, the figure is not canon, and saying so is more useful than repeating it.

The Sherlock Holmes case is the cleanest, because the dates settle it. A Study in Scarlet was published in 1887. Alfred Binet and Theodore Simon published their scale in 1905, and the ratio quotient that produced the first IQ numbers arrived after that, with the classification vocabulary appearing in Lewis Terman's 1916 Stanford revision. Doyle could not have given Holmes an IQ, because the concept was not yet available to give. The chronology is laid out on the history page.

The unsourced figures share a tell with the ones attached to historical figures: they are always high, always specific, and never accompanied by a test name, a date, or an examiner. That combination is what a transmitted claim looks like, and several more of them are catalogued on the myths page.

8 The One Show That Got the Reporting Convention Right

Family Guy staged an entire episode around an IQ test and never stated the score, which is closer to how results are actually reported than anything else on this list. In "Petarded," the sixth episode of season 4, broadcast on June 19, 2005 and written by Alec Sulkin and Wellesley Wild, Peter Griffin takes a test to prove he is a genius. The record of the episode notes that the results show he is not a genius and instead place him in a diagnostic category, using the clinical vocabulary of the period. A classification is given. A number is not.

That is, accidentally, the correct convention. A real report leads with a band and a confidence interval rather than a bare figure, because the band is what the evidence supports. A single point estimate implies a precision that no instrument has. The episode also gets something else right: a classification of that kind is never made from a test score alone. Adaptive functioning has to be documented independently, which is why a score can never be a diagnosis on its own.

ACIS is not a clinical instrument, is not diagnostic, and is not appropriate for hiring, accommodations, or society admission. It is an unsupervised online assessment that reports a profile, and that boundary is stated here rather than buried in a footer because this is the section where it belongs. What a supervised clinical process involves, and where it differs, is set out on the professional versus online comparison and on the page on where testing is done.

The contrast with the Reid line is instructive. Reid states a number and two impossible capabilities in a single sentence, and the audience accepts all three. Peter Griffin gets a category and no number, and the scene works anyway. The number was never the thing carrying the meaning.

9 The Low Numbers Are the Plausible Ones

Every figure in the ledger that a real test could actually produce is at or below average, which is the opposite of what a page about the smartest characters would predict. Homer at 105, Forrest Gump at 75, Charlie Gordon at 68, and Homer again at 55 are all inside the reportable range of a current battery. Every score written to signal exceptional ability is outside it.

Forrest Gump, in the 1994 Robert Zemeckis film, is established at school age with an IQ of 75. That is 1.67 standard deviations below the mean, about the fifth percentile, roughly one person in 21. It is a real and common position on the scale, and the film's plot turns on an institutional threshold rather than on the number itself, which is how such figures are used in practice.

Flowers for Algernon does the most careful work of any fiction on this list. Daniel Keyes published the novelette in The Magazine of Fantasy and Science Fiction in April 1959 and the novel with Harcourt Brace in 1966. The record of the book gives Charlie Gordon an IQ of 68 before the surgery and has it reach 185 within three months afterward. The starting figure sits about 2.13 standard deviations below the mean, roughly the bottom 1.6 percent. The ending figure is 5.67 above, about one in 137 million.

The asymmetry has a technical cause as well as a dramatic one. The lower half of the scale is populated. Norm samples contain plenty of people below the mean, so scores there are anchored to real observations, and clinicians read them constantly. The upper tail is empty by construction. So when fiction reaches downward it lands on numbers that a psychologist would recognize from a case file, and when it reaches upward it lands on numbers no psychologist has ever printed. Whether a score can be moved at all, in either direction, is a separate question handled on the page on changing a score.

10 Depicting Extraordinary Ability Versus Putting a Number on It

The best written intelligent characters in the ledger are the ones with no number, and the reason is a craft reason rather than a psychometric one. Sherlock Holmes has no score. Neither does Walter White, whose competence is demonstrated across sixty-two episodes of chemistry, manipulation, and error. Neither does Gregory House. In each case the audience is shown what the character does, and the showing is the characterization.

A number does the opposite. It replaces demonstration with assertion, which is why it tends to appear in the first episode of a series and rarely again. Reid's 187 arrives in the pilot, in the scene that establishes who he is. Sheldon's arrives in the first season. Malcolm's is revealed in the pilot, per the series record, which describes him as a genius with an IQ of 165 and a photographic memory. The number is an introduction. Once the audience knows the character, the show stops needing it.

There is a real distinction underneath the craft one. Depicting ability means showing behavior: the inference nobody else made, the pattern spotted early, the argument held together under pressure. Quantifying ability means placing a person against a reference group on a defined set of tasks. Those are different operations, and the second one is much narrower than the first. A test samples reasoning with unfamiliar material, accumulated verbal knowledge, quantitative reasoning, spatial manipulation, working memory, and speed. It does not sample judgment, courage, taste, or the willingness to keep going, and no report claims otherwise. What a battery does and does not cover is set out on the page on what IQ measures.

This is also why the stated numbers rarely match the depicted behavior. Rick Sanchez is written as the most capable mind in the multiverse and has no score, and the absence costs the show nothing. Charlie Gordon's 185 is doing something different again: the number is the plot, because the story is about what happens when the quantity moves and everything else in a life stays where it was. Keyes needed the figure precisely because his subject was measurement.

11 The Case Where a Canon Number Came From a Real Claim

One entry in the ledger is not an invention at all, and it is the one that shows what happens when an unverifiable number moves from a person into a script. Scorpion premiered on CBS on September 22, 2014, created by Nick Santora, and ran for four seasons. The lead character is a genius with an IQ of 197. The character is named after and based on a real person, the Irish businessman Walter O'Brien.

The real claim did not survive contact with reporters. The documented record states that O'Brien said he scored 197 on a test administered by a teacher in primary school and did not keep the paperwork. Brian Boyd wrote in The Irish Times in August and October 2014 that childhood scores are scaled by age and that it is impossible to substantiate some of the claims. Susan Karlin asked in Fast Company on October 15, 2014 why he had not simply retaken a supervised test if his credibility rested on the score. Mike Masnick noted at Techdirt on September 25, 2014 that the name appears on none of the published high score lists. Andrew Fenton made the age scaling point again at news.com.au on September 5, 2015.

Every one of those objections is a standard one, and every one of them applies to real claims constantly. A childhood ratio score and an adult deviation score are different quantities computed differently, a point developed on the mental age page. An undocumented administration cannot be checked. A score with no named instrument, no date, and no examiner is not a result, it is a memory of one.

What is unusual here is the direction of travel. Normally an unsupported number attaches to a person and stays there. In this case it attached to a person, could not be verified, and then became canon by being written into a character, where it is now unfalsifiable in the only way that matters: the character really does have a stated 197, because the show says so. The claim about the man remains exactly as thin as it was. The problem of estimating your own standing without a real administration, which is where most such claims start, is examined on the page on self estimated ability.

12 What a Real Result Reports Instead of One Number

The single figure that fiction reaches for is the least informative thing on an actual report. A composite is an average across tasks that measure different abilities, and averaging is exactly the operation that destroys the pattern. Two people with the same composite can be strong and weak in opposite places, and the profile is what tells them apart.

The structure that recovers the pattern is the CHC framework, which organizes ability into broad domains rather than a single quantity. ACIS reports six of them across 20 subtests: verbal comprehension, fluid reasoning, quantitative reasoning, visual spatial processing, working memory, and processing speed, reported as the VCI, FRI, QRI, VSI, WMI, and PSI indices. Subtest scaled scores use a mean of 10 and a standard deviation of 3; the indices use the familiar mean of 100 and standard deviation of 15. The framework is described on the CHC model page and the domain by domain breakdown is on the cognitive domains page.

The published reliability figures show why the profile carries information the composite cannot. In the technical manual, VCI has an omega of .9745 with a standard error of measurement of 2.40 and a g loading of .864, while FRI has an omega of .9727, a standard error of 2.48, and a g loading of .922. PSI is the loosest of the six, with an omega of .8790, a standard error of 5.22, and a g loading of .648. Those indices do not measure the same thing and they do not agree with each other: VCI correlates .801 with FRI, but WMI correlates only .475 with PSI. A character described by one number is described by an average of six quantities that disagree.

Dispersion across those six is the normal case at high ability, not a warning sign. A person can sit two standard deviations above the mean on verbal comprehension and near the middle on processing speed, and the composite describes neither of those accurately on its own. That is the argument for reading the shape rather than the level, and it is developed further on the fluid versus crystallized comparison and on the gifted range page.

Which leads to the distinction that actually matters, and it is not how high a score is but whether the administration can be described. A result you can defend has four things attached to it: a named instrument, a date, the conditions it was taken under, and a confidence interval. A fictional 187 has an author instead, which is a perfectly good answer to a different question.

A proctored session with a licensed psychologist using a current battery is what institutions accept, because identity was verified and a qualified examiner observed the performance. Nothing administered over the internet substitutes for that. An unsupervised online assessment answers a narrower question honestly: where does this profile sit, on tasks that have been normed, with the error stated. The difference between a normed battery and a quiz is set out on the free versus validated comparison, and the case for a longer administration is on the accuracy page.

ACIS is the second kind. It runs 20 subtests across the six CHC domains for adults aged 16 to 90, in three tiers: Quick at 15 dollars covering six subtests in about 45 minutes, Optimized at 30 dollars covering 13, and Full Scale at 50 dollars covering all 20 and reporting a Full Scale IQ with every composite. Payment is one time, paid access lasts 30 days, and there is a 5 day guarantee. The free trial covers 5 subtests and returns scaled scores and a provisional estimate. What the session involves is described on the adult test page and on the online test page, and the difficulty question that this article's ceiling discussion raises is handled on the page on hard tests.

For research and educational work there is a separate route. ACIS Professional provides private participant links, no participant personal data, verified scores, and CSV export, with a free account that includes three Quick administrations. Details are on the research landing page and in the administration section of the home page.

13 Sources Behind This Page

Every figure in the ledger is traced to a document that anyone can open, and the ones that could not be traced were placed in a separate table rather than quietly included. The distinction between quoted from a transcript, stated in a plot record, and attributed without a locatable line is preserved throughout, because that distinction is the entire method of this page.

• Criminal Minds, season 1 quotations. The Spencer Reid line stating an I.Q. of 187, an eidetic memory, and a reading speed of 20,000 words per minute, from "Extreme Aggressor," first broadcast September 22, 2005.
• The Big Bang Theory, series 1 episode 11 transcript. Sheldon Cooper's line stating an IQ of 187 in "The Pancake Batter Anomaly," first broadcast March 31, 2008.
• HOMR, season 12 episode 9 of The Simpsons, written by Al Jean, first broadcast January 7, 2001. Homer's IQ rises fifty points to 105 after the crayon is removed.
• Smart and Smarter, season 15 episode 13 of The Simpsons, written by Carolyn Omine, first broadcast February 22, 2004. Maggie's IQ of 167 against Lisa's 159.
• The reference record for Professor Frink, which attributes an IQ of 197, down from 199, to "They Saved Lisa's Brain," season 10 episode 22, first broadcast May 9, 1999.
• Petarded, season 4 episode 6 of Family Guy, written by Alec Sulkin and Wellesley Wild, first broadcast June 19, 2005. A classification is stated and no number is given.
• Malcolm in the Middle, created by Linwood Boomer for Fox, first broadcast January 9, 2000. Malcolm is revealed in the first episode to have an IQ of 165.
• Flowers for Algernon by Daniel Keyes, novelette in The Magazine of Fantasy and Science Fiction, April 1959, novel published by Harcourt Brace in 1966. Charlie Gordon's IQ of 68 rising to 185.
• Forrest Gump, directed by Robert Zemeckis, 1994. The character is established at school age with an IQ of 75.
• The record on Walter O'Brien, including the claimed score of 197 and the objections published by Brian Boyd in The Irish Times in 2014, Susan Karlin in Fast Company on October 15, 2014, Mike Masnick at Techdirt on September 25, 2014, and Andrew Fenton at news.com.au on September 5, 2015.
• Pearson's WAIS-5 sample score report, 2024. The primary composite score profile is plotted on an axis running from 40 to 160.
• Kaneda and Haub, Population Reference Bureau, November 15, 2022. About 117 billion members of the species have ever been born.
• Terman, The Measurement of Intelligence, 1916. The Stanford revision that introduced the classification vocabulary the modern scale inherited.

All rarity figures on this page were computed the same way: convert the stated score to a distance from 100 in units of 15, then read the tail of the normal distribution. Expected headcounts use a world population of 8.2 billion and the Population Reference Bureau figure of 117 billion births. The normal model is an idealization and real distributions depart from it in the extreme tails, which means these figures should be read as the arithmetic the scale implies rather than as demographic predictions.

The professional framework governing score interpretation is explicit about all of this. The International Test Commission guidelines on test use and the Standards for Educational and Psychological Testing (2014), published jointly by the American Educational Research Association, the American Psychological Association, and the National Council on Measurement in Education, require that a reported score be tied to documented evidence for its intended use, that classification be reported with its measurement error, and that the limits of the reference sample be disclosed. APA standards on test use make the same demand of anyone reporting a classification: state what the number supports, state what it does not, and never let a label carry weight the evidence cannot back. A number with an author instead of an administration fails every one of those requirements, which is exactly why it belongs in a script and nowhere else.

14 Frequently Asked Questions