Epistemic status: moderately confident
Robert Kegan’s developmental stages have become popular in my corner of the social graph, and I was asked by Abram Demski and Jacob Liechty to write a literature review (which they kindly funded, before I started my new job) of whether Kegan’s theory is justified. Since Kegan’s model is a composite that builds on many previous psychologists’ work, I had to do an overview of several theories of developmental stages. I cover the theories of Piaget, Kohlberg, Erikson, Piaget, Maslow, and Kegan. All of these developmental stage theories posit that there are various levels of cognitive, moral, or psychological maturity and sophistication; children start at the low levels and progress to the higher ones; only a few of the very “wisest” adults reach the very top stages.
This makes intuitive sense and is a powerful story to tell. You can explain conflicts and seemingly strange behavior by understanding that some people are simply on more primitive levels and cannot comprehend more sophisticated ones. You can be motivated to reach towards self-improvement by a model of a ladder of development.
But for the moment I want to ask more from developmental theories than being interesting or good stories; I want to ask if they’re actually correct.
In order for a developmental theory to be correct, I think a few criteria must be met:
- The developmental stages must be reliably detectable, e.g. by some questionnaire or observational test that has high internal consistency and/or inter-rater reliability
- The developmental stages must improve with age, at least within a given cohort (most people progress to later stages as they grow older)
- The developmental stages must be sequential and cumulative (people must learn earlier stages before later ones, and not skip stages)
- In cases where the developmental stages are supposed to occur at particular ages, they must actually be observed being attained at those ages.
Most of the theories do not appear to meet these criteria.
Jean Piaget was one of the pioneers of child development psychology. Beginning in the 1930’s, his observations of children led him to a sequential theory of how children gain cognitive abilities over time.
Piaget’s stages of cognitive development are:
- Sensorimotor, ages 0-2, hand-eye coordination and goal-directed motion
- Pre-operational, ages 2-7, speech, pretend play and use of symbols
- Concrete operational, age 7-11, inductive logic, perspective-taking
- Formal operational, ages 11-adult, deductive logic, abstraction, metacognition, problem-solving
Piaget’s first study, The Origins of Intelligence in Children, published in 1952, was conducted on his own three children, from birth to age 2. He and his wife made daily observations of the children.
Reflexes, Piaget noticed, are present from birth: the sucking reflex, upon contact with the nipple, happens automatically. In the first month of life, he notes that the babies become more effective at finding the nipple. From one to two months, babies self-stimulate even when there is no breast — they suck their thumbs or make sucking motions on their own. This, Piaget calls the “primary circular reaction”. A reflex has been transformed into a self-generated behavior. At first, the baby can’t reliably find his thumb; he flails his arms until they happen to brush his face, and then engages the sucking reflex. There are “circular reactions” to grasping, looking, and listening as well. Later, babies learn to coordinate these circular reactions across senses, and to move their bodies in order to attain an objective (e.g. reaching for an object to take it).
Some of Piaget’s conclusions have been disputed by modern experiments.
In his studies of infants, he tested their ability to reason about objects by occluding the object from view, subjecting it to some further, hidden motion, and then having the child search for the object.
However, younger infants have less physical ability to search, so this task is less appropriate for assessing what young infants know. In the 1980’s, Leslie and Ballargeon used looking time as a metric for how much infants were surprised by observations; since this doesn’t require physical coordination, it allows for accurate assessment of the cognitive abilities of younger infants. Leslie’s experiments confirmed that babies understand causality, and Ballargeon’s confirmed that babies have object permanence — in both cases, looking times were longer for “impossible” transformations of objects that violated the laws of causality or caused objects to transform when behind a screen. 4-month-old infants are, contra Piaget, capable of object permanence; they understand that objects must move along continuous paths, and that solid objects cannot pass through each other.
Kittens go through Piaget’s sensorimotor stages: first reflexes, then habits (pawing, oscillating head), then secondary circular formations (wrestling, biting, dribbling with objects), and finally means-end coordination (playing hide-and-seek.) This supports the ordering of sensorimotor skills in Piaget’s classification.
A 1976 study of 9-14-year-olds given a test and subjecting it to factor analysis found that there were three axes: formal operational systematic permutations; concrete operational addition of asymmetric relations; and formal operational logic of implications. This supports something like Piaget’s classifications of cognitive tasks.
Different studies conflict on which operational stages come before others: is class inclusion always required before multiplication of classes? Ordinal before cardinal? Logical and number abilities before number conservation? There’s no consistent picture. “By 1970, it was evident in the important book, Measurement and Piaget, that the empirical literature functioned poorly as a data base on which the objective evaluation of Piagetian theory could be effectively attempted (Green, Ford, & Flamer, 1971)…For example, Beard (1963) found that 50% of her 5- to 6-year-old samples conserved quantity (solid). In contrast, Lovell and Ogilvie (1960) and Uzgiris (1964) reported that it is in the 8- to 9-year-old range that children conserve quantity. Elkind (1961) reported that 52% of 6-year-old children conserved weight, but Lovell and Ogilvie (1961) reported this percentage for 10-year-old children.”
According to Piaget’s “structured whole” theory, when children enter a new stage, they should gain all the skills of that stage at once. For instance, they should learn conservation of volume of water at the same time as they learn that the length of a string is conserved. “ However, point synchrony across domains has never been found. To the contrary, children manifest high unevenness or decalage (Feldman 1980, Biggs & Collis 1982, Flavell 1982). Piaget acknowledged this unevenness but never explained it; late in his life he asserted that it could not be explained (Piaget 1971).” However, it’s overwhelmingly true that success at cognitive tasks is age-dependent. On a host of tasks, age is the most potent predictor of performance.
Piaget claimed that children develop cognitive skills in discrete stages, at particular ages, and in a fixed order. None of these claims appear to be replicated across the literature. The weaker claims that children learn more cognitive skills as they grow older, that some skills tend to be learned earlier than others, and that there is some clustering in which children who can perform one skill can also perform similar skills, have some evidentiary support.
Lawrence Kohlberg, working in the 1960’s and 70’s, sought to extend Piaget’s developmental-stage theories to moral as well as cognitive development.
Kohlberg’s stages of moral development are:
- Obedience and punishment (“how can I avoid punishment?”)
- Instrumentalist Relativist (“what’s in it for me?”)
- Interpersonal Concordance (“be a good boy/girl”, conformity, harmony, being liked)
- “Law and Order” (maintenance of the social order)
- Social contract (democratic government, greatest good for the greatest number)
- Universal ethical principles (eg Kant)
The evidence for Kohlberg’s theory comes from studies of how people respond to questions about hypothetical moral dilemmas, such as “Heinz steals the drug”, a story about a man who steals an expensive drug to save his dying wife.
Kohlberg did longitudinal studies of adolescents and adults over a period of six years, in the US, Taiwan, Mexico, and isolated villages in Turkey and the Yucatan. In all three developed-country examples, the prevalence of stages 1 and 2 declined with age, while the prevalence of 5 and 6 increased with age. In the isolated villages, stage 1 declined with age, while stage 3 and 4 increased with age, and stages 5 and 6 were always rare. Among 16-year-olds, Stage 5 was the most common in the US, while stages 3 and 4 were the most common in Taiwan and Mexico; in the isolated villages, Stage 1 was still the most common by age 16.
In adults there was likewise some change in moral development over time — Stage 4 (law and order) increased with age from 16 to 24, in both lower- and middle-class men, and the highest rates of stage 4 were found in the men’s fathers. Most men stabilize at Stage 4, while most women stabilize at stage 3.
Kohlberg’s experiments show that there is change with age in how people explain moral reasoning, which is similar in direction but different in magnitude across cultures.
In subsequent studies from around the world, 85% (out of 20 cross-sectional studies) showed an increase in moral stage with age, and none of them found “stage skipping” (all stages between the lowest and the highest were present.) Contra Kohlberg, most subsequent studies do not show significant sex differences in moral reasoning. There are some cultural differences: stage 1 does not show up in children in Iran or Hutterite children; most folk tribal societies do not have stages 4, 5, or 6 at all.
Subsequent studies have shown that children do in fact go through Kohlberg’s stages sequentially, usually without stage skipping or regression.
Juvenile delinquents have lower scores on Kohlberg’s moral development test than nondelinquents; moreover, the most psychopathic delinquents had the lowest scores.
Jonathan Haidt has critiqued Kohlberg’s theory, on the grounds that people’s verbal reasoning process for justifying moral hypotheticals does not drive their conclusions. In hypothetical scenarios about taboos — like a pair of siblings who have sex, using birth control and feeling no subsequent ill effects — people quickly assert that incest is wrong, but can’t find rational explanations to justify it. People’s affective associations with taboo scenarios (such as claiming that it would upset them to watch) were better predictors of their judgments than their assessments of the harm of the scenarios.
If the social intuitionists like Haidt are correct, then research in Kohlberg’s paradigm may tell us something about people’s verbal narratives about morality, but not about their decision-making process.
There is also the possibility that interviews about hypotheticals are not good proxies for moral decision-making in practice; people may give the explanations that are socially desirable rather than the real reasons for their judgments, and their judgments about hypotheticals may not correspond to their actions in practice.
Still, Kohlberg’s stages are an empirical phenomenon: there is high inter-rater reliability, people advance steadily in stage with age (before stabilizing), and industrialized societies have higher rates of the higher stages.
Erik Erikson was a psychoanalyst who came to his own theory of stages of psychosocial development in the 1950’s, in which different stages of life force the individual to confront different challenges and develop different “virtues.”
Erikson’s developmental stages are:
- Trust vs. Mistrust (infancy, relationship with mother, feeding and abandonment)
- Autonomy vs. Shame (toddlerhood, toilet training)
- Initiative vs. Guilt (kindergarten, exploring and making things)
- Industry vs. Inferiority (grade school, sports)
- Identity vs. Role Confusion (adolescence, social relationships)
- Intimacy vs. Isolation (romantic love)
- Generativity vs. Stagnation (middle age, career and parenthood)
- Ego integrity vs. Despair (aging, death)
This theory had its origins in subjective clinical impressions. There has been some attempt to correlate a measure of identity achievement with other positive attributes, but, for instance, it has no association with self-esteem or locus of control, which would seem counterintuitive if the “identity achievement” score really corresponded to the development of an independent self.
A self-report questionnaire, in which people rated themselves on Trust, Autonomy, Initiative, Industry, Identity, and Intimacy, was found to have moderately high Kronbach alpha scores (0.57-0.75). Males scored higher on autonomy and initiative, while females scored higher on intimacy, as you’d expect from sex stereotypes.
Two studies, one of 394 inner-city men, and one of 94 college sophomore men, classified them as stage 4 if they never managed to live independently or made lasting friendships, stage 5 if they managed to live apart from their family of origin and become financially independent, stage 6 if they lived with a wife or partner, and stage 7 if they had children, managed others at work, or otherwise “cared for others”. They added a stage 6.5 for career consolidation. Adult life stages in this sense were independent of chronological age, and men who didn’t master earlier stages usually never mastered later ones.
Erikson’s Stage 5, identity development, has some observational evidence behind it; children’s spontaneous story-telling exhibits less concern with identity than adolescents’. One researcher “found the white adolescents to show a pattern of “progressive identity formation” characterized by frequent changes in self-concept during the early high school years followed by increasing consistency and stability as the person approached high school graduation. In contrast, the black adolescents showed a general stability in their identity elements over the entire study period, a pattern Hauser termed “identity foreclosure.” He interpreted this lack of change as reflecting a problem in development in that important developmental issues had been dodged rather than resolved.” Of course, it may also mean that “identity development” is culturally contingent rather than universal.
A study that gave 244 undergraduates a questionnaire measuring the Eriksonian ego strengths found that “purpose in life, internal locus of control, and self-esteem bore strong positive relations with all of the ego strengths, with the exception of care.” But there were no significant correlations between the ego strengths and age, nor any indication that they are achieved in a succession.
A study giving 1073 college students an Erikson developmental stage questionnaire found that it did not fit the “simplex” hypothesis (where people’s achievement of stage n would depend directly on how well they’d achieved step n-1, and less on other stages.)
A 22-year longitudinal study showed that people continued to develop higher scores on Erikson developmental-stage questionnaires between the ages of 20 and 42, even “younger” stages; there was a significant increase over time in stages 1, 5, and 6, for several cohorts.
While people do seem to in some cases gain more of Erikson’s ego strengths over time, this finding is not reliable in all studies. People do not climb Erikson’s stages in sequence, or at fixed ages.
Psychologist Abraham Maslow, inspired by the horrors of war to learn about what propels people to “self-actualized”, developed the concept of a “hierarchy of needs” in which the lower ones must be fulfilled before people can pursue the higher ones.
Maslow’s hierarchy of needs are:
- Physiological (food, air, water)
- Safety (security from violence, disease, or poverty)
- Love and belonging
- Esteem (self-respect, respect from others)
- Self-actualization (realizing one’s potential)
The theory is that lower needs, when unsatisfied, “dominate” higher needs — that one cannot focus on esteem without first satisfying the need for safety, for instance. Once one need is satisfied, the next higher need will “activate” and start driving the person’s actions.
Three researchers (Alderfer, Huizinga, and Beer) developed questionnaires designed to measure Maslow’s needs, but all had weaknesses, “particularly a low convergence among items designed to measure the same constants.” None of the studies showed Maslow’s five needs as independent factors. Both adjacent and nonadjacent needs overlap, contradicting Maslow’s theory that needs are cumulative. People also do not rank the importance of those needs according to Maslow’s order. Also the “deprivation/domination” paradigm (that, the more deprived you are of a need, the higher its importance to you) is contradicted by studies that show that this is not true for safety, belonging, and esteem needs. The “gratification/activation” theory, that when need n is satisfied, need n becomes less important and need n + 1 becomes more important, was also not borne out by studies.
The author of the review concludes, “Maslow’s Need Hierarchy Theory is almost a nontestable theory…Maslow (1970) criticized what he called the newer methods of research in psychology. He called for a “humane” science. Accordingly, he did not attempt to provide rigor in his writing or standard definitions of concepts. Further, he did not discuss any guides for empirical verification of his theory. In fact, his defense of his theory consisted of logical as well as clinical insight rather than well-developed research findings.”
However, a more recent study of 386 Chinese subjects found Cronbach alpha scores in the 80-90% range, positive correlations between the satisfaction of all needs, and higher correlations between the satisfaction of adjacent needs than nonadjacent needs. This seems to suggest a stagelike progression, although the satisfaction of all needs still overlap. Also, satisfaction of the physiological needs was a predictor of the satisfaction of every one of the four higher-level needs.
A global study across 123 countries found that subjective wellbeing, positive feelings, and negative feelings were all correlated in the expected ways with certain universal needs: basic needs for food and shelter, safety and security, social support and love, feeling respected and pride in activities, mastery, and self-direction and autonomy. The largest proportion of variance explained globally in life evaluation was from basic needs, followed by social, mastery, autonomy, respect, and safety. The largest proportion of variance explained in positive emotions was from social and respect. The largest proportion of variance explained in negative emotions was from basic needs, respect, and autonomy. There are “crossovers”, people who have fulfillment of higher needs but not lower ones: “For example, respect is frequently fulfilled even when safety needs are not met.”
It is unclear whether Maslow’s needs are distinct natural categories, and it is clear that they do not have to be satisfied in sequence, that the most important needs to people are not necessarily the lowest ones or the ones they lack most, and that people do not develop stronger drives towards higher needs when their lower needs are fulfilled. The only part of Maslow’s theory that is borne out by evidence is that people around the world do, indeed, value and receive happiness from all Maslow’s basic categories of needs.
Robert Kegan is not an experimental psychologist, but a practicing therapist, and his books are works of interpretation rather than experiment. He integrates several developmental-psychology frameworks in The Evolving Self, such as Piaget, Kohlberg, and Maslow.
Kegan’s stage 0 is “Incorporative” — babies, corresponding to Piaget’s sensorimotor stage, no real social orientation.
Stage 1 is “impulsive”, corresponding to Piaget’s concrete operational stage, Kohlberg’s punishment/obedience orientation, and Maslow’s physiological satisfaction orientation: the subject is impulses, the objects are reflexes, sensing and moving. This is roughly toddlers.
Stage 2 is “imperial”, corresponding to Piaget’s concrete operational stage, Kohlberg’s “instrumental” orientation, and Maslow’s safety orientation; this is roughly grade-school-aged children. The subject is needs and wishes, the objects are impulses.
Stage 3 is “interpersonal”, corresponding to Piaget’s early formal operational, Kohlberg’s interpersonal concordance orientation, and Maslow’s belongingness orientation. The subject is mutuality and interpersonal relations, the objects are needs and wishes. These are young teenagers.
Stage 4 is “institutional”, corresponding to Piaget’s formal operational, Kohlberg’s social contract orientation, and Maslow’s self-esteem orientation. The subject is personal autonomy, the objects are mutuality and interpersonal relations. This is usually young adulthood and career socialization.
Stage 5 is “interindividual”, corresponding to Maslow’s self-actualization orientation and Kohlberg’s principled orientation. This is usually mature romantic relationship.
The Subject Object Interview is Kegan’s scale for measuring progression along the stages.
In a study of West Point students, average inter-rater agreement on the Subject-Object Interview was 63%, and students developed from stage 2 to stage 3 and from stage 3 to stage 4 over their years in school. Kegan stage in senior year had a correlation of 0.41 with MD (military development) grade.
A study of 67 executives found that Kegan stage was correlated with leader performance at a p < 0.05 level; Kegan stage was also positively correlated with age.
I was not able to find any studies that indicated whether people skip Kegan stages, regress in stage, or exhibit characteristics of other stages, or other psychometric instruments that decompose into Kegan stages with factor analysis. Kegan’s stages do appear to be relatively observable and higher stage seems to correspond fairly well with external evaluations of leadership skill.
Piaget’s stages are not distinct (they overlap) or sequential (they can be skipped or attained in different orders. Later stages do correlate with greater age, but the stages do not arise at consistent ages.
Kohlberg’s stages are sequential; they are defined as distinct by the measurement instrument; and they increase with age (as well as with social class and socioeconomic development of the community.) Stages don’t arise at fixed ages.
Erikson’s stages do not appear to be distinct, sequential, or even consistently increasing with age.
Maslow’s needs do not appear to be sequentially satisfied.
Kegan’s stages are defined to be distinct by the measurement instrument, and they increased with age in two studies. I could not find evidence that they are attained sequentially.
Overall, the experimental evidence that distinct, cumulative stages of human development exist is rather weak. The strongest evidence is for Kohlberg’s stages, and these (like all the other stages considered) are limited by the fact that they are measures of how people talk about moral decision-making, rather than what they decide in practice.
Higher stages correlate with positive results in many cases: people at higher Kohlberg stages are less likely to be criminals or delinquents, positive psychological strengths like self-esteem correlate with the Eriksonian ego strengths, and leadership development measures correlate with Kegan stage. This is evidence that developmental stages do often correspond to real psychological strengths or skills with external validity. We just don’t generally have strong reason to believe that they progress in a developmental fashion.
 Piaget, Jean. The origins of intelligence in children. Vol. 8. No. 5. New York: International Universities Press, 1952.
Spelke, Elizabeth S. “Physical knowledge in infancy: Reflections on Piaget’s theory.” The epigenesis of mind: Essays on biology and cognition (1991): 133-169.
Dumas, Claude, and François Y. Doré. “Cognitive development in kittens (Felis catus): An observational study of object permanence and sensorimotor intelligence.” Journal of Comparative Psychology 105.4 (1991): 357.
Gray, William M. “The Factor Structure of Concrete and Formal Operations: A Confirmation of Piaget.” (1976).
Shayer, Michael, Andreas Demetriou, and Muhammad Pervez. “The structure and scaling of concrete operational thought: Three studies in four countries.” Genetic, Social, and General Psychology Monographs 114.3 (1988): 307-375.
Fischer, Kurt W., and Louise Silvern. “Stages and individual differences in cognitive development.” Annual Review of Psychology 36.1 (1985): 613-648.
Kohlberg, Lawrence. “Stages of moral development.” Moral education 29 (1971).
Snarey, John R. “Cross-cultural universality of social-moral development: a critical review of Kohlbergian research.” Psychological bulletin 97.2 (1985): 202.
Walker, Lawrence J. “The sequentiality of Kohlberg’s stages of moral development.” Child Development (1982): 1330-1336.
Haidt, Jonathan. “The emotional dog and its rational tail: a social intuitionist approach to moral judgment.” Psychological review 108.4 (2001): 814.
Chandler, Michael, and Thomas Moran. “Psychopathy and moral development: A comparative study of delinquent and nondelinquent youth.” Development and Psychopathology 2.03 (1990): 227-246.
Rosenthal, Doreen A., Ross M. Gurney, and Susan M. Moore. “From trust on intimacy: A new inventory for examining Erikson’s stages of psychosocial development.” Journal of Youth and Adolescence 10.6 (1981): 525-537.
Vaillant, George E., and Eva Milofsky. “Natural history of male psychological health: IX. Empirical evidence for Erikson’s model of the life cycle.” The American Journal of Psychiatry (1980).
Waterman, Alan S. “Identity development from adolescence to adulthood: An extension of theory and a review of research.” Developmental psychology 18.3 (1982): 341.
Markstrom, Carol A., et al. “The psychosocial inventory of ego strengths: Development and validation of a new Eriksonian measure.” Journal of youth and adolescence 26.6 (1997): 705-732.
Thornburg, Kathy R., et al. “Testing the simplex assumption underlying the Erikson Psychosocial Stage Inventory.” Educational and psychological measurement 52.2 (1992): 431-436.
Whitbourne, Susan K., et al. “Psychosocial development in adulthood: A 22-year sequential study.” Journal of Personality and Social Psychology 63.2 (1992): 260.
Wahba, Mahmoud A., and Lawrence G. Bridwell. “Maslow reconsidered: A review of research on the need hierarchy theory.” Organizational behavior and human performance 15.2 (1976): 212-240.
Taormina, Robert J., and Jennifer H. Gao. “Maslow and the motivation hierarchy: Measuring satisfaction of the needs.” The American journal of psychology 126.2 (2013): 155-177.
Tay, Louis, and Ed Diener. “Needs and subjective well-being around the world.” Journal of personality and social psychology 101.2 (2011): 354.
Lewis, Philip, et al. “Identity development during the college years: Findings from the West Point longitudinal study.” Journal of College Student Development 46.4 (2005): 357-373.
Strang, Sarah E., and Karl W. Kuhnert. “Personality and leadership developmental levels as predictors of leader performance.” The Leadership Quarterly 20.3 (2009): 421-433.