Thinking, Language & Intelligence

Every waking moment, your mind is busy. It sorts the world into groups. It hunts for answers. It wraps ideas in words. It sizes up how smart the person across from you seems. This chapter opens up the machinery of higher thought. You will see how we build the concepts that let us think at all. You will see the shortcuts and traps that shape our judgments. You will meet language, an amazing system that a toddler learns without a single lesson. And you will follow a hundred-year effort to define and measure intelligence. Along the way you will meet Rosch, Tversky and Kahneman, Chomsky, Spearman, Gardner, Binet, and Wechsler. They turned these everyday wonders into science you can test. The goal is not to memorize a long list. It is to see how thinking works and where it breaks. It is also to see why measuring a mind is much harder than it looks.

Section 1Concepts & Problem Solving

Concepts: the mind's filing system

Imagine if you had to treat every object, event, and idea as something brand new. Thinking would be impossible. Instead, the mind sorts things into concepts. A concept is a mental group of objects, events, or ideas that share features. The concept dog lets you see that a Chihuahua and a Great Dane are the same kind of thing. You can guess that both will bark and need food. You do all this without trying. Concepts come in two rough kinds. Natural concepts come from everyday life. Their edges are fuzzy. Think of fruit or game. Artificial (formal) concepts are set by a fixed list of features. A triangle is any closed shape with three straight sides. There are no exceptions.

We rarely store a concept as a list of rules. Eleanor Rosch offered prototype theory. In it, each group has a prototype at its center. A prototype is the best, most typical example. For most North Americans, a robin is a very typical bird. A penguin or an ostrich sits out at the edge of the group. The closer an item is to the prototype, the faster we see that it belongs. Concepts also link together into schemas. A schema is a bigger framework that organizes information and what we expect. A role schema tells you how a librarian is likely to act. An event schema, also called a script, tells you the usual order of events at a restaurant: get seated, order, eat, pay. Schemas make life easier. But they also plant the seeds of stereotypes and false expectations. Later sections show how those can mislead us.

Solving problems: algorithms and heuristics

A problem exists any time there is a gap between where you are and where you want to be. Psychologists sort the ways across that gap into two broad plans. An algorithm is a step-by-step method. If you follow it correctly, it is guaranteed to give the right answer. Examples are a long-division rule, a recipe, or trying every possible code on a lock. Algorithms are reliable but often slow. For big problems they can take far too long. A heuristic is a mental shortcut, or rule of thumb. It usually gets you there faster, but it comes with no promise. "Start with the corners and edges" is a heuristic for a jigsaw puzzle. Two other heuristics solve many everyday problems. One is working backward from the goal. The other is means-ends analysis, where you keep taking the step that most shrinks the distance to the goal. The trade is clear. Algorithms buy certainty with effort. Heuristics buy speed with a risk of error.

Good solutions also get blocked by roadblocks we can predict. A mental set is the habit of using a method that worked in the past, even when a simpler way exists. In Abraham Luchins's 1942 water jar studies, people who learned a long formula kept using it on later problems that had a two-step answer. Functional fixedness is a special kind of mental set. It means seeing an object only in its usual role. You may miss that a coin can work as a screwdriver, or that a book can prop open a window. Spotting these traps is the first step to escaping them. That is just what the next section is about.

Section 2Judgment & Bias

Two heuristics that bend our judgment

Starting in the 1970s, Amos Tversky and Daniel Kahneman showed something important. The shortcuts from the last section don't cause random mistakes. They cause predictable ones. The availability heuristic judges how likely or common something is by how easily examples come to mind. Events that are vivid, recent, or all over the news feel more likely than they are. People fear plane crashes more than car crashes. They fear shark attacks more than the backyard swimming pool, which is far more deadly. The dramatic cases are so easy to bring to mind, or available. Yet the everyday risks kill far more people. The representativeness heuristic judges how likely something is by how well it matches a prototype. Suppose you hear, "Steve is shy, tidy, and pays close attention to detail." Most people guess he is more likely a librarian than a farmer. They ignore the base rate: there are far more farmers than librarians. The same heuristic causes the famous conjunction fallacy. People rate "Linda is a bank teller and a feminist" as more likely than "Linda is a bank teller." That can never be true. Two things together can't be more likely than one of them alone. People choose it because the fuller description fits their prototype better.

Confirmation bias and framing

Confirmation bias is the habit of looking for, favoring, and remembering information that supports what we already believe. At the same time, we brush off evidence that goes against it. Peter Wason ran classic studies with cards and number rules. People tested their hunches almost only by looking for cases that fit. They rarely tried the test that could have shown their idea was wrong. This bias quietly drives many things, from stubborn political arguments to a doctor sticking with a first diagnosis. Last comes the framing effect. It shows that the wording of a choice, not just the facts, moves decisions. Tversky and Kahneman used the "Asian disease" problem, about an imagined disease outbreak. People chose between programs described as lives saved or lives lost. When the choice was framed as a gain, people played it safe. When the same odds were framed as a loss, people took the risk. Meat labeled "90% lean" sells better than the same meat labeled "10% fat." The facts stay the same. Only the frame changes, and so does the answer.

Section 3Language

The building blocks of language

Language is a system of symbols joined by rules. It lets us share an endless range of meanings. The symbols are arbitrary. The word "dog" does not look or sound like a dog. Language gets its power from being built in layers. The smallest units of sound are phonemes. English uses about forty-four of them. A phoneme is the difference between the b and the p that separates bat from pat. Phonemes combine into morphemes. A morpheme is the smallest unit that carries meaning. The word cats holds two morphemes: cat (an animal) and -s (more than one). Morphemes build words. The study of what words mean is called semantics. Words are arranged by the rules of grammar. Grammar includes syntax, the rules for word order. This layered design gives language its most remarkable feature, called generativity. From a limited set of sounds and rules, we can make and understand sentences no one has ever said before, including this one.

How children learn language

Children learn this system amazingly fast. Through the toddler years they pick up roughly a new word every two waking hours. They also move through the stages in an order that is strikingly similar around the world (see the table). One telling clue is overregularization. A child who once said "went" starts saying "goed." This shows she has figured out the past-tense rule, add "-ed." She is not just copying what she heard. That fact sits at the heart of a famous debate. B. F. Skinner was a behaviorist. He argued that language is learned like any other behavior: through copying, reinforcement, and shaping. Parents reward correct speech and fix wrong speech. Noam Chomsky pushed back. He said reinforcement can't explain how fast and how creatively children make sentences no one taught them. It also can't explain errors like "goed," which no adult says. Chomsky proposed that humans are born with a language acquisition device. This is an inborn readiness for a universal grammar that underlies all languages. It just needs exposure to language to switch on. Today's view blends both sides. Biology supplies the equipment and a sensitive critical period for learning. A rich language setting supplies the input. A related question is whether the language you speak shapes the thoughts you can think. This is the linguistic relativity (Sapir-Whorf) hypothesis. Today it is seen as real but modest. Language nudges what we notice and remember. It does not lock thought in a cage.

Section 4Theories of Intelligence

One intelligence, or many?

Broadly, intelligence is the ability to learn from experience, solve problems, and adapt to new situations. But psychologists have argued for a century about whether it is one thing or many. Charles Spearman used a new math tool called factor analysis. It finds the hidden factors behind a set of scores. He noticed that people who score well on one kind of mental test tend to score well on others. He concluded there must be one underlying general intelligence, which he called g. In his view, g sits beneath all the specific abilities. Later thinkers refined this single factor. Raymond Cattell split it in two. Fluid intelligence is the raw ability to reason and solve new problems. It peaks in early adulthood. Crystallized intelligence is the knowledge and skills you build up. It grows with age. L. L. Thurstone rejected the single number entirely. He argued that intelligence is a cluster of seven "primary mental abilities," not one master factor.

Widening the definition

The strongest challenge to a single g came from Howard Gardner. His theory of multiple intelligences proposes at least eight fairly separate kinds. They are linguistic, logical-mathematical, musical, bodily-kinesthetic, spatial, interpersonal, intrapersonal, and naturalist. Gardner points to two kinds of evidence. One is savants, people with one striking ability. The other is brain injuries that knock out one ability and spare the rest. Many researchers are not convinced. Scores on the eight abilities still tend to rise and fall together. There is little direct test support for the theory. Some say the eight are better called talents. Also, the eight are not "learning styles." Many people believe you learn best when a lesson matches your style, such as "visual" or "auditory." Careful studies have not backed that up. It is a myth.

Robert Sternberg's triarchic theory offers a neater set of three. Analytical intelligence is the school-style problem solving that tests measure well. Creative intelligence is inventing and imagining. Practical intelligence is the "street smarts" of getting things done. A related idea spread far beyond the lab: emotional intelligence. Peter Salovey and John Mayer first described it, and Daniel Goleman made it popular. It is the ability to notice, understand, use, and manage emotions in yourself and in others. It predicts some parts of success at work and in love and friendship that an IQ score misses. But critics warn that it is hard to measure fairly. It can also turn into a label for every good trait.

Section 5Measuring Intelligence

From mental age to IQ

The modern intelligence test began with a practical request. In 1905, the French government asked Alfred Binet and Théodore Simon to find schoolchildren who needed extra help. Their scale introduced the idea of mental age. Mental age is the age level of the problems a child can solve. Lewis Terman of Stanford adapted the test into the Stanford-Binet. He also borrowed a formula for the intelligence quotient (IQ): mental age divided by real age, times 100. An eight-year-old who performed like a ten-year-old scored an IQ of 125. Modern tests dropped that formula. The main ones are David Wechsler's scales: the WAIS for adults and the WISC for children. They use a deviation IQ instead. It compares your score with people your own age on a normal distribution. This is the bell curve. Its average is 100, and about two-thirds of people score between 85 and 115. Wechsler's earlier tests also gave separate Verbal and Performance IQs. Today's versions dropped that split. They report a Full Scale IQ plus a set of index scores. The indexes are verbal comprehension, visual-spatial (perceptual) reasoning, fluid reasoning, working memory, and processing speed. This fits the many-part view of ability from the last section even better.

What makes a test trustworthy

A number is only as good as the test that made it. Two qualities decide whether an intelligence score means anything. Reliability means consistency. A reliable test gives nearly the same result if you take it twice. This is called test-retest reliability. It also gives matching results when you compare its two halves, called split-half reliability. Validity means accuracy. Does the test really measure what it claims to? A test has predictive (criterion) validity if scores predict a related outcome, such as school grades. It has construct validity if it truly taps the trait of intelligence and not something else, such as reading speed. Here is the key point. A test can be reliable without being valid. Picture a bathroom scale that always reads five pounds heavy. It is perfectly consistent and perfectly wrong. Both qualities depend on standardization. That means giving and scoring the test the same way every time. It also means comparing scores against the norms, or typical scores, of a large sample that represents the population.

Last comes the oldest question: where does intelligence come from? Twin and adoption studies give a clear answer that is easy to misread. Identical twins raised apart still have much closer IQs than adopted siblings raised together. So heritability estimates for intelligence are large. Heritability is about the differences between people in a group. It is the share of those differences that comes from differences in genes. The share changes with age. Estimates run near 0.4 to 0.5 in childhood. They rise to about 0.6 to 0.8 by adulthood. This pattern is called the Wilson effect. But heritability describes a group, not a person. It also says nothing about why groups differ. Environment matters a great deal. Enrichment, nutrition, and schooling raise scores. Average scores climbed steadily across the twentieth century. This rise is the Flynn effect. In some countries the rise has recently slowed or reversed, which also points to environmental causes. Claude Steele's work on stereotype threat found that reminding people of a negative stereotype about their group can lower their test scores. Later, larger studies suggest the effect is real but smaller and less consistent than first reported. Intelligence is neither fixed at birth nor endlessly changeable. It grows out of genes and environment working together all the time.

ReferenceWords worth knowing

A quick list of the terms in this chapter. Skim it now, or come back any time a word trips you up.

Algorithm

A step-by-step method that, if followed correctly, guarantees a solution to a problem.

Availability heuristic

Judging how likely or common something is by how easily examples come to mind.

Concept

A mental group of objects, events, or ideas that share features.

Confirmation bias

The habit of looking for, favoring, and remembering information that supports what you already believe.

Crystallized intelligence

Built-up knowledge and word skills, which tend to grow with age (Cattell).

Emotional intelligence

The ability to notice, understand, use, and manage emotions in yourself and others.

Fluid intelligence

The ability to reason and solve new problems, mostly apart from what you already know (Cattell).

Framing effect

The finding that how a choice is worded (as a gain or a loss) changes the decision.

Functional fixedness

Not being able to see an object as useful for anything except its usual job.

General intelligence (g)

Spearman's proposed single factor that underlies all specific mental abilities.

Heuristic

A mental shortcut or rule of thumb that speeds up problem solving but does not guarantee a correct answer.

Intelligence quotient (IQ)

A score that shows how a person performs compared with people the same age. It was first mental age divided by real age, times 100.

Language acquisition device

Chomsky's proposed inborn mental structure that readies humans to learn grammar from hearing language.

Mental set

The habit of tackling a new problem with a method that worked in the past.

Morpheme

The smallest unit of language that carries meaning, such as a root word or the plural "-s."

Phoneme

The smallest unit of sound in a language that can change a word's meaning.

Prototype

The best or most typical example of a concept. New items are compared against it.

Reliability

How consistent a test is: how much it gives the same result when measured again.

Representativeness heuristic

Judging how likely something is by how closely it matches a prototype, often ignoring base rates.

Standardization

Giving and scoring a test the same way every time, and comparing scores with norms from a sample that represents the population.

Telegraphic speech

The early two-word stage in which children get their meaning across while dropping small words ("want cookie").

Validity

How well a test measures what it claims to measure.