[Paleopsych] Satoshi Kanazawa: Why productivity fades with age: The crime-genius connection

[Premise Checker]

Satoshi Kanazawa: Why productivity fades with age: The crime-genius connection
Journal of Research in Personality 37 (2003) 257-272
Department of Psychology, University of Canterbury, Private Bag 4800, 
Christchurch, Canterbury, New Zealand

Abstract

The biographies of 280 scientists indicate that the distribution of their age 
at the time of their greatest scienti.c contributions in their careers 
(age-genius curve) is similar to the age distribution of criminals (age-crime 
curve). The age-genius curves among jazz musicians, painters and authors are 
also similar to the age-crime curve. Further, marriage has a strong desistance 
e.ect on both crime and genius. I argue that this is because both crime and 
genius stem from men's evolved psychological mechanism which compels them to be 
highly compet­itive in early adulthood but "turns o." when they get married and 
have children. Fluctuating levels of testosterone, which decreases when men get 
married and have children, can provide the biochemical microfoundation for this 
psychological mechanism. If crime and genius have the same underlying cause, 
then it is unlikely that social control theory (or any other theory speci.c to 
criminal behavior) can explain why men commit crimes and why they desist.

1. Introduction

A person who has not made his great contribution to science before the age of 
thirty will never do so.
Albert Einstein (Brodetsky, 1942, p. 699)
*
Fax: +64-3-364-2181.E-mail address: Satoshi.Kanazawa at canterbury.ac.nz.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272

Anecdotal evidence abounds that artistic genius or productivity fades with age. 
Paul McCartney has not written a hit song in years, and now spends his time 
painting.
J.D. Salinger now lives as a total recluse and has not published anything in 
more than three decades. Orson Welles was mere 26 when he wrote, produced, 
directed and starred in Citizen Kane, which many consider to be the greatest 
movie ever made.
The relationship between age and genius appears to be the same in science. It 
is often said that physics and mathematics are young men's games, and 
physicists and mathematicians tend to think they are over the hill at age 25 
(Mukerjee, 1996). John von Neumann, putatively the most brilliant scientist who 
ever lived, used to assert brashly when he was young that mathematical powers 
decline after the age of 26, and only the bene.ts of experience conceal the 
decline—for a time anyway. (As von Neumann himself aged, however, he raised 
this limiting age.) (Poundstone, 1992, p. 16). James D. Watson made the 
greatest discovery in biology in the 20th century at the age of 25, winning the 
Nobel prize for it, but has not made any other signi.cant scienti.c 
contribution for the rest of his career.
This paper addresses two questions. Does productivity truly fade with age? If 
so, what explains this phenomenon? While the question of why productivity fades 
with age in itself may be of trivial scienti.c importance, I will argue that 
the study of the age trajectories of scientists and other geniuses illuminates 
a very important question in behavioral science: Why men commit crimes and why 
they desist. I will note that the relationship between age and genius, not only 
among scientists but among mu­sicians, painters, and authors as well, is very 
similar to the relationship between age and criminality, and suggest that this 
is because the same mechanism produces the expressions of both genius and 
criminality. I will further note that marriage has the same negative e.ect on 
both genius and criminality, and thus any criminological theory that explains 
the desistance e.ect of marriage purely in terms of social control is not 
su.cient (because scientists, unlike criminals, are not subject to social 
control, and because scienti.c work is not illegal or deviant in any way).
2. Does productivity really fade with age?
In order to examine the relationship between age and scienti.c productivity, I 
study a random sample of the biographies of 280 scientists (mathematicians, 
physi­cists, chemists, and biologists) from The Biographical Dictionary of 
Scientists (Porter, 1994). There are a few scientists from the 16th and 17th 
centuries, but the over­whelming majority comes from the 18th century to the 
present. The biography of each scientist in this dictionary follows the same 
format. The .rst, brief paragraph lists the scientist's full name, years of 
birth and death, his nationality and .eld of re­search, and the most signi.cant 
scienti.c contribution in his entire career. (97.8% of the scientists in my 
sample are male.) For most Nobel laureates, this is the discovery or research 
for which they won the Nobel prize.
The next one or two paragraphs detail the scientist's educational career and 
the history of institutional a.liations—where he received his degrees and which 
positions he held at what institutions. Then the next few paragraphs summarize 
the research career of the scientist, enumerating the dates of major 
discoveries and publications. I use the date of the discovery or experiment 
which is listed in the .rst paragraph as the scientist's most signi.cant 
contribution in his career to denote the peak of his career. If the date of the 
discovery or experiment is di.erent from the date of its publication, I use the 
former date. Then I calculate the scientist's age at the peak of his career, by 
subtracting the year of his birth from that of his peak.
Fig. 1 presents the distribution of the peak age among the 280 scientists in my 
sample. It is apparent from the histogram that scienti.c productivity indeed 
fades very rapidly with age. Nearly a quarter (23.6%) of all scientists makes 
their most sig­ni.cant contribution in their career during the .ve years around 
age 30. Two-thirds (65.0%) will have made their most signi.cant contributions 
before their midthirties; 80% will have done so before their early forties. The 
mean age for the peak of scien­ti.c career is 35.4; the median is 34.0. Most 
signi.cantly, the interquartile range (the distance between the 75th and 25th 
percentile, encompassing the middle half of the

Fig. 1. The age of peak scienti.c achievement, 280 scientists.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
distribution) is merely 12 years. Peak scienti.c productivity appears to occur 
in a quick burst within a few years of the scientists' lives around the age 30.
My data replicate Lehman's (1953) classic study of the history of scienti.c 
discov­eries, which shows that more signi.cant discoveries are made by younger 
scientists than by older ones, and thus the age of the scientist has a negative 
e.ect on the like­lihood of making a signi.cant discovery. My data are also 
consistent with Cole's (1973) and Levin and Stephan's (1991) studies of 
representative samples of contem­porary scientists, which show that scienti.c 
productivity rapidly increases shortly af­ter the Ph.D. and gradually declines 
thereafter. Taken together, the evidence does seem to indicate that scienti.c 
productivity appears to fade with age.
3. What about other types of productivity?
Fig. 1 demonstrates the age distribution of scienti.c productivity, but what 
about other types of productivity? Scienti.c discoveries are not the only way 
genius ex­presses itself. What about more artistic forms of genius? Music? 
Literature?
Fig. 2 presents the relationship between age and productivity in jazz music 
(Mill­er, 1999, Fig. 5.1). It plots, separately for men and women, the age at 
which 719 jazz musicians released their 1892 albums. (Unlike the age 
distribution of the greatest sci­enti.c discoveries in Fig. 1, the 
distributions in Fig. 2 counts the same musician more than once. However, 
Simonton's (1988, 1997) equal-odds rule asserts that scientists make the most 
signi.cant contributions when they make the largest number of con­tributions. 
If Simonton is correct, then these two measures, one of quantity and the other 
of quantity, are equivalent.) Fig. 2 shows that the relationship between age 
and productivity in jazz music among male musicians is virtually identical to 
the relation­ship between age and scienti.c discoveries among largely male 
scientists in Fig. 1. There appears to be no discernible relationship between 
age and jazz productivity among female musicians. In this random sample of jazz 
albums produced between

Fig. 2. The age-genius curve among jazz musicians. Source: Miller (1999).
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
the 1940s and 1980s in the United States or Britain, the male musicians 
outnumber the female musicians by 20 to 1 (male:female ¼ 685:34).
Fig. 3 presents the same relationship among modern painters (Miller, 1999, Fig. 
5.2). It plots, separately for men and women, the age at which 739 artists 
painted 3274 painting. Once again, Fig. 3 clearly shows that the relationship 
between age and productivity in modern paintings among male artists is 
virtually identical to the age distribution of scienti.c discoveries in Fig. 1. 
Once again, the same relation­ship does not hold among female painters. In this 
exhaustive sample of every datable painting owned by the Tate Gallery, London, 
as of 1984, where the artist's last name begins with A through K, the male 
artists outnumber the female artists by roughly seven to one (male:female ¼ 
644:95).

Fig. 3. The age-genius curve among painters. Source: Miller (1999).

Fig. 4. The age-genius curve among authors. Source: Miller (1999).
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
Finally, Fig. 4 presents the same relationship among authors (Miller, 1999, 
Fig. 5.3). It plots, separately for men and women, the age at which 229 writers 
published 2837 books. Once again, Fig. 4 demonstrates that the relationship 
between age and literary productivity among male authors is virtually identical 
to the age distribution of scienti.c genius in Fig. 1. The same relationship 
among female authors, if it exists at all, is far weaker and seems to peak 
somewhat later. In this random sample of 20th century English-language .ctions 
and non.ctions, the male authors outnumber fe­male authors by roughly four to 
one (male:female ¼ 180:49).
Thus the relationships between age and productivity in .elds as varied as 
science, music, art and literature share two characteristic in common. First, 
in all .elds, the age distribution among male practitioners has the virtually 
identical form. Second, in all .elds, men far outnumber the women. What can 
possibly explain these common features in the age distribution of genius in 
such varied .elds?
4. The crime-genius connection
The most curious aspect of the relationship between age and genius represented 
in Figs. 1-4 is that these distributions (which I would like to call the 
"age-genius curves") very closely resemble another very well-known age 
distribution: The invari­ant age-crime curve (Hirschi & Gottfredson, 1983), 
presented in Fig. 5. Criminolo­gists widely recognize that criminal behavior, 
especially among men, rapidly rises during adolescence, peaks in late 
adolescence and early adulthood, and then equally rapidly declines through 
adulthood, reaching a plateau at a very low level around

Fig. 5. The age-crime curve. Source: Kanazawa and Still (2000, p.435, Fig. 1).
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272 263
age 40. (For empirical illustrations of the invariant age-crime curve, see 
Blumstein, 1995, Figs. 2 and 3; Daly & Wilson, 1990, Fig. 1; Hirschi & 
Gottfredson, 1983, Figs. 1-78). While the validity and universality of the 
invariant age-crime curve, with some minor variations, are beyond dispute in 
the criminological literature, there cur­rently is no satisfactory theory that 
can explain why the relationship between age and criminal behavior takes the 
shape that it does.1
Kanazawa and Still (2000) o.er an evolutionary psychological explanation for 
the invariant age-crime curve. They extend Daly and Wilson's (1988, 1990) 
theory of ho­micide and explain all types of violent and property crimes as 
consequences of young men's competition for access to women's reproductive 
resources. The theory posits that young men become rapidly violent and criminal 
during the years right after pu­berty. There is no point for prepubertal boys 
to compete for women, but the repro­ductive bene.ts of competition quickly 
rises after puberty, since post-pubertal men can translate increased access to 
women's reproductive resources into greater repro­ductive success (see Fig. 
6a). The theory also explains the rapid decline in criminal behavior among 
adult men as a function of increased costs of competition and its potentially 
harmful e.ects on reproductive success (see Fig. 6b). While men can al­ways 
increase their reproductive success by gaining greater access to women's 
repro­ductive resources, competition for women can result in their own death or 
injury, which would be detrimental to the welfare of their existing o.spring. 
In other words, while the reproductive bene.ts of competition (interpersonal 
violence and property malappropriation) remain high for men for their entire 
lives (as Fig. 6a shows), the reproductive costs of such competition quickly 
increase after they have had chil­dren (as Fig. 6b shows). Their children will 
su.er if they are injured or killed in the course of the competition. Kanazawa 
and Still argue that this is why men desist quickly during early adulthood, 
when they were likely to have had their children in the ancestral environment. 
The age-crime curve is the mathematical di.erence be­tween the reproductive 
bene.ts and costs of competition (see Fig. 6c).
It is important to keep in mind two signi.cant points in any discussion of 
evolu­tionary psychological theory of human behavior (Kanazawa, 2001). First, 
evolved psychological mechanisms, such as the ones that compel young men to act 
violently
1
There is another uncanny resemblance between crime and scienti.c productivity. 
Cole's (1979) study of a representative sample of contemporary mathematicians 
in the United States demonstrates that, while the career trajectories of a 
majority of mathematicians follow what I call the "age-genius curve," where 
their productivity, measured both by the quality and quantity of their 
publications, peaks very early in their careers and gradually declines 
thereafter, there is a small minority of mathematicians who produce a large 
quantity of high-quality work throughout their careers. This dichotomy of 
mathematicians is reminiscent of Mo.tt's (1993) taxonomy of 
"adolescence-limiteds" and "life-course persistents" among criminals. Mo.tt 
argues that most men's antisocial behavior peaks in adolescence and then 
declines throughout the rest of their lives (following the age-crime curve), 
while there is a small minority of career criminals who continue to engage in 
anti-social behavior throughout their lives. While my focus in this paper is on 
the majority of scientists and criminals whose expressions of genius and 
criminality follow a predictable life-course pattern, I would not be surprised 
if the same hormonal factors underlie the behavior of what Cole (1979) calls 
life-long "strong publishers" and that of what Mo.tt calls "life-course 
persistents."
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272

Fig. 6. The bene.ts and costs of competition and the age-crime (and age-genius) 
curve. (a) Reproductive bene.ts of competition. (b) Reproductive costs of 
competition. (c) Propensity toward competition ¼ ben­e.ts ) costs.
toward each other, operate mostly behindconscious thinking. Young men feel like 
act­ing violently or want to steal others' property, but they do not know why. 
Organisms (including humans) are usually not privy to the evolutionary logic 
that placed the psychological mechanisms in the brain to solve adaptive 
problems. Criminals them­selves are therefore unaware of the ultimate causes of 
their behavior; they are not consciously pursuing reproductive success when 
they engage in criminal behavior. Their preferences and desire for violence and 
crime serve as the proximate causes of their behavior.
Second, all evolved psychological mechanisms are adapted to the ancestral 
envi­ronment where humans evolved for millions of years. Behavior that stems 
from evolved psychological mechanisms (such as criminal behavior) is therefore 
often mal­adaptive in the current environment, which is so vastly di.erent from 
the ancestral environment. In particular, the psychological mechanism that 
compels young men to be violent and steal from others assume that there are no 
third-party enforcers of norms in the form of the police and the courts 
(because such things did not exist in the ancestral environment). The fact that 
criminals today can have lower repro­ductive success than law-abiding citizens 
is immaterial for the claim that the psycho­logical mechanism that produces 
criminal behavior was once adaptive in the ancestral environment.
The logic of the theory requires that this psychological mechanism have evolved 
before informal norms against violence and theft emerged in the protohuman 
pri­mate society in the course of evolution. Such psychological mechanism could 
not
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272 265
have emerged after the emergence of norms against violence and theft, because 
then men would not be able to attract mates by eliminating competitors through 
violence and accumulating resources through theft. In the context of such 
informal norms, men with tendencies toward violence and theft would be 
ostracized and would not have attained greater reproductive success.2 In fact, 
the norms against violence and theft probably emerged in response to men's 
evolved psychological mechanism that compels them to behave in antisocial ways. 
The fact that violent and predatory acts that would be classi.ed as criminal if 
committed by humans are quite common among nonhuman species that do not have 
informal norms against such acts (Ellis, 1998) supports this speculation.
I suggest that the age-genius curve looks similar to the age-crime curve 
because the same psychological mechanism that compels men to commit crimes also 
compel them to make great scienti.c contributions and express their genius in 
other forms. This also explains why men far outnumber women both in crime and 
in various ex­pressions of genius. Miller (1999, 2000) argues that the 
production of jazz music, modern paintings and books is an example of "cultural 
display" designed to attract mates. I contend, counterintuitive though it might 
sound at .rst, that the same psy­chological mechanism that compels men to 
engage in cultural display in order to at­tract mates, by producing cultural 
products or making scienti.c discoveries, also compel other men to engage in 
criminal activities. Both crime andgenius are expres­sions of young men’s 
proximate competitive desires, whose ultimate function in the an­cestral 
environment wouldhave been to increase reproductive success.
I contend that productivity (observable expressions of genius such as scienti.c 
dis­coveries, jazz albums, paintings, and books) is a function of two 
components: Genius and e.ort. Genius (or talent in some endeavor), while 
unobservable, clearly varies be­tween individuals. Some have it, others do not. 
Further, di.erent people have genius in di.erent endeavors. J.D. Salinger could 
not have been the .fth Beatle; Paul Mc­Cartney could not have written The 
Catcher in the Rye. E.ort, I contend, results from competitiveness, and all men 
have the universal age pro.le of competitiveness, which is probably identical 
to the age-crime curve and peaks in late adolescence and early adulthood. From 
this perspective, genius per se does not have to decline with age. It is 
instead the life-course .uctuations in e.ort (competitiveness) that makes 
productivity fade with age. Paul McCartney probably still has the genius which 
would allow him to write another Yesterday; he just does not feel like it, 
especially after his recent remarriage (see below).
Crime may be thought of as the "default" expression of male competitiveness, in 
two senses. First, unlike scienti.c and artistic endeavors, crime (young men 
killing each other to get access to available women) probably happened in the 
ancestral en­vironment. (Our ancestors might have had primitive art and music, 
but they certainly did not produce CDs, portraits, and books.) Second, once 
again unlike scienti.c and artistic endeavors, criminal behavior does not 
require any special talent (or "Genius" in the equation: Productivity ¼ Genius 
+ E.ort). This is why I believe the age-crime
2
I thank Barbara J. Costello and Allan Mazur for independently making this 
point.
curve more closely resembles the age pro.le of competitiveness in men's life 
course than the age-genius curves. Crime is the product of men's 
competitiveness when they have no genius (that is, when genius ¼ 0 in the 
equation Productivity ¼ Genius + E.ort). This is consistent with the well-known 
fact that criminals on average have lower intelligence than noncriminals.
Today, men can express their competitiveness ("e.ort") in evolutionarily novel 
ways in science, music, art and literature, if they have talent ("genius") in 
these endeavors. This is probably why the age-genius curves (in Figs. 1-4) peak 
some­what later than the age-crime curve (Fig. 5). Productivity in arts and 
sciences, un­like crime, requires men to respond to evolutionarily novel 
stimuli and situations, and their response to such evolutionarily novel 
environments might be delayed. Their evolved psychological mechanism 
(competitive urge) may not respond to evolutionarily novel pursuits such as 
science and art as quickly or reliably. This is similar to the fact that our 
desire to reproduce, which we share with and inherit from our ancestors, is 
expressed much later in our lives (in terms of actual repro­duction), compared 
to our ancestors, in the evolutionarily novel environment of post-industrial, 
monogamous society with compulsory education and reliable con­traception. 
Likewise, the competitive urge of men who lack talent in any endeavors is 
expressed earlier in the evolutionarily familiar, default form of crime and 
vio­lence, but the same competitive urge of men who have talent in some 
endeavors is expressed somewhat later in evolutionarily novel forms of science, 
music, art and literature.
Consistent with this reasoning, there is evidence to show that criminals, whose 
productivity peaks early, also marry earlier than noncriminals. In their 
prospective longitudinal study of 500 delinquents and 500 nondelinquents in the 
Boston area, Glueck and Glueck (1968) show that delinquent men on average marry 
earlier than their nondelinquent counterparts. For instance, more than twice as 
many delin­quents marry at age 18 or younger as nondelinquents do (7.4% vs. 
3.6%) while a lar­ger proportion of nondelinquents postpone their .rst marriage 
until after 25 than do delinquents (33.8% vs. 28.1%) (v2 ¼ 11:01; p <:05) 
(Glueck & Glueck, 1968, p. 82, Table VIII-3).3
In the ancestral environment, most (if not all) competition between men was 
phys­ical and its potential costs included death and physical injury. This is 
why men be­come increasingly less competitive as they age, because they must 
shift their reproductive e.ort from mating to parenting once they have 
children, and dead or injured men do not make good fathers (see Fig. 6). This 
is no longer true in the cur­rent environment, where men compete in scienti.c 
and artistic endeavors. There are no physical costs to competition in these 
evolutionarily novel endeavors; scientists do not literally perish when they 
fail to publish. However, men's competitive urge, adapted to the ancestral 
environment and the default form of competition (crime
3
One reviewer points out that criminals mostly pursue resources, not status, 
whereas artists and scientists mostly pursue status, not resources. This 
di.erence in reproductive strategy can also potentially account for the 
di.erence in age peaks between crime and genius curves, if it takes men longer 
to attain status than resources.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
and violence) nonetheless compels them to desist from competition as they get 
older, if more gradually than was the case in the ancestral environment. Their 
evolved psy­chological mechanism compels them to act as if competition always 
carries physical costs.
Miller (1999, 2000) argues that women judge men's underlying genetic quality by 
their "cultural displays" of artistic expressions. In the course of sexual 
selection, wo­men have been selected to be attracted to men whose competitive 
urge manifests it­self in arts and sciences. Men who can win the Nobel prize or 
the Grammy are obviously more capable than those who cannot. These men will, 
therefore, make bet­ter fathers and providers for their o.spring, even though 
their competitive urge will soon decline after marriage and parenthood, and 
their productivity will fade. How­ever, fathers do not have to win the Nobel 
prize or the Grammy every year to earn su.cient resources to make parental 
investment into the o.spring. Their superior ge­netic quality has already been 
demonstrated when they were young and highly com­petitive. This is why highly 
competitive and successful men (in whatever endeavor) attract mates; they can 
bring in more resources and be better fathers even when they are not being 
highly competitive later in life.
5. The comparable e.ect of marriage on crime and genius
Crime and genius share something else in common: Marriage depresses both. Fig. 
7 presents the age-genius curve separately for scientists who were married 
sometime in their lives (n ¼ 186) and for scientists who remained unmarried for 
their entire lives (n ¼ 72). (I used Debus (1968) and Gillispie (1970-1980) to 
obtain information on the scientists' marital history, but I was not able to 
ascertain the marital history of 22 scientists.) The histograms clearly show 
that the age-genius curve holds only for married scientists. The age-genius 
curve among these scientists is essentially the same as that for the entire 
sample, but the peak occurs a bit earlier in an even quicker burst (mean ¼ 
33.9, median ¼ 32.5; IQR ¼ 11.3).
In contrast, expressions of genius among scientists who never married do not 
de­cline sharply. Half as many (50.0%) unmarried scientists make their greatest 
contri­butions in their late 50s as they do in their late 20s. The 
corresponding percentage among the married scientists is 4.2%. The mean peak 
age among the unmarried sci­entists is 40.0, the median is 38.5, and the IQR is 
16.8. The di.erence in the mean age between the married and unmarried 
scientists is statistically signi.cant (t ¼ 4:83; p <:0001).
Given that science did not exist in the ancestral environment, men's evolved 
psy­chological mechanism appears to be rather precisely tuned to marriage as a 
cue to "desistance." Nearly a quarter (23.4%) of all married scientists make 
their greatest contributions, and thus "desist," within .ve years after their 
marriage. The mean de­lay (the di.erence between their marriage and their peak) 
is mere 2.6 years; the me­dian is 3.0 years. It, therefore, appears that 
scientists rather quickly desist after their marriage, while unmarried 
scientists continue to make great scienti.c contributions later in their lives. 
Similarly, Hargens, McCann, and Reskin's (1978) study demon­
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272

Fig. 7. The age-genius curve among the married and unmarried scientists.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272 269
strates that childless research chemists are more productive than those with 
chil-dren.4
This is exactly the pattern observed among criminals. Criminologists have known 
that one of the strongest predictors of desistance from criminal careers is 
good mar­riage (Laub, Nagin, & Sampson, 1998; Sampson & Laub, 1993). Criminals 
who get married, and especially those who maintain strong marital bonds to 
their wives, sub­sequently stop committing crime, whereas criminals at the same 
age who remain un­married tend to continue their criminal careers.
Sampson and Laub (1993) and Laub et al. (1998) explain the strong desistance 
ef­fect of marriage from the social control perspective (Hirschi, 1969). 
Marriage creates a bond to the conventional society, and investment in this 
bond, in the form of a strong marriage, makes it less likely that the criminal 
would want to remain in the criminal career, which is incompatible with the 
conventional life. Marriage also in­creases the scope and e.ciency of social 
control. Now there is someone living in the same house and monitoring the 
criminal's behavior at all times. It would be more di.cult for the criminal to 
escape the wife's watchful eye and engage in illicit activ­ities.
However, Sampson and Laub's social control theory, and its explanation of the 
desistance e.ect of marriage, could not be the whole answerif marriage has the 
same desistance e.ect on scientists. Unlike criminal behavior, scienti.c 
activities are com­pletely within the conventional society, and are thus not at 
all incompatible with marriage and other strong bonds to conventional society. 
Unlike criminals, scientists are not subject to social control (by their wives 
or otherwise) since scienti.c activities are not illegal or deviant in any way.
I believe an evolutionary psychological theory provides a more parsimonious 
ex­planation for the desistance e.ect of marriage for both crime and science in 
the form of a single psychological mechanism that compels young men to compete 
and excel early in their adulthood but subsequently turns o. after the birth of 
their children. Further, there seems to be a biochemical microfoundation to the 
desistance e.ect of marriage. David Gubernick's unpublished experiment 
(discussed in Blum, 1997,
p. 116) demonstrates that the testosterone levels of expectant fathers 
precipitously drop right after the birth of their children. Mazur and Michalek 
(1998) show that marriage decreases, and divorce increases, men's testosterone 
levels. If high levels
4
Contemporary readers might suggest that unmarried scientists continue to make 
scienti.c contributions much later in their lives because they have more time 
to devote to their careers. Unmarried, and therefore childless, scientists do 
not have to spend time taking care of their children, driving them back and 
forth between their soccer practices and ballet lessons, or doing half of the 
household chores, and that's why unmarried scientists can continue making great 
contributions whereas married scientists must desist. This is precisely Hargens 
et al.'s (1978) interpretation of the negative correlation between parenthood 
and productivity among research chemists. I would remind the readers, however, 
that almost all the scientists in my sample lived in the 18th and 19th century, 
when married men made very little contribution in the domestic sphere and their 
wives did not have their own careers. Hargens et al.'s data come from 1969 and 
1970, when this was probably still true to a large extent. I would, therefore, 
contend that, if anything, married scientists probably had more (rather than 
less) time to devote to science, because they had someone to take care of their 
domestic needs at all times.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
of testosterone predispose men to be more competitive, then the sudden drop in 
tes­tosterone after their marriage and the birth of their children might 
provide the bio­chemical reason why men's psychological mechanism to commit 
crime or make great scienti.c discoveries "turns o." when they get married and 
become fathers, and si­multaneously why the same mechanism does not "turn o." 
when the men (be they criminals or scientists) do not get married.
Now there are other phenomena which exhibit similar age distributions, such as 
automobile accidents, and other risk-taking behavior. In fact, men who engage 
in crime and deviance are also prone to have accidents and engage in 
risk-taking behav­ior (Hirschi & Gottfredson, 1994). Criminologists have known 
that criminals do not specialize; men who engage in one type of crime also 
engage in many others. I believe it is entirely possible that di.erent types of 
crime and deviance, accidents and other forms of risk-taking behavior are all 
manifestations of the same underlying psycho­logical mechanism that compels 
young men to be highly competitive. For one thing, we know from automobile 
insurance statistics that marriage depresses men's ten­dency to have automobile 
accidents.
6. Conclusion
Perhaps the tragic life of the French mathematician
E
Evariste Galois (1811-1832) best illustrates my argument (Singh, 1997, pp. 
210-228). Despite the fact that he died at age 20, Galois made a large number 
of signi.cant contributions to mathematics. (His work was integral to Andrew 
Wiles' celebrated proof of Fermat's Last Theorem in 1994.) Galois was involved 
in an a.air, and the woman's.ance
challenged him to a duel. The night before the duel, Galois stayed up all night 
and wrote down all of his mathematical ideas on paper. (It is due to these 
notes, written on the last night of his life, that many of Galois' ideas 
survived to the posterity.) From other comments written on the paper, next to a 
series of mathematical notations, however, it is clear that Galois spent the 
night, intensely thinking about the woman over whom he was to have a duel the 
next morning. Something compelled this young man of 20 to pro­duce so many 
brilliant mathematical ideas in one night and then go to a duel the next 
morning, ready to kill or be killed over a woman. It is my contention that the 
same psychological mechanism was responsible for both.
If the age-crime curve and the age-genius curve have similar shapes, and if 
mar­riage has the desistance e.ect on both crime and genius, then it is highly 
unlikely that social control theory of criminal behavior and desistance (Laub 
et al., 1998; Sampson & Laub, 1993), or, for that matter, any theory that is 
speci.c to criminal behavior, can hold the whole key to why men commit crimes 
and why they desist. Following Daly and Wilson (1988) and Kanazawa and Still 
(2000), I argue that a single psychological mechanism is responsible for making 
young men highly competitive during early adulthood and then quickly making 
them desist after their marriage in later adult­hood. It is my contention that 
both crime and genius are manifestations of young men's competitive desires to 
gain access to women's reproductive resources, which, in the ancestral 
environment, would have increased their reproductive success.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
Acknowledgments
I thank Barbara J. Costello, Steven W. Gangestad, Travis Hirschi, Rosemary L.
Hopcroft, Christine Horne, Alan S. Miller, Joanne Savage, and Dean Keith Simon-
ton for their comments on earlier drafts.
References
Blum, D. (1997). Sex on the brain: The biological di.erences between men and 
women. New York: Penguin. Blumstein, A. (1995). Youth violence, guns, and the 
illicit-drug industry. Journal of Criminal Law and
Criminology, 86, 10-36.
Brodetsky, S. (1942). Newton: Scientist and man. Nature, 150, 698-699.
Cole, S. (1979). Age and scienti.c performance. American Journal of Sociology, 
84, 958-977.
Daly, M., & Wilson, M. (1988). Homicide. New York: De Gruyter.
Daly, M., & Wilson, M. (1990). Killing the competition: Female/female and 
male/male homicide. Human Nature, 1, 81-107.
Debus, A. G. (Ed.). (1968). Worldwho's who in science: A biographical 
dictionary of notable scientists from antiquity to the present. Chicago: A.N. 
Marquis.
Ellis, L. (1998). Neodarwinian theories of violent criminality and antisocial 
behavior: Photographic evidence from nonhuman animals and a review of the 
literature. Aggression andViolent Behavior, 3, 61-110.
Gillispie, C.C., (Editor in Chief.) 1970-1980. Dictionary of scienti.c 
biography (16 Volumes.) New York: Charles Scribner's Sons.
Glueck, S., & Glueck, E. (1968). Delinquents andnond
elinquents in perspective. Cambridge: Harvard University Press. Hargens, L. L., 
McCann, J. C., & Reskin, B. F. (1978). Productivity and reproductivity: 
Fertility and
professional achievement among research scientists. Social Forces, 57, 154-163.
Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California 
Press.
Hirschi, T., & Gottfredson, M. (1983). Age and the explanation of crime. 
American Journal of Sociology, 89, 552-584.
Hirschi, T., & Gottfredson, M. R. (Eds.). (1994). The generality of deviance. 
New Brunswick: Transaction Publishers.
Kanazawa, S. (2001). De Gustibus Est Disputandum. Social Forces, 79, 1131-1163.
Kanazawa, S., & Still, M. C. (2000). Why men commit crimes (and why they 
desist). Sociological Theory, 18, 434-447.
Laub, J. H., Nagin, D. S., & Sampson, R. J. (1998). Trajectories of change in 
criminal o.ending: Good marriages and the desistance process. American 
Sociological Review, 63, 225-238.
Lehman, H. C. (1953). Age andachievement. Princeton: Princeton University 
Press.
Levin, S. G., & Stephan, P. E. (1991). Research productivity over the life 
cycle: Evidence for academic scientists. American Economic Review, 81, 114-132.
Mazur, A., & Michalek, J. (1998). Marriage, divorce, and male testosterone. 
Social Forces, 77, 315-330.
Miller, G. F. (1999). Sexual selection for cultural display. In R. Dunbar, C. 
Knight, & C. Power (Eds.), The evolution of culture: An interdisciplinary view 
(pp. 71-91). New Brunswick: Rutgers University Press.
Miller, G. F. (2000). The mating mind: How sexual choice shaped the evolution 
of human nature. New York: Doubleday.
Mo.tt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial 
behavior: A developmental taxonomy. Psychological Review, 100, 674-701.
Mukerjee, M. (1996). Explaining everything. Scienti.c American, 274, 88-94.
Porter, R. (Ed.). (1994). The biographical dictionary of scientists (2nd ed). 
New York: Oxford University Press.
S. Kanazawa / Journal of Research in Personality 37 (2003) 257-272
Poundstone, W. (1992). Prisoner's dilemma. New York: Anchor.
Sampson, R. J., & Laub, J. H. (1993). Crime in the making: Pathways andturning 
points through life. Cambridge: Harvard University Press.
Simonton, D. K. (1988). Age and outstanding achievement: What do we know after 
a century of research? Psychological Bulletin, 104, 251-267.
Simonton, D. K. (1997). Creative productivity: A predictive and explanatory 
model of career trajectories and landmarks. Psychological Review, 104, 66-89.
Singh, S. (1997). Fermat's enigma: The epic quest to solve the world's greatest 
mathematical problem. New York: Anchor.