Where do Savant Skills Come From?

Where do Savant Skills Come From?

Where do Savant Skills Come From?

Unraveling the mysteries surrounding the exceptional performance of savants.

There’s a scene in the 1988 movie Rain Man in which Raymond Babbitt (played by Dustin Hoffman) recites a waitress’s phone number. Naturally the waitress is shocked. Instead of mental telepathy, Raymond had memorized the entire telephone book and instantly recognized the name on her nametag.

Hoffman’s character was heavily influenced by the life of Kim Peek, a real memory savant who recently passed away. Peek was born without a corpus callosum, the fibers that connect the right and left hemispheres of the brain. He was also born missing parts of the cerebellum, which is important for motor control and the learning of complex, well-rehearsed routines.

When Peek was 9 months old, a doctor recommended he be institutionalized due to his severe mental disability. By the age of 6, when Peek had already memorized the first eight volumes of the family encyclopedia, another doctor recommended a lobotomy. By 14, Peek completed a high school curriculum.

Peek’s abnormal brain wiring certainly came at a cost. Though he was able to immediately move new information from short-term memory to long-term memory, there wasn’t much processing going on in between. His adult fluid reasoning ability and verbal comprehension skills were on par with a child of 5, and he could barely understand the meaning in proverbs or metaphors. He also suffered deficits in the area of self-care: he couldn’t dress himself or brush his teeth without assistance.

But what Peek lacked in brain connections and conceptual cognitive functioning, he more than made up for in memory. He had the extraordinary ability to memorize any text in just one sitting. With two pages in front of him, he had the uncanny ability for each eye to focus on a different page. His repertoire included the Bible, the complete works of Shakespeare, U.S. area codes and zip codes, and roughly 12,000 other books. He was known to stop performances to correct actors and musicians who had made a mistake! He could also tell you what day of the week your birthday fell on in any year.

Toward the end of Peek’s life, Peek showed a marked improvement in his engagement with people. He also began playing the piano, made puns, and even started becoming more self-aware. During one presentation at Oxford University, a woman asked him if he was happy, to which he responded: “I’m happy just to look at you.”

The trade-off between memory and meaning is common among savants. The purpose of memory is to simplify experience. We didn’t evolve memory to be precise. Instead, we extract meaning wherever we can so that we can organize the regularities of experience and prepare for similar situations in the future. But without the imposition of meaning, savants can focus on literal recall. Some savants even have hyperlexia, which is the opposite of dyslexia. They are precocious readers, but have no comprehension of what they are reading.

Descriptions of savant syndrome first appeared in the scientific literature as early as 1789. In 1887 the British doctor J. Langdon Down (who discovered Down syndrome) described 10 people with savant syndrome and used the term “idiot savant” (which is no longer used, because of its pejorative connotation). Today, one in ten people with autism have savantism, although only half of the documented savants are autistic. The rest have some other kind of developmental disorder.

Savantism disproportionately affects males, with about five male savants for every one female, and the syndrome generally occurs in people with IQs between 40 and 70. Like others with ASD, when savants take IQ tests they tend to score higher on nonverbal problems than verbal problems. As Darold Treffert, a world-renowned expert on savant syndrome, observes, “IQ scores, in my experience with savants, fail to adequately capture and reflect the many separate elements and abilities that contribute to ‘intelligence’ overall in everyone.”

Like others on the autism spectrum, savants display a narrow repertoire of skills, which tend to be highly structured, rule-based, and nonverbal. Common savant domains include music, art, calendar calculating, lightning calculating, and mechanical/visual spatial skills. Most musical savants are blind and have perfect pitch, most artistic savants express themselves through realistic drawing and sculpture, and most rapid calculating savants have a fascination and facility with prime numbers.

Even so, savants vary markedly in their abilities. Savant skills fall along a continuum, ranging from “splinter skills” (such as memorization of license plates), to “talented” savants who have musical or artistic skills that exceed what is expected based on their handicap, to “prodigious” savants where the skill is so remarkable it would be impressive with or without the disability. To date, fewer than 100 prodigious savants have been documented. Interestingly, there is almost always no “dreaded trade-off ” between the incredible skills of savants and their development of language, social skills, and daily living functioning.

Savant syndrome can be congenital or acquired. When congenital, the skill appears early in childhood, and when acquired, abilities appear to spring forth suddenly following stroke, brain injury, or dementia. Many savants are creative, in additions to being highly skilled. Here is a sampling of some of these remarkable individuals:


How can we explain the extraordinary feats of savants? No one knows the whole story, but there are some clues. Bernard Rimland, who passed away in 2006, maintained the largest database in the world of people with autism (more than 34,000 cases). He observed that the savant skills that were most frequently present were right-hemisphere skills, and their deficits were most strongly associated with left-hemisphere functions.

Treffert argues that we may all have a reservoir of inherited knowledge, but savants are able to access this knowledge through a process of rewiring, recruitment, and release. According to Treffert’s account, left-hemisphere damage or dysfunction causes recruitment of still-available intact cortical tissue elsewhere in the brain to compensate for the loss, and this rewiring releases dormant capacity within that still-intact cortical tissue. Although it’s typically the case that left-hemisphere brain damage causes right-hemisphere brain recruitment, rewiring, and release, Treffert acknowledges this is not always the case.

Another prominent theory, proposed in the 1980s by Norman Geschwind and Albert Galaburda, offers an explanation as to why disorders involving the disruption of the left hemisphere of the brain (such as savantism, autism, dyslexia, delayed speech, stuttering, hyperactivity) occur so much more often in males than in females. The left hemisphere typically completes its development later than the right hemisphere, so it is susceptible to prenatal influences for a longer period. Therefore, in the developing male fetus, circulating testosterone can slow the growth of the left hemisphere. This can cause compensation, in which the right hemisphere becomes bigger and more dominant in males.

This left-brain damage/right-brain compensation hypothesis is consistent with some recent threads of research. Allan Snyder and his colleagues have had some success in increasing savant-like skills (e.g., proofreading ability, artistic style) in University students by using low-frequency magnetic pulses to suppress the left fronto-temporal lobe.

Also consistent is research involving elderly patients with frontotemporal dementia (FTD). Bruce Miller and colleagues found that some of their patients with degeneration in very specific regions of the frontal and anterior temporal lobes on the left side of the brain suddenly expressed interest and skill in art and music as their dementia progressed. The paintings that the patients produced were generally realistic or surrealistic without symbolism or abstraction, and the patients approached their art in a compulsive way, repeating the same design many times. For some reason the colors purple, yellow, or blue were commonly repeated. Some of these patients described their experience as though their brain was “taken over” by a compulsive need to create.

Treffert believes savants are the best example of innate talent and “represent ‘nature’ in its most basic form.” To explain the emergence of savant skills, Treffert proposes the notion of “genetic memory,” which he defines as the biological transfer of proclivities and knowledge that don’t require additional instruction or learning. He argues that this knowledge is “factory- installed” in all of us but remains dormant because we tend to use the same well-worn pathways and circuits that serve us well. He believes this inhibits the “little Rain Man in all of us.”

There are a few separate aspects of Treffer’s theory that are worth pulling apart. The first idea is that we have factory-installed “templates” that ease learning in a domain, the second notion is that we are born with full domain knowledge, and the third idea is that the major difference between savants and the rest of us is that they have more direct access to these templates and knowledge.

On the first part, Treffert is most surely correct. We aren’t born blank slates. In recent years researchers have begun to map out the various domains of the human mind. An accumulating body of research suggests that humans have an innate facility to learn and reason about spatial relations, number, probability, logic, language, physical objects, living things, artifacts, music, aesthetics, and the beliefs and desires of other minds.

So Treffert is on firm ground when arguing that we have built-in mental structures. I’d also agree that we all have our own unique brain wiring that influences our interests in mastering a particular domain of human knowledge and even greatly facilitates such learning. That’s how I define talent.

What remains far more controversial, in my view, is Treffert’s notion that genetic memory consists of built-in knowledge that doesn’t require any learning or experience. In a recent review paper entitled “Savant Syndrome: Realities, Myths and Misconceptions,” Treffert writes: “The only way to know things one never learned– sometimes at complex levels– is for that knowledge to be factory installed, genetically transmitted.”

It’s not clear to me how this is possible. After all, our genes don’t code for traits; they code for the production of proteins. Although it’s true that proteins are the building blocks of everything we do— they contribute to the formation of cells and the transport of elements from one location to another, and are the foundation for chemical reactions—they are far removed from anything we would recognize as psychological traits.

Allow me to suggest another way.


Most, perhaps as much as 95 percent, of the learning that takes place in our day-to-day lives operates implicitly—no explicit instruction was available or necessary. Much of this learning is incredibly complex, considering that information in the environment includes so many multidimensional and interactive relations.

Compared to conscious, deliberate thought, our implicit learning structures are much faster and more efficient. In their review of the literature on the implicit acquisition of knowledge, Paul Lewicki, Thomas Hill, and Maria Czyzewska argue that implicit learning must be considered intelligent, if “intelligence is understood as ‘equipped to efficiently process complex information.’ ”

Perhaps savant skills are learned, just like any other skills. But the reason their skills appear to burst forth without any prior practice or explicit instruction is that their learning operated implicitly, facilitated by their unique brain wiring.

Consider the domain of calendar calculation. The calendar has many sequential regularities and redundancies that make it particularly attractive to savants. While formulas exist online that anyone can use to become a calendar calculator, the underlying rules can be discovered with enough exposure to calendars. For instance, if instead of going to school dances and playing cops and robbers, you spent every waking second of your life obsessively studying calendars, your powerful implicit learning system would eventually discover the following regularities:

  • There are always 7 days to a week.
  • The 1st, 8th, 15th, 22nd, and 29th must always be the same day of the week.
  • In any given year, particular pairs of months (such as April and July) always have the same starting day, regardless of leap year.
  • The pair January and October and the triad February, March, and November have the same starting day only on non-leap years.
  • Only in leap years do February and August have the same starting day. Same for January, April, and July.
  • January 1st advances one day of the week each succeeding year, except leap years, when it advances two days.
  • Because a day is added every 4 years, the calendar shifts 5 days every four years.
  • Except for a few exceptions, the calendar repeats every 28 years.
  • There is a 400-year cycle.

Systematic studies on calendar-calculating savants—based on reaction time information and a look at systematic errors—suggests that calendar- calculating savants do indeed incorporate these redundancies into their strategy for making their calculations. As one researcher noted, “Something mysterious, though commonplace, is operating here—the mysterious human ability to form unconscious algorithms on the basis of examples.”

Theoretically, all that is required to explain savant skills is an innate predisposition to find redundancies and sequential regularities fascinating and an intact implicit learning system that gradually extracts those regularities over many hours of experience. Consistent with this idea, my colleagues and I have found that although the explicit learning functions of individuals with autism spectrum condition (ASD) are impaired, their implicit learning mechanisms are preserved.

Now consider the music domain, which also consists of highly structured sequential regularities. In recent years a number of studies have shown that, just as implicit learning plays an important role in acquiring the rules of language, various musical structures such as melody, harmony, timbre, and rhythm are implicitly acquired through mere exposure. One study found that after being exposed to sequences of tones that conformed to a complex artificial grammar, participants later showed faster and more accurate processing for a new set of grammatical tones compared to ungrammatical tones, without any explicit awareness that they learned such information.

Due to their language impairment, perhaps for musical savants, their improvisation becomes their main source of communication. Also, perhaps they are drawing on the same implicit learning mechanisms we all implicitly employ whenever we listen to music or engage with language.

Consider a telling study by Jonathan Sloboda and colleagues. They studied a musical savant they referred to as NP who lived in a residential home for people with autism. From an early age, NP listened obsessively to music on the radio, and he showed an early propensity for mimicking speech and imitating sounds. A tape recording made between the ages of 5 and 8 shows that his musical skills were progressing at a fast rate.

At the time the researchers encountered NP he had a substantial repertoire of classical pieces and was able to learn a new sonata-length piece in three or four hearings. His verbal IQ was found to be 62 and his performance IQ was 60. His short-term memory was unexceptional: he could only remember 5 digits forward and 4 digits backward. Like other people on the autism spectrum, NP had an almost complete absence of speech, averted his eye gaze, and displayed obsessive behaviors. The researchers noted that NP would probably always live in an institutional setting.

The researchers played two unfamiliar piano recordings for NP. One piece was by Bartok and had a clear melody, whereas the other piece was atonal and consisted of whole tone scales. Compared to the Bartok, the atonal piece contained fewer notes, fewer voices, and an equally simple rhythmic and formal structure. The key difference was that it had no systematic tonal structure. After hearing each piece, NP was asked to play as much of the piece as he could. On another occasion the same pieces were given to a professional pianist roughly the same age and years of experience to memorize under the same conditions.

After about 12 minutes and hearing the piece no more than four times, NP played back the melodic piece almost perfectly. He also appeared to have perfect pitch, because he never needed to try out different positions before starting to play. NP was also able to play the melodic piece back perfectly 24 hours later, although his repeat performance had become “wooden” and “metronomic.” The researchers observed that he retained the structural “husk” but discarded the expressive “flesh.” In contrast, the professional pianist’s performance was good only on the first eight bars, but she rapidly became overwhelmed with the rapid succession of notes. The researchers tested other professional pianists and found the same thing: none were able to perform even half as well as NP.

The atonal piece told a completely different story. By the third trial, the professional pianist could play the first 12 bars without error, whereas NP was still making errors on the third trial. NP (but not the professional pianist) declined to go any further and was visibly distressed that he couldn’t memorize the atonal piece. The researchers analyzed NP’s pattern of errors and concluded that they were “structure preserving.”

NP seemed to be utilizing a framework that enabled him to rapidly encode material in terms of tonal structures and relations, but when the music went outside that framework he was unable to cope with the new information. Interestingly, they found that he was highly sensitive to structure in domains other than music. While his short-term memory for digits was below average, he was able to memorize conventionally ordered sentences.

They also noticed that NP was able to put appropriate chords under a melodic beginning, and then improvise a conventional conclusion—“the kind of elaboration that one might expect to find in a proficient improviser.” In contrast, the professional pianists’ errors were more evenly distributed. The researchers conclude that a high IQ is not necessary for the development of high levels of musical memorization skill, and that when we look at the memory capacity of NP and Mozart, “we are looking at essentially the same phenomenon.”

I wouldn’t go that far—clearly Mozart’s musical skills went well beyond the ability to memorize pieces and improvise within a particular structured framework. But these findings do highlight the notion that prodigious skills of savants and prodigies aren’t superhuman. Their phenomenal feats can be explained by the same powerful learning and motivational mechanisms that the rest of us use to make sense of our world.

Of course, this shouldn’t take any of the magic out of their truly magnificent performances. They significantly enrich our world. But let’s not forget that these individuals have the same fundamental basic needs as the rest of us. They too, want to find a place in the world where they can engage their unique minds.

Acknowledgement: I am deeply appreciative for Darold Treffert’s friendship, advice, and feedback. Even though we don’t always see eye to eye on how savant skills are acquired, he is always respectful and thoughtful.

Portions of this article were excerpted from Ungifted: Intelligence Redefined.

image credit #1; illustration by George Doutsioulos.

This post originally appeared at Scientific American.

comments powered by Disqus