The mechanism of emergenesis


  • D. T. Lykken

    Corresponding author
    1. Department of Psychology, University of Minnesota, MN, USA
      *D. T. Lykken, 825 Summit Avenue, Apt. 1106, Minneapolis, MN 55403, USA. E-mail:
    Search for more papers by this author

*D. T. Lykken, 825 Summit Avenue, Apt. 1106, Minneapolis, MN 55403, USA. E-mail:


Since each individual produced by the sexual process contains a unique set of genes, very exceptional combinations of genes are unlikely to appear twice even within the same family.

E. O. Wilson (1978)

The intraclass correlations of monozygotic twins who were separated in infancy and reared apart (MZA twins) provide estimates of trait heritability, and the Minnesota Study of Twins Reared Apart [MISTRA: Bouchard et al. (1990), The sources of human psychological differences: the Minnesota study of twins reared apart, Science 250, 223–228] has demonstrated that MZA pairs are as similar in most respects as MZ pairs reared together. Some polygenic traits—e.g. stature, IQ, harm avoidance, negative emotionality, interest in sports—are polygenic-additive, so pairs of relatives resemble one another on the given trait in proportion to their genetic similarity. But the existence and the intensity of other important psychological traits seem to be emergent properties of gene configurations (or configurations of independent and partially genetic traits) that interact multiplicatively rather than additively. Monozygotic (MZ) twins may be strongly correlated on such emergenic traits, while the similarity of dizygotic (DZ) twins, sibs or parent–offspring pairs may be much less than half that of MZ pairs. Some emergenic traits, although strongly genetic, do not appear to run in families. MISTRA has provided at least two examples of traits for which MZA twins are strongly correlated, and DZA pairs correlate near zero, while DZ pairs reared together (DZTs) are about half as similar as MZTs. These findings suggest that even more traits may be emergenic than those already identified. Studies of adoptees reared together (who are perhaps more common than twins reared apart) may help to identify traits that are emergenic, but that also are influenced by a common rearing environment.

It is by now generally accepted that most consistent individual differences— psychological, physiological or psychophysiological—are from partly to strongly heritable. The within-pair similarity of monozygotic (MZ) twins who were separated in infancy and reared apart (MZA twins), who share all their genes but minimal shared environmental influence, provides one estimate of heritability but, because MZA correlations for most traits are approximately equal to those for MZ pairs reared together (Bouchard et al. 1990), these more accessible MZT pairs constitute the main heritability database.

The simplest examples of heritable traits, such as stature, are polygenic-additive: different genes determine the vertical dimensions of the head, neck, trunk and the long bones of the legs, and their contributions to the sum of these dimensions are additive. Because MZ twins share all their genes, while dizygotic (DZ) twins share, on average, only half, the within-pair similarity of DZ twins on additive traits is expected to be about half that of MZ pairs. But DZ pairs are often less than half as similar as MZ pairs, and geneticists early discovered two mechanisms that can yield this result. One is dominance and the other epistasis, defined as an interaction between non-allelic genes, especially (as in dominance) an interaction in which one gene suppresses the expression of another. In fact, the Greek word epistanai means to stop or check, as in another, medical definition of epistasis, viz. the suppression of a bodily discharge or secretion.

But a more frequent and important form of polygenic interaction is when gene effects combine in a multiplicative fashion, rather like workers on an assembly line, each of whose contributions may be qualitatively essential to the final product. The term emergenesis (Lykken 1982; Li 1987; Lykken et al. 1992) has been proposed—and used mainly, thus far, in the psychological literature—as more general and descriptively more accurate than epistasis.

Emergenic (adj) Arising as a novel or emergent property resulting from the interaction of more elementary and partly genetic properties. An emergenic trait may be determined by a configuration of independently segregating polygenes interacting in a multiplicative rather than an additive fashion, or, at a more molar level, an emergenic trait might be a property of a configuration of independent traits that are themselves partly genetically determined. In some instances, situational factors may figure as configural components. The distinctive feature of emergenesis is the notion of configurality, which implies that a change of any one component may result in a qualitative, or a large quantitative, change in the emergenic trait
(Lykken et al. 1992; p.1569.)

In a study of 221 pairs of twins, plus 3924 additional sibling or parent–offspring pairs, Heath et al. (1984) reported that the total number of fingerprint triradii yields MZ twin correlations of 0.91, while DZ twins, sibs and parent–child (PC) pairs all average about 0.29. Because the PC pairs were about as similar to each other as were the DZ twins, thus excluding dominance as the mechanism, the authors claimed these findings to be ‘the strongest evidence so far for polygenic epistasis in man.’ (p. 719). Because Heath et al. speak of ‘additive × additive epistasis’, rather than the non-allelic dominance mechanism suggested by the Greek source, they seem to regard fingerprint triradii frequency as what I would call an emergenic trait.

But there are many emergenic traits that are of greater interest to behavior geneticists than fingerprint triradii. Research with twins, including twins separated in infancy and reared apart, has identified numerous important emergenic traits of personality (e.g. neuroticism, aggressiveness, impulsiveness, positive emotionality; see Bouchard & Lykken 1999; Grove et al. 1990; Keller et al. 2005; McGue et al. 1993; Tellegen et al. 1988), and interests (e.g. arts and crafts, husbandry; see Lykken et al. 1993; Waller et al. 1995). Using data from MISTRA, Bouchard & Lykken (1999) report on the two different measures of creativity, MZA and DZA correlations of 0.63 and 0.06, respectively, and, on the psychopathic deviate, psychasthenia and schizophrenia scales of the Minnesota Multiphasic Personality Inventory (MMPI), mean MZA and DZA correlations of 0.62 and 0.12, respectively. University of Minnesota students majoring in music had significantly more first-degree relatives who were musically talented than did non-music majors, but those majoring in voice did not have more singers in their families than did those majoring in instrumental music (Stearns S & VanderWoude A, unpublished data); musical talent runs in families, but voice quality is a multiplicative function of properties of the lungs, throat, vocal chords, larynx, palate, uvula, etc., and is an emergenic trait.

Not only traits of interest and personality but even some social tendencies, such as risk for divorce, and some social attitudes, such as right-wing authoritarianism, have been shown to be not only strongly heritable but also apparently emergenic. McCourt et al. (1999) administered Altemeyer's right wing authoritarianism scale (RWA, Altemeyer 1988) to a large sample of reared-together, middle-aged twins, getting the twin correlations summarized in Table 1.

Table 1.  Reared-together twins
 MZT TwinsDZT Twins
Number of pairs423434
Within-pair correlations0.63 (0.57–0.68)0.42 (0.34–0.49)

Thus, it would at first appear that RWA is heritable as a polygenic-additive trait. Because DZ twins share, on average, at least half of their polymorphic genes, they should be at least half as similar, in respect to such polygenic-additive traits, as are the MZs, who share all their genes. But these DZs were even more than half as similar as the MZs. This might be due to the fact that husbands and wives tend to be nearly as similar in RWA as those MZ twins are, and such assortative mating increases the genetic similarity of the offspring of those parents. Another possibility is that growing up together in the same family (and the same neighborhood) tends to make DZ twins more than half as similar as MZ twins in the extent of their authoritarian attitudes. It makes sense, of course, that this centripetal influence would be stronger for those DZ pairs whose genetic difference is larger than for those who happen to be genetically similar.

However, McCourt et al. (1999) also administered the RWA scale to 77 pairs of reared-apart twins from MISTRA. These pairs had the same genetic similarity (although perhaps not as much parental similarity) as did the twins reared together, but they were influenced not by one another nor by the same parents while growing up.

In Table 2, those high MZA correlations corroborate that RWA is strongly influenced by genetic effects, yet the similarity of the DZA twins is not significantly different from zero. One interpretation of these two sets of results would be that RWA is emergenic but also subject to a strong centripetal environmental influence in sibs reared together.

Table 2.  Reared-apart twins
 MZT TwinsDZT Twins
Number of pairs3938
Within-pair correlations0.69 (0.48–0.82)0.00 (−0.31–0.33)

Are there any other variables that act this way? McGue and Lykken (1992) studied risk of divorce in a large sample of middle-aged twins. For twins reared together, divorce risk goes up about 50% if either the parents or the DZ co-twin has divorced (100% if both), and about 250% if the MZ co-twin has divorced. This could be a genetic effect, but it could just as well result from environmental influence. If your parents or your sibling solved the problem by divorce, then divorce may seem to you to be more acceptable or more inevitable than if yours was the first in your family, and your risk might really go up, if your MZ twin gets divorced, because you know that you and your ‘identical’ twin are very likely to want or to dislike the same things. But the MISTRA data again suggest that an emergenic interpretation may be the correct one.

(Table 3) The DZA twins divorced at about the same average rate irrespective of what their co-twins may have done, while the MZA twins, if their co-twins (whom in most cases they had not yet met nor known about) had divorced, were nearly five times more likely to be divorced themselves than if those phantom co-twins had remained married!

Table 3.  Twins reared-apart
 MZT TwinsDZT Twins
Number of pairs where both had married5748
Number of pairs concordant for no divorce3620
Number of pairs concordant for divorce105
Number of pairs discordant for divorce1123
Risk if co-twin not divorced11/(72+11) = 13%23/(40+23) = 37%
Risk if co-twin is divorced20/(20+11) = 65%10/(10+23) = 30%

Thus, both for RWA and divorce risk, it appears that the genetics may be emergenic, but that there is also a strong within-family influence, so that the reared-together data alone look more-or-less like the usual Rmz = 2*Rdz situation, which would suggest additivity. These hypotheses could be tested definitively by studies of unrelated same-sex adoptees reared together in the same homes: do such pairs correlate (nearly) as strongly as DZT pairs in RWA scores and in risk for divorce? Should such tests support the emergenesis interpretation, then there may be many more interesting traits that have been assumed to be polygenic-additive that may in fact be emergenic yet strongly familial. Especially for psychological traits and tendencies, we may be greatly underestimating the frequency and importance of the emergenic effect, because the reared-together data look like additive gene action.

It will be noted from the examples cited that much of the evidence for the more interesting emergenic traits is based on relatively small samples. Twins reared apart are rare, and studies in which many pairs of both MZ and DZ twins are retested over extended intervals are expensive. For most variables, one can be confident of the strong MZ correlations; the problem is to establish that the similarity of the DZ pairs is close to zero. One hopes that future research using sibling or parent–offspring pairs, as did Heath et al. (1984), together with studies of adoptive pairs, will establish more certainly which strongly genetic traits are in fact emergenic. One reason why this is important is that molecular identification of emergenic polygenes is likely to be different and more difficult than the identification of genes that contribute additively to the trait value.

Emergenic psychophysiological traits

One of the first individual difference dimensions to be shown to be emergenic was the mid-frequency (Phi) of the alpha rhythm of the electroencephalogram (EEG: Lykken et al. 1974). In a replication (Lykken et al. 1982), the relative magnitudes of EEG activity within each of the four classical frequency bands gave MZ and DZ correlations averaging 0.80 and 0.49, respectively, indicating additivity, while those for Phi were 0.81 and 0.16, suggesting emergenesis.

The electrical resistance of the thick skin of the palms and soles is extraordinarily reactive to psychological stimulation. The hundreds of sweat glands in these volar regions subserve a special function; instead of helping with thermoregulation, they moisten grasping surfaces in preparation for action. Dry palmar skin is both slippery and more subject to abrasion, and its tactile sensitivity also is reduced. Neural circuits arising in the activating systems of the midbrain control volar sweating, which increases tonically with CNS arousal, and which also shows wavelike, phasic increases in response to any stimulus important enough to produce an orienting response. In part, because the sweat gland tubules provide a low-resistance pathway through the epidermis, the electrical conductance of the skin varies with sweat gland activity.

Volar skin conductance level (SCL) is lowest in a drowsy or somnolent subject, rises sharply with awakening and rises still further during mental effort or emotional storm. Superimposed upon these tidal changes of SCL are the wavelike, phasic skin conductance responses (SCRs) to discrete stimuli. After a latency of perhaps 1.5 s, conductance rises rather quickly to a peak, and then returns more or less rapidly to the prestimulus level. SCR amplitude seems to vary with the psychological impact of the eliciting stimulus; other things being equal, strong stimuli produce larger SCRs than weak ones but an unexpected or especially significant weak stimulus will produce a larger SCR than an expected but meaningless strong stimulus.

When subjects are presented with repeated, unpredictable, loud (105 db) blasts of white noise, their SCRs, at first very large, gradually habituate. Lykken et al. (1988), using both reared-together and reared-apart twins, found that the slopes of these SCR habituation curves gave within-pair correlations of 0.63 and −0.05 for MZ and DZ pairs, respectively. But that same study also displayed other examples of emergenesis which make obvious the configural or multiplicative character of the mechanism.

Using both reared-together and reared-apart twins (Lykken 1982), it was found that, when measured in comparable circumstances, MZ twins correlated 0.61 in both their maximum and minimum levels of SCL and 0.55 in the size of the SCR elicited by a strong stimulus. DZ twin pairs, in contrast, are no more similar in these respects than pairs of unrelated subjects. These findings have been replicated with larger samples (Lykken et al. 1988) and raw SCL and SCR thus qualify as emergenic traits. What is especially interesting about these electrodermal variables is that the source and nature of their emergenesis is easily demonstrated.

Because the same level of arousal produces various levels of SCL in different people, while the same stimulus may elicit an SCR from some subjects that is much larger or smaller than that elicited from others, it is common practice to ‘range-correct’ electrodermal variables (Lykken et al. 1966; Lykken & Venables 1971; Lykken 1972). This is done by expressing SCL values as a fraction of the range between that subject's minimum SCL, obtained during rest toward the end of the experiment, and his maximum SCL, obtained during stress imposed at the start of the session. SCRs, similarly, are range-corrected by expressing them as a fraction of that individual's maximum SCR elicited by a strong, unexpected stimulus. Range-corrected SCL and SCR values show the same strong MZ correlations found with the uncorrected measures, but the DZ correlations for the corrected values now are approximately half the size of the MZ correlations.

Raw SCL is determined by (at least) the subject's current level of arousal, the reactivity of that subject's sudomotor system, the average reactivity of that subject's volar sweat glands to sudomotor innervation, the density of sweat glands in that subject's volar skin and the conductivity of that subject's dry epidermis. Moreover, raw SCL is a function of the product of these several variables, each of which is determined in part by a different group of polymorphic genes.

Sharing all their genes, MZ twins will display similar raw SCLs (or SCRs). DZ twins, in contrast, are likely to differ significantly in one or more of these components, and a significant change in any multiplier will yield a significant difference in the product (the emergenic trait value), whereas the same change in any addend would cause only a slight change in the sum (the polygenic-additive trait value). Range-correction of either SCL or SCR serves to eliminate the effect of individual differences in all these non-psychological variables and reveals that individual differences in arousal or in the strength of the orienting reflex, each measured under standardized conditions, seem to be largely determined by a polygenic-additive mechanism.


A growing number of metrical traits—especially psychological traits such as neuroticism, aggressiveness, impulsiveness, positive emotionality, artistic interests, vocal quality—appear to be determined in part by configurations of genes, or of partly genetic traits, which combine multiplicatively rather than additively. This mechanism, nicely illustrated by electrodermal variables, results in high MZ but low or zero DZ correlations. Because the term epistasis was created to refer to dominance relations between non-allelic genes, the new term emergenesis (thus far seen mainly in the psychological literature) seems a better name for this mechanism. The risk for divorce and the trait of right-wing authoritarianism appear to be covert emergenic traits for which the correlations of DZ twins reared together are strongly augmented by shared environment. Future research may show that many important traits are emergenic; if so, their genes will be harder to identify than is true for Mendelizing or additive traits.