Psychology, oriented

Introspection can mislead, because many mental operations are not open to conscious inspection, so psychology relies on evidence others can check. Wundt's laboratory, formally recognized in 1879, and Titchener's trained observers tried to describe conscious experience exactly, while James asked what the mind is for. When trained observers could not settle their disputes, behaviorism limited the science to observable behavior. It left behavior therapy but ignored how a person represents a situation, and around 1956 the cognitive revolution brought mental processes back as models tested by experiment and simulation. Freud's psychoanalysis, enormously influential but weakly supported, and the humanistic psychology of Rogers and Maslow ran alongside.

Engel's 1977 biopsychosocial model holds that biological, psychological and social factors all shape health and illness. Say a nurse on rotating night shifts sleeps badly. Her body clock is out of step, she lies awake rehearsing worries, and her schedule leaves no room for a regular bedtime. Each level contributes, and none tells the whole story.

The trap is using the model as a license to pick. Because it names no method, you can stress whichever level you already favor; the usual guard is to walk through every level on purpose.

Rule: for any explanation of behavior, name its level and ask what evidence another person could check.

A correlation shows that things go together, but not which way the influence runs or whether an unmeasured third variable drives both. An experiment manipulates the presumed cause, the independent variable, and measures the dependent variable. Random assignment makes groups equal on average, a placebo gives both groups the same expectations, and a double-blind design hides who got what from researchers as well as participants.

Say a survey finds that students who sleep more get better grades. Sleep might help grades, good grades might ease sleep, or a lighter work schedule might drive both; randomly assigning sleep schedules would separate those stories.

A score must be reliable, or consistent, and valid, meaning it reflects the construct it claims to measure: a quiz that gives you the same score every time can still be tracking reading speed rather than understanding. Studies from the US find that WEIRD participants, Western, educated, industrialized, rich and democratic, make up roughly 70 to 90 percent of psychology samples.

Ethics rules followed real harm: Milgram ran his obedience studies before review boards existed, and the Tuskegee study withheld treatment from 1932 to 1972 without informed consent. Under US rules, consent must now be voluntary and revocable, and a review board weighs risks against benefits first.

The trap is reading a p-value as the chance the result is true. It is the chance of data at least this extreme if every assumption of the test, the null hypothesis included, held. A typical psychology effect needs about 99 people per group for 80 percent power in a two-group design, and when researchers repeated 100 studies from three leading journals, well under half came out significant again.

Rule: before believing a finding, ask what design supports its causal claim, how large the sample was, and whether anyone has repeated it.

A neuron fires an action potential only when its combined input reaches threshold; below threshold there is no spike at all. Excitatory transmitters such as glutamate push the next cell toward firing, and inhibitory ones actively push it away, so inhibition is a signal, not just silence. At the synapse, vesicles release transmitter, it binds receptors on the next cell, and the signal ends when transporters pull it back or an enzyme breaks it down. Drugs act on that cycle: an agonist mimics a transmitter's action, an antagonist blocks one, and cocaine blocks the transporter that clears dopamine.

Some functions lean to one side, such as forming words, but a scan study of 1,011 people found that lateralization varies network by network, so the left-brained versus right-brained idea fails.

Say a news story says a brain region lit up while people scrolled social media and calls that proof of craving. That is reverse inference, and it fails whenever the region also does other jobs.

The trap is reading heritability as a fraction of you. It is the share of variation across a population that can be attributed to genetic variation, so a heritability of one half says nothing about how much of your own trait comes from genes.

Rule: keep levels apart; a population statistic never describes one person, and an active region never proves one particular process.

Consciousness is graded, not a switch, and bedside behavior can understate it: scans have shown signs of following commands in some patients diagnosed as unresponsive. Your body clock, run by the suprachiasmatic nucleus in the hypothalamus, keeps a roughly 24-hour rhythm with light as its strongest cue, which is why shift work and jet lag throw it off. You cycle through three stages of non-REM sleep and rapid eye movement (REM) sleep roughly every 80 to 100 minutes. Deep slow-wave sleep comes in longer stretches early, REM periods grow longer toward morning, and during REM your arm and leg muscles are temporarily paralyzed so you do not act out dreams.

Say you set your alarm two hours early: you cut off the end of the night, when REM periods run longest.

With repeated drug use the brain adapts, for example with fewer receptors in reward circuits, so the same dose does less: tolerance. Stop abruptly and the adaptation is left unopposed: withdrawal. Alcohol taken with sleep medicines or opioids raises overdose risk because the effects stack.

The trap is treating dependence and addiction as one thing. Dependence is the body's adaptation, revealed by withdrawal and usually manageable by tapering; addiction is compulsive use despite harm. Someone taking a prescribed opioid for weeks can become dependent without being addicted.

Rule: ask what the brain adapted to, and separate the body's adaptation from compulsive use before you name either.

Each sense has its own converter: photoreceptors turn light into electrical signals, and in the cochlea hair cells turn the motion of a traveling wave into signals. Perception is what the brain makes of those signals, and it can go beyond them, as when a few wedge-shaped gaps make you see a triangle nobody drew. Color shows how stages stack: three cone types code color first and opponent cells downstream compare them, so the trichromatic and opponent-process theories describe successive steps rather than rivals.

Grouping settles on the simplest overall organization, constancy keeps a familiar object the same size as it moves away, and illusions such as the Ponzo exploit that machinery. Attention limits what you notice: about half of viewers counting ball passes miss a person in a gorilla suit. Weber's law says the smallest change you can detect grows with the size of the starting stimulus.

Say two lifeguards each spot 9 of 10 swimmers in real trouble. One also blows the whistle at every splash; the other almost never does. Their hit rates match; only the false alarms show that one guard is quicker to say yes.

The trap is judging a detector by hits alone. Signal detection theory separates sensitivity from the criterion for saying yes, and a falling hit rate can come from either, so you need the false alarms to tell which.

Rule: judge any detector, whether a sense, a person or a test, by its hits and false alarms together.

In classical conditioning, an unconditioned stimulus such as food triggers a response without learning, and a neutral cue paired with it becomes a conditioned stimulus that triggers a response on its own. Learning depends on how well the cue predicts the outcome, not on pairing alone. Extinction does not erase the link; it layers a second, context-bound association on top, so the response can return with time, in a new place, or when the outcome recurs. Taste aversion bends the usual rules: the taste-illness gap can be hours, and ill rats in Garcia and Koelling's 1966 study avoided a flavor but not the lights and noise that came with it.

In operant learning, consequences select behavior. Reinforcement increases a behavior and punishment decreases it; positive means something is added and negative means something is removed. Behavior rewarded only some of the time extinguishes less readily than behavior rewarded every time, though some studies found the reverse. Tolman's famous shortcut test of latent learning replicated poorly: across about a dozen replications, only 17 percent of experiments showed shortcutting.

Say a child whines at the checkout and a tired parent sometimes hands over candy. The candy positively reinforces the whining, the quiet that follows negatively reinforces the giving in, and the unpredictable payoff can make the whining slow to fade.

The trap is calling negative reinforcement punishment. "Negative" means removal, and removing something unpleasant makes a behavior more likely, the opposite of what punishment does.

Rule: label the effect first, up or down, then ask whether something was added or removed.

A memory has to be encoded, stored and later retrieved, and each step fails differently. In an alcohol blackout the memory was never stored; after misleading questions, a stored memory is rebuilt wrongly at retrieval. Short-term memory holds little, about seven items in Miller's classic estimate, and chunking works around that limit by recoding many items into a few meaningful units.

Retrieval is reconstruction. In Loftus and Palmer's study, people asked how fast the cars "smashed" gave higher speed estimates than people asked about cars that "hit", and later falsely recalled broken glass. The classic finding that divers recall best where they learned did not hold in a 2021 replication with sixteen divers.

Say a witness hesitantly picks someone from a fair lineup, later learns that person is the suspect, and feels certain by the trial. Under clean conditions, high confidence recorded right after a first, double-blind identification signals high accuracy, which is why it should be written down before anything can contaminate it. The certainty at trial is not the one to trust.

The trap is trusting how a study method feels. Retrieval practice, spacing and interleaving take more effort than rereading or cramming and can leave you less confident, and that difficulty is what helps long-term learning.

Rule: ask which stage failed, treat vivid recall as a reconstruction, and study by pulling material back out of memory.

Prototype theory says something belongs to a concept when it has enough of the concept's typical features, which is why an apple feels like a better fruit than a plum. Heuristics are fast rules of thumb whose errors are systematic enough for researchers to name. Availability rates an event as likely when instances spring readily to mind; representativeness rates a case by resemblance to a type, neglecting base rates. Estimates adjust too little from an anchor, and confirmation bias favors what you already believe.

Scores on very different mental tests correlate positively, which Spearman summarized as a general factor, g, though what g is remains debated. An IQ is not a count of right answers: it places a performance against same-age peers on a scale with a mean of 100 and a standard deviation of 15. Average scores rose about two to four points per decade over much of the twentieth century, so old norms inflate scores and tests are re-normed; the gains have since slowed, stalled or reversed in some countries.

Say a child scores 115 on a test normed two decades earlier: one standard deviation above those old peers, but possibly lower against current norms.

The trap is assuming a famous effect is settled. Stereotype threat is contested: a meta-analysis of 47 lab studies of gender and math warned that publication bias may have seriously distorted the estimate. The separate brain systems Gardner's multiple intelligences predict have not been found, and a group score gap alone does not prove a test is biased.

Rule: treat a quick judgment as a heuristic that can misfire, and read any score against its norm group and its date.

Development is gene-environment interaction: a gene's influence can change with the environment, and the reverse. Most birth defects arise in the first three months of pregnancy, though no stage is fully protected. Piaget proposed four stages, from infants gaining object permanence to abstract thought from about 11, with conservation, grasping that rearranging a quantity leaves the amount unchanged, from about 6 or 7. Vygotsky's zone of proximal development covers tasks just beyond what a learner manages alone, and scaffolding is guidance matched to that zone, withdrawn as skill grows.

Ainsworth's Strange Situation sorts infants by how they respond when a caregiver returns, and meta-analyses link those patterns only modestly to later peer competence and behavior. Planning and decision-making rely on the prefrontal cortex, which finishes maturing late; the brain goes on developing into the mid-to-late twenties.

Say a study tests 30-year-olds and 70-year-olds, and the older group scores lower. The gap may be aging or a cohort difference between generations; following the same people over time separates the two, at the cost of dropouts and practice effects.

The trap is reading a link between parenting style and child outcomes as one-way cause. Much of that evidence is cross-sectional, so the influence could run either way, and longitudinal work is still needed to settle it.

Rule: for any age or parenting difference, ask whether the design separates the cause you care about from cohort and from direction.

Drive-reduction theory treats motivation as an unpleasant state, like hunger, that pushes you to act, which fits restoring set points but not the need for stimulation: in one study people gave themselves painful shocks rather than sit with nothing to do. In Yerkes and Dodson's mouse experiments, performance rose steadily with shock intensity on an easy task but peaked at a middle level on a hard one, so their result is two patterns, not one curve. Maslow's pyramid is intuitive, but a review of 13 studies found little evidence that needs emerge in a fixed order.

An emotion bundles appraisal, bodily change, feeling, expression and an urge to act, and theories differ over which parts are essential. James put the bodily response first, Cannon and Bard made feeling and arousal simultaneous, and Schachter and Singer had arousal take its label from the situation. The pen-in-the-teeth facial feedback effect failed in a multi-lab replication with 1,894 participants.

At work, job analysis comes first, showing which requirements are genuinely job-related. Cognitive ability tests, structured interviews and work samples all predict performance.

Say a colleague writes the team newsletter because she enjoys it, and a manager starts paying her per issue. A performance-based reward can lower intrinsic motivation, the undermining effect, though only where interest existed and not in every study.

The trap is assuming more reward always means more motivation, when for an already enjoyed task it can mean less.

Rule: ask whether a behavior is pushed by a need or pulled by an incentive, and check what a new reward does to the reason someone was already acting.

Most personality researchers today are interactionists: behavior depends on both person and situation, and the open question is how to describe stable differences, not whether they exist. Factor analysis of everyday trait words repeatedly found five broad traits: openness, conscientiousness, extraversion, agreeableness and neuroticism. Vague, flattering profiles feel accurate, the Barnum effect, which helps low-validity personality tests keep being accepted.

Situationism holds that minor features of a situation often sway behavior more than character does; in one study, hurried passersby helped a person in distress far less often, 10 versus 63 percent. Explaining someone's behavior by their disposition even when an obvious outside cause is present is the fundamental attribution error. In a 2023 replication of Asch's line studies, people gave the group's wrong answer on 33 percent of judgments.

Say a coworker snaps at you in a meeting. Your first explanation is that she is rude; the one you skip is that she just got bad news.

The trap is quoting Milgram as proof that most people obey. In his best-known version 65 percent went to the maximum shock, but across about 30 variants obedience ranged from 0 to nearly 100 percent, most participants overall stopped short, and replications have had mixed results.

Rule: before explaining a behavior by character, check the situation, and before citing a classic study, check its variants and its replications.

A stressor is the event; stress is the response, shaped by appraisal, which in Lazarus and Folkman's model weighs how much a stressor matters and whether your resources can meet it.

No single mark defines a mental disorder. Distress, dysfunction and statistical rarity each fail alone, since being unusual is not being disordered and ordinary grief can resemble depression. The diathesis-stress model proposes that stress activates a latent vulnerability and symptoms appear once a threshold is crossed, which would explain why the same stress affects people differently; in one large genetic test of the idea, the effect was tiny. The popular claim that low serotonin causes depression overreached: a large umbrella review found no consistent evidence for it.

Say a man who fears elevators always takes the stairs. In cognitive models, each avoided ride brings relief and denies him evidence that the feared outcome does not happen, so the fear survives. Exposure therapies target that loop with feared situations faced in brief, supported doses. Cognitive behavioral therapy teaches people to notice and question unhelpful thoughts; for depression, psychotherapy's effects are comparable to antidepressants', with evidence that CBT's last longer by preventing relapse.

The trap is treating one mark, or a list of symptoms, as a diagnosis; none of this is a basis for judging yourself or anyone else. Among the ingredients therapies share, the working alliance between client and therapist is the most robust predictor of outcome.

Rule: ask what keeps a problem going, and what evidence a treatment rests on, before naming either.