This one's a preview, not a review. The syllabus core takes you through memory in depth, but "cognition" as a field is bigger than encoding and retrieval — it's the whole machinery of thought: how you form categories, solve problems, judge probabilities, understand and produce language, and how psychologists have tried (and repeatedly struggled) to measure something called intelligence. You may not see all of this on an exam, but it's some of the most useful psychology you'll ever learn, because you use it — or misuse it — every single day, usually without noticing.
Thinking in categories, not definitions
Start with something you do thousands of times a day without thinking about it: categorizing. You see a four-legged animal and instantly know it's a dog, not a cat, not a fox. You'd assume your brain stores a tidy definition — "a dog is an animal with X, Y, and Z features" — and checks new cases against it. It doesn't work that way. Most everyday categories don't have clean definitions at all (try defining "game" so it includes chess, tag, and solitaire but excludes nothing that isn't a game). Instead, you think in prototypes: a mental "best example" of a category, built from the features that show up most often in the category members you've encountered. A robin is a more prototypical bird than a penguin — faster to classify, more likely to come to mind first — even though both are, technically, equally birds. Categorization by resemblance-to-prototype rather than by strict rule is fast and usually works fine. It's also the seed of a much bigger theme in this unit: your mind trades accuracy for speed constantly, and mostly you never notice the trade being made.
Problem-solving: guaranteed-but-slow vs. fast-but-fallible
The same trade-off shows up in problem-solving. When you face a problem, you have two broad strategies available. An algorithm is a step-by-step procedure that is guaranteed to produce a correct solution if you follow it completely — like trying every possible combination of a lock, or a formula you plug numbers into. Algorithms are reliable but often slow, sometimes absurdly so; trying every possible move in chess would take longer than the universe has existed. A heuristic, by contrast, is a mental shortcut or rule of thumb that gets you to a solution quickly and usually correctly, without guaranteeing it — for example, "when packing to move, start with the room you use least." Heuristics are how you solve most real-world problems, because most real-world problems are too big for exhaustive algorithms and don't need a perfect answer, just a good-enough one, fast.
Then there's a third experience entirely: insight. Sometimes a solution doesn't arrive through step-by-step progress at all — you're stuck, you set the problem aside, and then it suddenly appears, complete, often with a felt "aha" moment. Insight problems (classic examples involve seeing an object used in an unexpected way) feel qualitatively different from grinding through an algorithm, which is part of why they're of continuing interest to cognitive psychologists — the sense of sudden restructuring suggests your brain was doing something in the background that wasn't available to conscious step-tracking.
Step 1 · Try the algorithm, if there is one
If a guaranteed procedure exists and the problem is small enough, it's the safest route — no risk of a wrong shortcut.
Step 2 · Reach for a heuristic when the problem is too big
Most real problems (what to make for dinner, how to word an email) have no practical algorithm. A rule of thumb gets you a workable answer fast.
Step 3 · Watch for the insight moment
Sometimes neither works consciously, and stepping away lets restructuring happen outside awareness — then the answer arrives all at once.
Judging under uncertainty: the heuristics that quietly run your life
This is the centerpiece of the unit, and arguably one of the most consequential findings in all of psychology. Amos Tversky and Daniel Kahneman spent years documenting that when people judge probability and frequency — how likely is this, how common is that — they don't do formal statistical reasoning. They use heuristics, and those heuristics produce systematic, predictable errors (Tversky & Kahneman, 1974).
The availability heuristic says: you judge how likely or frequent something is by how easily examples come to mind. The problem is that "easy to recall" and "actually common" are not the same thing. Plane crashes are extensively covered by news media, vivid, and emotionally charged, so they come to mind easily — which is why many people rate flying as more dangerous than driving, when the reverse is true by a wide margin. Nothing about a car crash is less real; it's just less available in memory, because it's less dramatic and less reported.
The representativeness heuristic says: you judge how likely something is by how much it resembles your mental prototype of a category — and in doing so, you tend to ignore base rates, the actual statistical frequency of something in the population. Told that someone is quiet, detail-oriented, and enjoys organizing things, you might guess "librarian" over "farmer," even though farmers vastly outnumber librarians, making farmer the better bet on base rates alone before you even factor in the description. Resemblance to the stereotype overrides the math.
Two more biases round out the toolkit. Anchoring is the tendency for an initial number you're exposed to — even an arbitrary one — to pull your subsequent estimate toward it; a first offer in a negotiation, or the first price you see on a menu, quietly reshapes what feels reasonable afterward. Confirmation bias is the tendency to seek out, notice, and remember information that supports what you already believe, while overlooking or discounting information that doesn't — which is a large part of why people rarely talk each other out of strongly held positions with a single well-reasoned counterargument. Closely related are framing effects: the same information can produce different decisions depending on how it's worded — a treatment described as having a "90% survival rate" sounds more appealing than the identical treatment described as having a "10% mortality rate," even though they describe exactly the same outcome.
◆ Why this is the centerpiece: two systems, one of them lazy
Kahneman (2011) later organized this entire research program into a dual-process framework that's become one of the most widely cited ideas in psychology outside the classroom too. System 1 is fast, automatic, intuitive, and effortless — it's what generates your gut reaction, your snap categorization, your immediate sense of "that seems risky" or "that person seems trustworthy." System 2 is slow, deliberate, effortful, and rule-based — it's what you engage when you consciously work through a math problem or force yourself to double-check a decision. System 1 runs almost all the time because it's cheap; System 2 is expensive and lazy, and mostly rubber-stamps whatever System 1 hands it. Heuristics and biases are System 1 doing its normal job — they're not malfunctions, they're features that are usually good enough and occasionally get it badly wrong, especially in situations (probability, statistics, rare risks) that human cognition wasn't built to handle intuitively.
Bodhi says
The trap with this material is thinking "okay, now that I know about these biases, I won't fall for them." You will. Knowing a heuristic exists barely dents how much it runs your judgment in the moment — these aren't facts you memorize and disable, they're the default settings of a fast system that doesn't check in with the slow one unless something forces it to. The useful skill isn't "never use System 1." It's noticing the situations — big decisions, rare risks, anything involving statistics — where it's worth paying System 2's tax.
Language: how sound becomes meaning
Step back from judgment and look at the tool you use to communicate all of it: language. Language has a layered design. Phonemes are the smallest units of sound that change meaning (the difference between "bat" and "pat" is one phoneme). Phonemes combine into morphemes, the smallest units of meaning (like "un-," "break," and "-able" in "unbreakable"). Morphemes combine according to syntax — grammatical rules for how words can be ordered into sentences — and the resulting sentences carry semantics, actual meaning. Each level builds on the one below it, and by the time you're following a sentence in real time, you're running all four levels simultaneously without any conscious effort.
Children acquire this entire system with startling speed and without formal instruction, which raises an obvious question: how? Noam Chomsky argued that this speed and universality can't be explained by imitation and reinforcement alone — children produce grammatical sentences they've never heard, and do so on a broadly similar timetable across every studied language. His argument is that the human brain comes prewired with some capacity specifically built to acquire grammar, not just language in general but its structural rules. Consistent with a built-in, biologically timed capacity, there also appears to be a critical or sensitive period for language acquisition — a developmental window (roughly early childhood into puberty) during which language is acquired easily and natively, and outside of which full native-like acquisition, particularly of grammar, becomes markedly harder.
Does your language shape what you can think?
Once you have language, a natural next question is whether the specific language you speak changes how you think — not just what you can say, but what you can perceive or reason about. This is linguistic relativity, associated with Benjamin Lee Whorf.
The myth
The strong version of this idea — sometimes called linguistic determinism — holds that language determines thought: if your language has no word for a concept, you literally cannot think it, and speakers of different languages live in genuinely different conceptual worlds because of their vocabulary.
What's actually true
The strong, deterministic version is rejected by contemporary psycholinguistics — people routinely think about, perceive, and later name concepts their language hasn't lexicalized yet, and translation between languages, while imperfect, is clearly possible, which shouldn't be true if thought were locked inside vocabulary. A moderate version, though, holds up well: language doesn't determine what you can think, but it does nudge what you habitually notice, attend to, and categorize. Languages that grammatically mark distinctions speakers have to attend to (for example, some languages require the speaker to mark direction with absolute compass terms rather than "left/right") measurably shift habitual attention toward those distinctions. Language shapes the well-worn paths of thought, not the boundaries of what's thinkable.
Intelligence: one thing, several things, or the wrong question?
Now to the argument that has run through psychology for over a century: what is intelligence, and can it be meaningfully measured with a single number? Charles Spearman proposed g, a general intelligence factor — his observation was that people who score well on one type of mental test tend to score well on others too, suggesting a shared underlying general ability running through all of them. That correlation among different test types is real and well replicated. Whether it's best explained by one general capacity, however, is where things split.
Spearman's g
One general factor underlies performance across different cognitive tasks; scores on varied mental tests correlate with each other, supporting a shared underlying ability. Highly influential in the psychometric tradition and still central to most modern IQ tests.
Multiple-intelligences views
Gardner (1983) proposed several relatively independent intelligences — for example linguistic, logical-mathematical, spatial, musical, bodily-kinesthetic, interpersonal, and intrapersonal — arguing a single g score misses real, distinct competencies. Sternberg's triarchic theory similarly split intelligence into analytical, creative, and practical components. Both views are popular in education, but remain contested among researchers who note the multiple "intelligences" aren't as independent, or as well measured, as the theories claim.
A separate, better-evidenced split within the psychometric tradition itself comes from Raymond Cattell: fluid intelligence is the capacity to reason and solve novel problems without relying on prior knowledge, while crystallized intelligence is accumulated knowledge and skill built up through experience and education. Fluid intelligence tends to peak relatively early in adulthood and gradually decline with age; crystallized intelligence tends to hold steady or even grow across most of adulthood — which is part of why a sixty-year-old can out-argue a twenty-year-old on accumulated expertise while being slower on a novel abstract-reasoning puzzle.
What an IQ score actually is — from Binet to Wechsler
The measurement story starts practically, not grandly. In 1905 Alfred Binet and Théodore Simon were hired by the French government for a modest task: identify schoolchildren who needed extra academic help. Their test compared a child's performance to age norms — a "mental age" — and Binet himself insisted the score was a snapshot of current functioning, not a fixed measure of worth. (History, as the next section shows, did not honor that insistence.) Lewis Terman's Stanford revision — the Stanford-Binet — brought the test to America and popularized the ratio "IQ": mental age ÷ chronological age × 100.
The ratio broke down for adults (a 40-year-old with a "mental age" of 40 is just… an adult), so David Wechsler rebuilt the scoring. His tests — the WAIS for adults and WISC for children, still the most widely used individual IQ tests today — introduced two upgrades. First, subtests: separate scores for vocabulary, working memory, processing speed, visual-spatial reasoning, and more, so the result is a profile rather than one bare number. Second, the deviation IQ: your score reflects where you stand relative to other people your age, placed on a normal curve with the mean set at 100 and a standard deviation of 15. An IQ of 115 doesn't mean "115 units of smart" — it means one standard deviation above the average of your age group, roughly the 84th percentile. The bell curve from your statistics toolkit is doing all the work.
Two report-card words tell you how good these tests are, and the honest answers differ. Reliability — does the test give consistent scores? — is excellent: modern IQ tests are among the most reliable instruments psychology has ever built, with retest correlations around .9. Validity — does it measure what matters? — is real but bounded: IQ scores predict school achievement substantially and job performance moderately, yet leave most of the variation in life outcomes unexplained, and they say nothing about honesty, creativity, wisdom, or what a person does with a supportive environment. A tool this reliable is genuinely useful; a tool this bounded should never be treated as a verdict on a person — which is exactly the misuse the next section documents.
A history worth knowing, not just a formula
IQ testing has a genuinely troubled history alongside its genuine usefulness, and an honest treatment doesn't skip either half. Early IQ tests were, at various points, misused to justify discriminatory immigration policy and coercive eugenics programs — a real historical harm caused by treating a test score as a fixed, complete measure of a person's worth or potential rather than the limited, context-bound estimate it actually is. That history is part of why any serious modern discussion of intelligence testing insists on being careful about what scores can and cannot support.
One finding cuts directly against a purely fixed, genetics-only view of intelligence: the Flynn effect — the documented rise in average IQ scores across many countries over the 20th century, substantial enough that test makers have had to periodically renorm their tests just to keep the average at 100. Genes don't change meaningfully across a few generations, so a genetics-only account can't explain a population-wide rise this fast; something environmental — better nutrition, more schooling, more cognitively complex daily environments — has to be doing real work. The Flynn effect doesn't mean genes are irrelevant to individual differences in intelligence. It does mean that the score itself is sensitive to environment and history, not a fixed biological ceiling stamped at birth.
◆ Test fairness and stereotype threat
Group differences in average test scores have also been used, historically and still today, to make claims about innate group ability — claims the field treats with real caution, because test performance is sensitive to more than raw capacity. Steele and Aronson (1995) demonstrated stereotype threat: when people are reminded, even subtly, of a negative stereotype about their group's ability on a task, their performance on that task can suffer, apparently because of the extra cognitive load of monitoring for and worrying about confirming the stereotype. It's an important finding precisely because it shows situational factors, not just fixed ability, can move a test score. That said, in good scientific practice this is presented even-handedly: the original stereotype threat studies have since been subject to replication debates, with some later, larger studies finding smaller or less consistent effects than the original reports. Neither the initial excitement nor a dismissal of the whole idea is the right final word — the responsible summary is that testing conditions matter for scores, the size of stereotype threat specifically is actively debated, and both genetic and environmental influences, along with the fairness of the tests themselves, remain genuinely open, actively studied questions rather than settled ones (Neisser et al., 1996).
- Early 1900s First IQ tests developed and, at points, misapplied to justify discriminatory policy — the origin of intelligence testing's troubled history.
- 1980s Alternative models challenge a single-score view: Gardner (1983) proposes multiple relatively independent intelligences.
- 1990s Steele and Aronson (1995) identify stereotype threat as a situational factor that can depress test performance; Neisser et al. (1996) publish a field-wide consensus statement on the "knowns and unknowns" of intelligence research.
- Across the 20th century The Flynn effect — the steady rise in average IQ scores — is documented across many countries, forcing periodic renorming of tests.
- 2000s–present Ongoing replication debates over stereotype threat's size and consistency; the field continues to insist on situating any test score within its environmental and cultural context.
A popular idea that isn't supported: learning styles
The myth
You've probably heard, maybe even been told directly by a teacher, that you're a "visual learner" or an "auditory learner," and that you'll learn better if instruction is matched to that style — visual materials for visual learners, lectures for auditory learners, and so on. It's one of the most widely believed ideas in education.
What's actually true
Pashler, McDaniel, Rohrer, and Bjork (2008) reviewed the evidence for the "meshing hypothesis" — that matching teaching style to a student's preferred learning style improves learning outcomes — and found it essentially unsupported. People do have preferences for how they like to receive information, but preference is not the same as improved learning. For the hypothesis to hold up, you'd need studies showing that visual learners taught visually outperform visual learners taught in another format, and vice versa for other "styles" — and that specific interaction pattern essentially doesn't show up in well-designed research. What does reliably improve learning, regardless of anyone's stated preference, are general good-evidence strategies like spaced practice and self-testing, not style-matching.
◆ Update: why the myth persists anyway
The learning-styles idea keeps circulating in schools and workplaces despite the evidence, partly because the underlying observation — "I like this format better" — is true and intuitive, even though the conclusion drawn from it ("so I'll learn more from it") doesn't hold up under testing. It's a useful case study for the whole unit: an intuitive, System-1-friendly belief that feels obviously true and spreads for that reason, independent of whether the evidence actually supports it.
The one thing to carry out of this unit
Your mind runs on shortcuts — in categorization, in problem-solving, in judging probability, and even in how confidently you hold beliefs about testing and ability — and those shortcuts are mostly good deals, not defects. The habit worth building isn't distrust of your own thinking. It's knowing which situations (rare risks, statistics, group-level claims, anything wrapped in a stereotype) are exactly the ones where the fast, automatic system is most likely to mislead you, and where it's worth paying the toll for the slow one.
References
Binet, A., & Simon, T. (1905). Méthodes nouvelles pour le diagnostic du niveau intellectuel des anormaux. L'Année Psychologique, 11, 191–244.
Gardner, H. (1983). Frames of mind: The theory of multiple intelligences. Basic Books.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Neisser, U., Boodoo, G., Bouchard, T. J., Boykin, A. W., Brody, N., Ceci, S. J., Halpern, D. F., Loehlin, J. C., Perloff, R., Sternberg, R. J., & Urbina, S. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51(2), 77–101.
Pashler, H., McDaniel, M., Rohrer, D., & Bjork, R. (2008). Learning styles: Concepts and evidence. Psychological Science in the Public Interest, 9(3), 105–119.
Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69(5), 797–811.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131.
Wechsler, D. (2008). Wechsler Adult Intelligence Scale (4th ed.). Pearson.