Science doesn't know everything

We understand much about human nature, but scientists are ignoring many important topics, like ethnicity and class

Published June 2, 2013 6:00PM (EDT)


Excerpted from "The Human Spark: The Science of Human Development"

The scientific discoveries of the past century that illuminate our understanding of human nature would have brought full smiles to a host of philosophers, novelists, poets, and historians who recognized some of the themes in this narrative. But empirical research added important new details to the humanist’s rough outline. We now have a richer appreciation of the biologically prepared biases of infants; recognize the developmental stages that are marked by the acquisition of a language, a moral sense, consciousness, and inferential talents; recognize that blends of temperaments and experiences create variation in personality and mental illness symptoms; and, hopefully, are ready to acknowledge that the display of a psychological property is often limited to specific settings.

Despite many victories, a number of important problems that are amenable to inquiry or reconceptualization are being ignored.

First, scientists studying psychological phenomena should replace their habit of linking one cause to one outcome with an examination of the relations between patterns of causal conditions and patterns of outcomes. A single condition (whether a gene, a secure attachment, premature birth, abuse, harsh socialization, or bullying) that ignores the child’s gender, temperament, ethnicity, social class, and culture usually explains little of the variation in most psychological outcomes.

Prediction of the probability of suicide by an American adolescent is aided in a major way by knowing that the youth is a member of a disadvantaged social class, resides in a rural area of a western state, and made the suicidal attempt on a Monday during the spring or summer months. Psychologists who assume that being a victim of bullying increases the likelihood of aggressive behavior, depression, or social anxiety in the absence of other conditions are ignoring the contributions of the child’s social class, ethnicity, gender, body size, school performance, and personality traits to the probability of being selected as a target for bullying, because the victims are not a random sample of all youths.

A team of Danish scientists found that the adolescents most likely to be bullied in each of thirty-five different countries came from poorer families but were attending schools that had many students from affluent families. High school youths from suburban Boston communities who reported being bullied were more likely to be deviant in some way; for example, they were gay, were small for their age, came from a poor family, or performed poorly on examinations. Hence, discovery of a relation between being bullied and a later trait is usually confounded with other properties that make a necessary contribution to the outcome. In the case of Mao Zedong, who was bullied by the wealthy boys at the school he attended because of his shabby clothes and peasant dialect, the experience of being a victim of derisive taunts increased his desire to become a person of importance who commanded respect. Adolescent boys from different homes and possessing different personalities might react to the same cruelty by deciding that they would never achieve the goals they wished to command. Investigators should also measure patterns of outcomes. Too often scientists measure a single variable—for example, the concentration of a molecule (cortisol or a sex hormone), answers to a questionnaire, a behavior (time looking at a stimulus or time to respond to a target), an autonomic response (blood pressure or heart rate), or a brain reaction. The problem with this strategy, however, is that each of these single measures is affected by more than one process and thus has an ambiguous meaning.

Adults who had initially seen many pictures of familiar objects, such as guitars and combs, were later shown three kinds of pictures: the same object in its original position, the same object at a different angle, and a totally new object. The fusiform area in the right hemisphere was activated by the familiar object presented at a different angle; the fusiform site in the left hemisphere was not. An investigator who examined only the evidence from the left hemisphere would arrive at the wrong conclusion regarding the brain’s response to the change in perspective. Examining the evidence from both hemispheres protects the scientist from the error of claiming that the brain did or did not detect the change in the angle of presentation.

An event as simple as a continuous loud noise, usually experienced as annoying or stressful, activates a circuit of at least nine interconnected brain sites. Scientists cannot know the person’s psychological state by measuring activity in only one location in this circuit. The act of reaching for a cup involves the simultaneous activation of a pattern of arm muscles by the spinal cord and cannot be explained by the activity of individual muscles.

Earlier I described a student who wanted to discover the features of sound that were most likely to alert newborn infants. Richard Kearsley recorded the behaviors of two-day-old infants in reaction to sounds that varied in frequency, loudness, and rise time. After spending several years gathering evidence and then analyzing the infants’ level of alertness to each stimulus, he was disappointed to learn that no particular frequency, loudness, or rise time had a significant influence on the infants’ behavior. When he came to me for advice, I suggested that he look for the patterns of frequency, loudness, and rise time that produced the most obvious alerting in a majority of the newborns. When he did so, he discovered one pattern that combined a specific frequency, loudness, and rise time that was maximally alerting. It is not a coincidence that that this sound bore the closest resemblance to the qualities of the human voice.


Investigators wishing to discover what infants know about an object or event should record looking time along with event-related potentials, eye movements, heart rate, or facial expressions. No single measure can reveal all that an infant knows about an event. Psychologists who study human memory have learned that the recall of a poem learned years earlier is not the most accurate index of how much information was preserved. That is why they also measure recognition of words from the poem and the ease of relearning the poem. Because scientists at the University of Lausanne measured both duration of body immobility and heart rate, they did not make the mistake of concluding that oxytocin reduced the intensity of a rat’s fear. Although high levels of oxytocin resulted in a shorter duration of body immobility, implying less fear, this molecule had no effect on the animal’s heart rate, implying no change in the level of fear.

Verbal reports, behaviors, and biological measures reveal different aspects of the invisible phenomena that psychologists wish to understand. Verbal descriptions imply that psychological traits, such as extraverted or agreeable, are displayed across different settings. Observations of behavior in natural contexts complement the verbal replies by revealing that traits such as extraversion or agreeableness are often displayed in select settings. The observations fine-tune the semantic concepts and allow the investigator to appreciate which contexts release the trait in question and which do not. I suspect that no more than 50 percent of a group of adults who describe themselves as extraverts on questionnaires would show the behavior that is expected of extraverts in a variety of settings. The other half belong to different categories.

Sadly, only a small proportion of psychologists study the effect of the context on the behaviors that define a trait, despite their acknowledgment that one of psychology’s major goals is to predict behavior. Instead, the evidence gathered in the vast majority of investigations of human participants comes from verbal reports, a brain measure, or the speed of a simple motor response in a laboratory setting.

Neither questionnaire replies nor BOLD signals are adequate proxies for the behaviors that are displayed in the settings in which individuals live. Adding measures of behavior in settings that approximate natural contexts to the verbal reports and brain profiles is mandatory if psychologists want to keep the promises they have made to the public and to the agencies that support them.

The assumption that a single measure, whose power depends on it being part of a pattern, has some of the power possessed by the complete pattern is analogous to conferring aesthetic qualities to a single thread from a Gobelin tapestry. Georg von Bekesy, who was awarded a Nobel Prize in 1961 for discovering how sounds affect the sensory cells of the inner ear, told a younger colleague that significant discoveries require studying a phenomenon under as many different conditions and with as many measures as possible in order to avoid a premature and incorrect conclusion. Although psychologists are familiar with the many discoveries in biology that reveal more complexity than had been assumed, they continue to favor simple cause-effect relations in which one condition—an insecure attachment, bullying, childhood neglect, or a gene—can make a significant contribution to an outcome that is measured with one procedure. This flight from complexity, which comes naturally to the human mind, is not a recipe for progress.


A second phenomenon requiring a more satisfactory explanation is the fact that the social class of rearing, which is correlated with the adult’s class, is a powerful predictor of many important outcomes in childhood and the adult years, whether children grow up in a developed nation like the United States or a poor country like Madagascar. The political attitudes, ethical positions, styles of social interaction, feelings of agency, and health of individuals born to college-educated parents differ from those of individuals born to parents who did not graduate high school. Yet scientists do not understand the processes that were responsible for these distinctive collections of properties.

What are parents from different classes doing in the home? How do children react to their parents’ practices and personal traits? When does an identification with one’s class emerge, and how is it manifested? What conditions differentiate the school environments of the poor and the affluent? Because we cannot answer these and other questions, we are deprived of a persuasive explanation of the robust relations between class of rearing and a host of outcomes that affect governmental expenditures for health care, prisons, and special education, as well as the person’s quality of life. The failure to understand the reasons why class is so important makes it difficult for experts to suggest strategies that might improve the lives of millions of individuals across the world’s societies.

Despite the obvious significance of this problem, I suspect that no social scientist is currently writing a grant request that proposes to gather behavioral and biological evidence—not just questionnaire data—on the temperaments, family experiences, peer relationships, school experiences, and identifications of one thousand children from different class and ethnic groups who will be studied from the early postnatal months through the adolescent years. One reason is that such a project would probably cost several billion dollars.

Social scientists are reluctant to make requests for that much money because they do not expect them to be approved. This timidity exacts a cost. The small-scale, relatively inexpensive studies that are most often published in technical journals are unlikely to yield discoveries that have a half-life greater than ten years. The reliance on this strategy over the past few decades persuaded government agencies to award a smaller proportion of funds to social science research than they did sixty years ago. Social scientists might take note of the fact that physicists regularly request and receive large amounts of money to study a feature of the material universe. The United States spent more than $2 billion for the project that allowed the landing of a vehicle on Mars in August 2012.

The high cost of studying the psychological consequences of class membership is not the only reason why social scientists are reluctant to explore the actual experiences that accompany growing up in families belonging to divergent classes. Another reason stems from an unwillingness among many Americans to acknowledge the profound class differences in a society that values egalitarianism and equal opportunity. Perhaps investigators are a little apprehensive over discovering facts that would require faulting some parents for their practices and criticizing some teachers for indifference, especially if both groups belong to an ethnic minority. The dissemination of these findings might also raise the average citizen’s consciousness—or, in some cases, guilt—over the serious health burdens that accompany a life of poverty and provoke a more vocal demand for allocating additional resources to the poor. This is politically dangerous territory.

A similar reluctance surrounded research on sexual behavior before the end of World War II. Frank Beach, who was my mentor when I was a graduate student at Yale in the 1950s, had been studying animal sexual behavior at the American Museum of Natural History in New York when he learned in the late 1940s that he was being considered for a distinguished professorial chair at Yale. Several members of the committee of Yale professors reviewing his credentials resisted his appointment because they felt it was unseemly for a Yale professor to study sex, even among animals. Fortunately, they were outvoted. Diluted, but similar, attitudes toward research on sexual behavior persist today. Neuroscientists interested in the brain sites that contribute to basic psychological systems usually condition a fear state in rats or mice, and less often condition a state of sexual arousal, despite their ability to do so and the evolutionary significance of sexual behavior.

A long-standing bias favoring the material causes of mental phenomena helps to explain an indifference to inventing novel ways to measure the psychological states created by rearing in an advantaged or disadvantaged class. Genes, brains, neurons, and molecules are material things with features that can be measured accurately and manipulated. The emotions that accompany class membership are invisible, non-material entities that emerge and disappear at unpredictable times and are more difficult to measure and manipulate.

The failure to develop more powerful methods to study psychological states prevents a deeper insight into the processes that mediate the relation between class of rearing and the child’s command of skills and traits. Most research on this issue relies on the crude evidence that questionnaires or interviews provide. Social scientists have insufficient enthusiasm for spending many years developing a method that might uncover new phenomena or measure a theoretically important process. Physicists, chemists, and biologists regularly support and celebrate colleagues who are willing to devote many years to perfecting a new machine or procedure. Anthony Greenwald of the University of Washington notes that 80 percent of the Nobel Prizes in physics, chemistry, physiology, and medicine from 1991 to 2011 were given to scientists who invented a new method or improved on an older one. The two scientists responsible for the development of the magnetic scanner, Paul Lauterbur and Sir Peter Mansfield, are examples.

I cannot think of many psychologists who worked for a decade or more trying to perfect a new, more accurate procedure to evaluate a person’s identification with his or her social class, beliefs about the self, or conception of society. As a result, questionnaires remain, as they have for a hundred years, the primary source of information on these concepts. The technologies that psychologists often use—cameras, DVD players, and computers—were invented by engineers and natural scientists for other purposes.

Natural scientists are regularly attracted to the challenge of inventing a new technology because of the promise of discovering a novel phenomenon with an elegant new machine. As I wrote this sentence I was reminded of an afternoon almost fifty years ago when I was sitting with fourteen other social scientists at a long, polished table in an elegant room of a mansion given to the Massachusetts Institute of Technology for conferences. New machines that recorded sounds with great fidelity on reels of tape had just become available. The speaker, wearing a dark suit, vest, and a carefully trimmed van Dyke beard, wheeled in a large, expensive recorder, cleared his throat, and told his audience: “We will now hear a baby cry.” He flicked on the recorder and that is what we heard. He stopped the recorder after about thirty seconds and said: “Now we will hear a second baby cry.” When these cries terminated he announced that his presentation was complete, and he sat down. I was surprised that no one was laughing.


A third issue deserving closer scrutiny is the psychologist’s continued attraction to abstract, semantic concepts that ignore the properties of the agent, the features of the setting, and the source of evidence. Natural scientists typically select a puzzling phenomenon as the primary reason for initiating research. Most eminent physicists during the 1920s agreed that accounting for the spectral lines emitted by an atom was a pressing problem deserving inquiry. The subsequent answers led to the body of equations called quantum mechanics. Crick and Watson chose the molecular structure of DNA as the puzzle to solve. They did not begin by assuming it was a double helix. Rather, they inferred that structure from Crick’s metal models of the molecule and from Rosalind Franklin’s photographs. Linus Pauling’s incorrect assumption that DNA was a triple helix made it easier for Crick and Watson to arrive at the correct solution first.

Many psychologists, by contrast, prefer to prove the validity of intuitively attractive, abstract ideas, perhaps because philosophy is one of the discipline’s parents. Philosophers typically analyze single words with a controversial meaning and generate a semantic argument that defends the utility, often temporary, of one particular meaning. Theory of mind, self-concept, well-being, impulsivity, stress, risk aversion, fear, intelligence, social cognition, cooperation, altruism, positive affect, and regulation are among the popular concepts that psychologists try to measure. Each term is indifferent to the specific agent to whom the property applies, the setting, and the fact that often only a single source of evidence is used to infer the property.

Most social scientists favor one of two research strategies. One group prefers to use a robust observation that is not yet understood as the reason for further inquiry—for example, discovering the biological properties and experiences of children who have great difficulty learning to read. The second, larger group prefers to begin with a semantic concept whose validity they want to prove. The psychologists wishing to demonstrate that infants understand causality, and use looking time as the only measure, provide an example. The psychologists who are attracted to the study of puzzles would have noted that infants vary in the duration of time they look at various events. Because this variation is not understood, they would have studied the conditions that produce long, medium, and short looking times. If the evidence they gathered required the conclusion that infants understand causality, they would have entertained that possibility—but only after the fact, not before they understood the factors modulating the amount of time infants of varied ages look at varied classes of events.

One problem trailing the strategy of beginning with intuitions about fruitful semantic concepts rather than puzzling phenomena is that the different schemata linked to a concept confer divergent meanings to the same term. For example, the term anxiety could evoke schemata of task failure, a parent’s harsh criticisms, losing money, a social blunder, the disruption of a satisfying relationship, or the health of a parent. Each schema awards a different meaning to the word. Similar ambiguities apply to the concepts stress, intelligence, and regulation.

Many commentators have noted that humans are susceptible to believing that if a word is used frequently it probably names something that exists. The psychologists’ frequent use of the word regulate invites the inference that this term must name a property that some humans display consistently across diverse settings. However, Eli Tsukayama and his colleagues at the University of Pennsylvania found that few adults regulate inappropriate, impulsive decisions across the domains of work, social relationships, sex, gambling, and investing money. Most who admit to making impulsive decisions in one domain—say, eating too much at restaurants—are not impulsive when spending money or working. Tiger Woods effectively regulated impulsive decisions in most areas of his life but failed to activate that property when he was tempted by sexual desire.

It is misleading to argue, as a few psychologists do, that youths from poor, single-parent families living in urban neighborhoods with drug dealers, pimps, and rival gangs do not regulate their aggression as well as children from comfortable, affectionate, two-parent families living in safer small towns. The latter youths encounter far fewer occasions where they have to regulate inappropriate actions. A marble in a groove is not effectively regulating its perfectly straight motion.

A popular illustration in journal articles on the effects of stress on children features a box enclosing the term adversity and an arrow pointing to a second box enclosing the phrase problems with self-regulation. Two additional boxes, sitting on top of the first pair, enclose the terms family environment and genetic background. Arrows from each of the four boxes point to a fifth box enclosing a list of undesirable outcomes that accompany poor regulation. The intended message in this array of boxes is that any adversity can, under some family and genetic conditions, compromise a child’s ability to control impulses and lead to a variety of maladaptive responses. The form of the adversity, the child’s interpretation, the specific family practices, the genes, and the settings in which regulation is required are free to vary, allowing readers to insert in the boxes any adverse event, parental practice, gene, form of regulation, and outcome they wish. This permissiveness makes it close to impossible to disprove the claim that some adversities compromise some aspect of regulation in some children born with certain genomes and growing up in certain families.

The concept victim of bullying is another example. Herbert Marsh and his colleagues at the University of Western Sydney found that some Australian boys who physically bullied others became victims of bullying months later. Psychologists do not have a category for youths who are both bullies and victims at different times during the same year. Some men who are bullied by their employer in the workplace bully their wives at home. The wives in turn bully their school-age sons, who bully timid boys at recess, who in turn bully their younger sisters. This sequence makes it difficult to decide who is a bully and who is a victim. The Chinese note that the moment the sun reaches the highest point in the sky, it is prepared to begin its descent.

I confess to being initially surprised and then saddened by reading a review essay, by two geneticists from North Carolina State University, titled “The Genetics of Aggression.” The authors never defined aggression and simply assumed that readers understood what it meant. More seriously, these biologists declared, without adequate evidence, that one set of genes contributed to all forms of aggression— whether ants defending a nest, sharks attacking a human, snakes killing a mouse, mice biting an intruder, alpha baboons defending their status, or jealous spouses stabbing their mate for infidelity.

The related, equally abstract term violence, intended to apply to humans, fails to specify the class of person committing the violence, their motives, the form the violence assumed, or the contexts in which a violent action occurred. These four features assumed different combinations when one of Napoleon’s troops killed a Russian soldier who was about to shoot him, a Nazi official ordered the gassing of innocent Jewish women, a new Ottoman sultan murdered his brothers and all the pregnant women in the harem of the recently deceased sultan, an estranged adolescent killed twelve classmates in a rampage, a bank robber murdered a teller, a white racist lynched an African-American, a Hutu man slaughtered a pregnant Tutsi woman, a Syrian soldier shelled a crowd protesting the regime’s legitimacy, one gentleman shot another in a duel because the latter questioned his honor, and an unmarried adolescent girl drowned her newborn infant.

Steven Pinker’s decision in "The Better Angels of Our Nature" to pool different types of violent actions by agents who varied in gender, age, ethnicity, and motive into a single category called violence is analogous to grouping the olfactory signals of ants, dances of bees, songs of birds, croaks of frogs, grimaces of monkeys, screams of infants, lullabies by mothers, lectures by professors, and spam messages over the Internet into an omnibus category called communication.

The probability of a violent act is palpably higher for males compared with females across the life span. Males between seventeen and thirty years of age commit the vast majority of violent behaviors, in the past as well as today. Older men and children are unlikely to kill, rape, or torture anyone. Males between age seventeen and thirty represented a large proportion of the European population between 1200 and 1700 because more than 50 percent of children died before age five and more than 80 percent of adults died before age fifty. By contrast, males between seventeen and thirty make up less than 20 percent of today’s European population. Hence, a scientist who bases the prevalence of violence on the ratio of the number of homicides over the total population would be likely to discover that the rate of violence decreased across the interval from 1200 to 2000 because the proportion of the total population that was male and between age seventeen and thirty had declined precipitously over those eight hundred years.

This statement is not intended to question the fact, noted by many scholars, that there has been a steady decline in deaths due to violence over the past eight centuries—a fact that seems secure rather than an artifact. However, I suspect that the decline might be less steep if the index of violence was the ratio of the number of violent acts, or deaths due to violence, divided by the total number of males between seventeen and thirty years of age, rather than divided by the total population. If, for example, the absolute number of violent acts committed by males between seventeen and thirty was the same in the fourteenth and twentieth centuries, the ratios based on the total population would indicate a decline in violence. Conclusions based on ratios are always tricky, however, because they can vary with the denominator that is selected. An estimate of the population density of the United States based on the ratio of the total population over the total land area of the country implies a low density. But a ratio whose denominator is the land area where two-thirds of Americans live implies a high density.

A personal experience taught me the advantages of resisting abstract concepts that have insufficient support from evidence. I was fond of the popular ideas of motive and conflict in 1957 when, as a twenty-eight-year-old psychologist at the Fels Research Institute in Ohio, I needed a battery of procedures to measure personality traits in young adults who had been studied since their early childhood. Many psychologists I respected believed that adults would reveal their motives and conflicts through their interpretations of ambiguous stimuli. A task that was popular at the time involved giving a person arrays of twenty-two small cardboard figures in postures that suggested anger, anxiety, sexuality, dependence, frailty, authority, or neutral traits. Each individual was asked to select groups of figures that he or she believed “went together.” The assumption behind this instruction was that a person who grouped the figures suggestive of anger was preoccupied with the motive of hostility, whereas a person who selected figures suggestive of sexual motives was concerned with or conflicted over this desire.

After poring over the groupings the adults produced, I was disappointed to learn that the data were totally unrelated to the extensive, objective information available on their child and adult personalities. Frustrated, I tried to figure out what biases, if any, guided the groupings. One evening, after many weeks of study, a possible answer emerged. Some individuals based their groups on the physical features of the figures—say, all the figures that were standing up, lying down, or with their arms raised. Others grouped figures that formed a narrative; for example, the figure of a woman with her arm raised was grouped with the figure of a child on his knees. Fortunately, I recorded the time each person took to create each group. Those who based their groupings on physical features took longer than the others. Equally important, these adults were more cautious and compulsive as children and as adults than those who grouped figures in narrative relations. It appeared that the groupings did reveal a personality trait, but not the one I had predicted. This new interpretation of the evidence led to the invention of the concept of a reflective cognitive style, which I pursued for many years. This example illustrates the problem trailing the premature acceptance of an intuitively attractive word and assuming the validity of a procedure to measure the concept underlying it.

Young psychologists are expected to test the validity of a theoretical idea and are often criticized by their elders for gathering observations that are not guided by a well-argued prediction that enjoys legitimacy among investigators working in the same domain. This practice is odd because a large number of the most important discoveries in psychology were unanticipated by any hypothesis. Ivan Pavlov had no strong expectations about what he would find when he began his experiments on the conditioning of salivation in dogs. The scientists studying the man called H.M., who lost his hippocampus in an operation intended to relieve his epilepsy, were surprised to learn that H.M. had a satisfactory memory for many facts of his childhood but could not remember a sentence heard twenty seconds earlier.

Torsten Wiesel, who shared the 1981 Nobel Prize with David Hubel for discovering the receptivity of neurons in the visual cortex to particular contours, confessed to a reporter: “We started out with no hypothesis but just forged ahead to the best of our intuition.” The discovery that led to the 2011 Nobel Prize in chemistry was equally unexpected. Daniel Shechtman’s decision to cool a molten glob of aluminum and manganese quickly led to a totally new class of crystal that no scientist believed existed. The three biologists who won the 1995 Nobel Prize in Physiology had absolutely no idea what might happen when they fed male fruit flies chemicals that they hoped would cause mutations in the flies’ sperm. They were surprised to find that the chemicals caused changes in a small number of genes that controlled the shape of the body axis. This fact has become one of the momentous discoveries of the past few decades. We need more playful curiosity in psychology laboratories and among the directors of the agencies awarding funds for the study of psychological phenomena.

Psychologists who prefer to design experiments that affirm the validity of big words, which the writer Helen Sword calls “zombie nouns,” should descend from the high tower—from which objects on the ground below look the same—to the street level where their distinctive features are obvious. The problem with concepts like introversion, attend, remember, fearful, bold, positive affect, stress, and regulate is that they fail to tell readers who, when, where, and what. Alecia Carter and her colleagues at Australian National University remind psychologists that the baboons who were exceedingly bold when presented with a stuffed version of a predator snake were not at all bold when the object was a novel food. And a team of scientists led by Frank Albert found that the trait called tameness, when applied to the domesticated species of dogs, pigs, and rabbits, compared with their less tame forms in the wild, can be the product of different genes and brain physiologies in each domesticated species. These examples illustrate why scientists must specify the context and the species when using adjectives like bold or tame to describe an animal or a human. Despite the reasonableness of this suggestion, one team used the words cooperators and cheaters to describe the activity of different kinds of yeast cells. A similar indifference to species and source of evidence is common among scientists who are fond of the concept stress. They assume that the state of rats who are restrained in a narrow tube or forced to swim in a tub of water resembles the state of adults who have lost a job, money, their home, or a loved one.

The mathematician Peter Woit described an occasion when the physicist Wolfgang Pauli became irritated with a young Werner Heisenberg because the latter insisted that he had come up with a unified theory of matter that was missing only a few technical details. The next day Pauli sent his friends a piece of paper containing a blank rectangle in the shape of a frame for a painting. Written below the frame was the sentence: “This is to show that I can paint like Titian. Only the technical details are missing.”

There is a wooden bench close to the edge of a small pond in a nature preserve about a mile from my home where I walk regularly. A plaque on the back of the bench contains a quote by the late Paul Samuelson, a Nobel Laureate in economics: “A professor must have a theory as a dog must have fleas.” That pithy statement has a complementary truth: “A professor must question her theory as her students and colleagues must question her.”

A number of abstract psychological concepts remain popular because they satisfy the need for consistency among the investigator’s semantic networks. The networks for the concepts positive emotion and negative emotion are an example. Joy is always classified as a positive emotion and guilt as a negative one. Hence, they should not occur together. Yet it is possible that some adolescents who experience joy upon learning that a disliked rival failed an examination might immediately experience guilt because feeling happy over another’s misfortune violates their ethical code. The young Rousseau felt the positive emotion of sexual arousal when he was spanked by his governess, although the spanking evoked the unpleasant sensation of pain, which should have created a negative emotion. The continued use of the terms positive and negative to describe emotions or experiences allows investigators to avoid specifying whether these words refer to the brain states or feelings evoked by objectively pleasant or unpleasant events (food, sex, and money or pain, loss, and frustration) or to the subjective appraisals of those feelings in a particular context (a masochist being whipped). Authors who let readers decide on the meaning of positive and negative are protected from being wrong. Biologists would never use these evaluative adjectives to describe the states of organs or cells. That is, they would not write that a heart was in a negative state because colleagues want to know the objective state of the heart.

The problems trailing attempts to preserve semantic consistency are clearest for concepts related to the antonyms good and bad. Many popular terms for human qualities belong to semantic networks that have good and bad as nodes. This contrast includes a secure versus an insecure attachment, effective versus ineffective regulation, and a vulnerable versus a resilient personality. Investigators hold the implicit premise that all bad properties, as well as all good properties, should be positively correlated.

The contemporary Western mind prefers a symmetry in which good experiences are followed by good outcomes and bad experiences by bad ones. This premise is falsified by the fact that many stressful events, classified as bad, motivate some children to acquire the coping defenses that lead to a good outcome later. By contrast, many good experiences early in life are followed by undesirable outcomes. One-year-old American girls with a secure attachment to the mother are apt to become anxious adolescents if their mothers socialize their daughters to be passive with boys, noncompetitive in school, and guilty over sexual feelings. Most Americans born to college-educated parents between 1987 and 1992 were the childhood recipients of the good family experiences of affection, gentle socialization, and consistent praise. But experts are describing this generation as narcissistic, which is a bad trait. Professional tennis players who avoid close attachments to other players when on tour, which some psychologists would regard as a bad trait, win more matches.

Psychologists should infer concepts from a diverse set of observations in different contexts and resist the temptation to posit correlations among concepts simply because they are semantically consistent. Natural phenomena are under no obligation to honor a semantic consistency among all good outcomes or all bad ones. The amygdala is activated by the unexpected sight of a tarantula as well as by a pile of hundred-dollar bills under a log. Maple leaves are most beautiful the week before they fall. Flamingos are least beautiful the moment they hatch.


About twelve years ago, while chatting with one of the world’s most respected neuroscientists, I asked whether he believed that if he had complete knowledge of my brain at that moment he would be able to predict whether I was about to crawl under the table at which we were sitting. To my surprise, he said “yes.” His reply captures the fourth puzzle that this final chapter considers—namely, explaining how any psychological phenomenon could emerge from a profile of brain activity. A great many psychologists have decided that this problem has an obvious priority over other candidates because of the new technologies permitting study of the brain and a long-standing preference for measuring material phenomena that can be observed rather than psychological processes that cannot be seen under a microscope or inferred from patterns on an oscilloscope.

Very few scientists question the popular assumption that the brain is the foundation of all psychological phenomena. We have gained an initial understanding of the bases for some aspects of visual, auditory, and olfactory perception as well as a few motor actions. However, the brain bases for beliefs, feelings, emotions, intentions, personality traits, complex actions, and consciousness, as well as the correlations between the development of the brain and the emergence of new cognitive talents, remain more mysterious.

Rarely does any brain measure predict a psychological outcome in 90 percent of research participants. This claim holds even for the perception of acute pain. Kay Brodersen and her colleagues at the University of Oxford found that about 60 percent of the time adult reports of pain to a stimulus were correlated with a pattern of brain activation. Reports of a painful sensation in the remaining 40 percent of the trials were unrelated to a particular brain profile.

Similarly, a team of neuroscientists studying monkeys at the Max Planck Institute in Tubingen discovered that even when the outcome is an extremely well-defined, simple motor response indicating whether the animal recognized an object as familiar or unfamiliar, an appropriate brain measure could predict the monkey’s behavior only about one-third of the time.

Despite the efforts of many brilliant investigators who have addressed this problem with new technologies over the past forty years (admittedly a short time), the veil separating patterns of brain activity from most psychological phenomena remains opaque. In The Organisation of Mind, Tim Shallice and Richard Cooper of the University of London wrote an extensive and fair summary of what is known about the relation of brain measures to perception, memory, consciousness, and thinking. Although the book reflects the authors’ belief that future scientists will discover more robust relations, Shallice and Cooper had to admit that every major conclusion wrested from nature as of 2011 was vulnerable to the criticism of one or more skeptics who could mount a reasonable, but different, explanation of the same evidence.

One reason for the persistent disagreements, which I noted earlier, is the habit of writing about abstract psychological processes without specifying agents, targets, and sources of evidence. The final chapter on thinking in Shallice and Cooper’s book contains an illustration with boxes labeled goal generation, working memory, monitoring system, and running of cognitive procedures. These boxes fail to specify the age of the subjects, the goals being sought, the representations being remembered or monitored, or the cognitive procedures being run.

The brain circuit activated when adults attend to a salient figure in the foreground of a scene is not the circuit activated when they attend to the spatial relationships between the figure and the scene’s cognitive processes have to specify what these processes are working on. There is no single brain circuit corresponding to the abstract term remember; there probably is a circuit, however, that is the foundation of remembering a sibling’s birthday.

Excerpted with permission from "The Human Spark: The Science of Human Development," by Jerome Kagan.  Available from Basic Books, a member of The Perseus Books Group. Copyright © 2013.

By Jerome Kagan

MORE FROM Jerome Kagan

Related Topics ------------------------------------------

Books Editor's Picks Neuroscience