Skip to content

Unit 2 of 5

Cognition

20% of the exam252 reviewed questions116 key terms25 min read

Cognition is worth about 20% of the AP Psychology exam. Our bank tests it with 252 reviewed questions across 116 distinct ideas — 55 easier, 138 medium and 59 harder.

On this page · 12 sections

This unit is about what your mind does with information once it gets in. You take in a flood of sensory data, keep a tiny slice of it in working memory, and store some of that slice for years. Later you pull it back out, often in a slightly rebuilt form. The same machinery then gets used to solve problems, make decisions, learn a language and score on intelligence tests. Memory is the largest piece of the unit and the one the exam returns to most, so most of this guide is about it.

The pieces connect through one theme: the mind is fast and efficient, and that efficiency has costs. Chunking lets you beat the limits of short-term memory, but the limit is still there. Retrieval cues make recall easy, but only when the cues at test match the cues at study. Memory rebuilds the past from fragments, which is why a leading question can plant broken glass that was never there. Heuristics let you decide in a second, which is why the availability heuristic makes plane crashes seem more common and more dangerous than they really are. Many people fear flying more than driving, even though driving is far riskier per mile.

Almost every idea here comes attached to a researcher and a study. Sperling flashed letter grids, Miller counted chunks, Craik and Lockhart tested depth of processing, Loftus asked about smashed cars, and Kahneman and Tversky catalogued judgment errors. Chomsky argued that children are built to learn grammar, and Sternberg split intelligence into three kinds. If you know what each study did and what it showed, you can answer most of the questions this unit produces.

The unit in a minute

One line per topic. Tap any line for the full notes.

  1. Perception: processing, organization, attention and depthExpectation shaping what you see is top-down processing and perceptual set; binocular cues need two eyes, monocular cues need only one.
  2. Sensory memory, short-term limits and working memorySperling's partial report shows sensory memory is large but fades fast; Miller's 7 plus or minus 2 counts chunks, not seconds.
  3. Encoding depth and what actually works when you studyMeaning beats sound, sound beats appearance; self-testing spread over time beats re-reading and cramming.
  4. Types of long-term memory, storage and what amnesia revealsHippocampal damage breaks new explicit memory but spares implicit skill learning; anterograde means after the injury, retrograde means before.
  5. Retrieval cues, matching and forgettingMatch the cue: room is context-dependent, body state is state-dependent; old blocks new is proactive, new blocks old is retroactive.
  6. Reconstructive memory, false memories and the serial position effectMemory is rebuilt, so post-event information can change it; primacy comes from long-term memory, recency from working memory.
  7. Problem solving, heuristics and cognitive biasesName the exact shortcut: easy-to-recall examples is availability, fits-the-type is representativeness, a streak of chance is the gambler's fallacy, already-paid is sunk cost.
  8. Language acquisition'Goed' is overregularization, evidence for rule learning and the nativist view, not for imitation or conditioning; Chomsky's core argument was the poverty of the stimulus.
  9. Intelligence and intelligence testingReliable means consistent, valid means accurate; Sternberg's three are analytical, creative and practical; fluid declines with age while crystallized holds or rises.
Grind Cognition questionsAlready know this? Skip the notes and prove it.

The big ideas

Memory is a set of systems, not one box

Sensory memory, working memory and long-term memory differ in how much they hold and how long they hold it. Long-term memory splits again into explicit memory, which you can consciously report, and implicit memory, which shows up only in performance. These systems run on different brain structures, so damage can wipe out one and spare another. That is why a patient with hippocampal damage can get better at a video game every day while insisting he has never played it.

How you encode decides what you can retrieve

A memory is only as good as the processing that built it and the cues that can reach it. Thinking about meaning builds stronger traces than thinking about sound or spelling. Retrieval works best when the cues present at test match the cues present at study, whether those cues are a room, a mood or a drug state. A lot of forgetting is retrieval failure, not erasure: the memory is stored, but the cues you have cannot reach it. Forgetting can also come from encoding failure, decay or interference.

Remembering is rebuilding

You do not replay memories like video. You reconstruct them each time from stored fragments, schemas and anything you learned afterward. That process is usually good enough, but it lets leading questions, later information and expectations slip into the memory itself. Confidence is not a reliable sign of accuracy, even for vivid flashbulb memories.

Shortcuts make thinking fast and sometimes wrong

Algorithms guarantee an answer but are slow, while heuristics are quick rules of thumb that usually work. Kahneman and Tversky showed that heuristics produce predictable errors, such as judging likelihood by how easily examples come to mind. Mental set, confirmation bias, overconfidence and framing are all cases of a normally useful habit misfiring. On the exam you name the specific shortcut that a scenario shows.

Language and intelligence are partly built in and partly built by experience

Chomsky argued that children are born ready to learn grammar, and errors like 'goed' support the nativist view because no adult models them. Intelligence also sits between nature and environment, and psychologists still argue over whether it is one general ability or several. Tests of it are only useful if they are standardized, reliable and valid. The exam asks you to judge both the theories and the tests.

Study notes

9 topics. Each one opens with its key idea, then the detail you need to use it.

1. Perception: processing, organization, attention and depth

Key idea: Expectation shaping what you see is top-down processing and perceptual set; binocular cues need two eyes, monocular cues need only one.

Perception is how your brain turns raw sensation into a meaningful scene. Bottom-up processing starts with the stimulus and builds up from its features, such as lines, edges and colors. Top-down processing starts with what you already know and expect. A schema is a mental framework for a type of thing or event, and it drives top-down processing. Perceptual set is a readiness to perceive something a certain way. Show people a shape between the letters A and C and they read it as B. Put the same shape between 12 and 14 and they read it as 13. The stimulus did not change, only the expectation did.

Gestalt psychologists argued that the brain organizes parts into wholes. Figure-ground means you split a scene into an object and its background, as in the vase that can also be seen as two faces. Proximity means you group things that are close together, so six dots in three pairs look like three units. Similarity means you group things that look alike, so rows of matching colors stand out. Closure means you fill in gaps, so a circle with a missing piece still looks like a circle.

Attention is limited, so you see far less than you think. Selective attention is focusing on one input while tuning out others, like following one voice at a loud party. Inattentional blindness is missing something in plain view because your attention is elsewhere. In Daniel Simons and Christopher Chabris's 1999 study, about half of viewers counting basketball passes missed a person in a gorilla suit. Change blindness is failing to notice that something changed, such as a stranger asking you for directions being swapped for a different person mid-conversation.

Depth cues let a flat image on the retina produce a 3D world. Binocular cues need both eyes. Retinal disparity is the difference between the two eyes' images, which is larger for near objects. Convergence is your eyes turning inward, which they do more for near objects. Monocular cues work with one eye, and artists use them to fake depth on a flat canvas. With relative size, the smaller of two similar objects seems farther away. With interposition, an object that blocks another seems closer. With linear perspective, parallel lines like railroad tracks seem to meet in the distance. Others include relative height, texture gradient, and light and shadow.

2. Sensory memory, short-term limits and working memory

Key idea: Sperling's partial report shows sensory memory is large but fades fast; Miller's 7 plus or minus 2 counts chunks, not seconds.

The Atkinson-Shiffrin model moves information through three stages: sensory memory, short-term memory and long-term memory. Sensory memory holds a raw copy of input for a moment. Iconic (visual) memory lasts well under a second, and echoic (auditory) memory lasts a few seconds. George Sperling showed this in 1960 by flashing a grid of 12 letters. When people tried to report the whole grid, they named only about 4 letters. But when a tone right after the flash told them which row to report, they could name 3 or 4 letters from any row. That partial report result means nearly the whole grid was briefly stored. When the tone came a second later, the advantage vanished, because the image had already faded.

Short-term memory is small and brief. George Miller put its capacity at about 7 plus or minus 2 chunks, and a chunk is any meaningful unit. 'FBI' is one chunk, not three letters, so chunking stretches capacity without breaking the limit. Duration is a separate property. Without rehearsal, items fade within roughly 15 to 30 seconds. In 1974 Alan Baddeley and Graham Hitch recast short-term memory as working memory, an active workspace rather than a waiting room. A central executive directs attention. A phonological loop holds sounds, and a visuospatial sketchpad holds images. Baddeley added a fourth part, the episodic buffer, in 2000, which links the other parts to long-term memory. Two verbal tasks at once interfere with each other, but a verbal task and a visual task interfere much less.

3. Encoding depth and what actually works when you study

Key idea: Meaning beats sound, sound beats appearance; self-testing spread over time beats re-reading and cramming.

Some encoding happens without effort. Automatic processing records things like where on a page you read a fact, or how often you have passed a coffee shop, with no attention needed. Effortful processing takes attention and work, like learning the terms in this unit. Effortful encoding is where your study choices make a difference.

Fergus Craik and Robert Lockhart argued in 1972 that durability depends on depth of processing, not on how long information sits in a store. Structural encoding, such as noticing whether a word is in capital letters, is shallow and fades fast. Phonemic encoding, such as noticing what a word rhymes with, is intermediate. Semantic encoding, which means thinking about what a word means, is deepest and lasts longest. Elaborative rehearsal goes further by linking new material to things you already know. The self-reference effect is one example, because relating a fact to your own life makes it stick.

The same logic explains which study habits work. Re-reading feels productive because the text grows familiar, but familiarity is not the same as being able to retrieve it. Highlighting and colorful summary sheets have the same weakness. Two strategies beat them in study after study. Retrieval practice, also called the testing effect, means quizzing yourself so that you practice pulling information out. Distributed practice, also called the spacing effect, means spreading study across several sessions instead of cramming it into one. Cramming can carry you through tomorrow morning, but it is forgotten quickly. Combining both, by self-testing in short sessions spread over days or weeks, gives the best long-term retention.

Metacognition is thinking about your own thinking, such as judging how well you really know a topic. Re-reading fools it, because familiar text feels learned. Self-testing gives it honest feedback, because you find out what you cannot recall. Mnemonics like 'My Very Educated Mother Just Served Us Nachos' help because they chunk a list and give each item a retrieval cue. The method of loci works another way. You picture each item at a spot along a route you know well, like the rooms of your house, then walk the route in your mind to recall them in order.

4. Types of long-term memory, storage and what amnesia reveals

Key idea: Hippocampal damage breaks new explicit memory but spares implicit skill learning; anterograde means after the injury, retrograde means before.

Long-term memory has two main branches. Explicit (declarative) memory is what you can consciously state. It includes episodic memory for personal events, like your last birthday, and semantic memory for facts, like the capital of France. Implicit (nondeclarative) memory shows up in behavior without conscious recall. Procedural memory for skills, such as riding a bike, is implicit. So is priming, where earlier exposure to a stimulus changes your response to a later one without your awareness. If you just saw the word 'yellow', you will recognize 'banana' slightly faster.

Storage has a physical basis. Long-term potentiation (LTP) is the strengthening of a synapse after repeated stimulation, so the receiving neuron fires more easily the next time. It is thought to be a cellular basis of learning and memory. Memory consolidation is the process that turns a fragile new trace into a stable long-term memory. The hippocampus helps hold new memories while they are consolidated and later stored more widely in the cortex. Sleep helps, because the brain replays the day's experiences during sleep, which is one more reason an all-nighter backfires.

These branches use different brain systems. The hippocampus is needed to form new explicit memories. The cerebellum and basal ganglia support procedural learning. The case of Henry Molaison (H.M.) made this clear. After surgery in 1953 removed most of both hippocampi to stop his seizures, he could not form new explicit memories. That is anterograde amnesia, the loss of memory for events after the injury. Yet when he practiced tracing a star in a mirror over several days, he improved each day while denying he had ever done the task. His procedural memory still worked. Retrograde amnesia runs the other way: it erases memories from before the injury, while new learning may be fine. The direction is the whole distinction, so check which side of the injury the lost memories fall on.

Infantile amnesia is a normal gap, not an injury. Most adults remember almost nothing from before about age 3, because the hippocampus was still maturing and they had little language to encode events with.

5. Retrieval cues, matching and forgetting

Key idea: Match the cue: room is context-dependent, body state is state-dependent; old blocks new is proactive, new blocks old is retroactive.

Retrieval cues are anything that helps you reach a stored memory. Recall makes you generate the answer with few cues, while recognition hands you the answer to identify, which is why multiple-choice feels easier than fill-in-the-blank. Endel Tulving's encoding specificity principle says a cue works best when it was part of the original encoding. Two specific cases follow from it. Context-dependent memory is about the external setting. In Godden and Baddeley's 1975 study, divers who learned words underwater recalled them better underwater, and words learned on land were recalled better on land. State-dependent memory is about your internal condition, such as a mood, caffeine or alcohol. Mood-congruent memory is a close cousin, because a sad mood makes sad memories easier to retrieve.

Hermann Ebbinghaus memorized nonsense syllables in the 1880s and found that forgetting is steep right after learning and then levels off. Several causes explain forgetting. Encoding failure means the information never got in, as with the details on a penny. Decay means unused traces fade. Retrieval failure means the memory is there but the cue cannot reach it, as in the tip-of-the-tongue feeling. Interference comes in two directions. In proactive interference, old learning disrupts new learning, as when Spanish vocabulary gets in the way of French. In retroactive interference, new learning disrupts old learning, as when your new phone number crowds out the old one. Pro means forward, so the old memory pushes forward into the new.

6. Reconstructive memory, false memories and the serial position effect

Key idea: Memory is rebuilt, so post-event information can change it; primacy comes from long-term memory, recency from working memory.

Memory is reconstructive. Each time you remember, you rebuild the event from fragments, schemas and anything you learned later. In Elizabeth Loftus and John Palmer's 1974 study, people watched a film of a car crash. Those asked how fast the cars were going when they 'smashed' gave higher speed estimates than those asked about cars that 'hit'. A week later, the 'smashed' group was more likely to report broken glass, though the film showed none. This is the misinformation effect. The verb activated a schema for a violent crash, and that schema was stitched into the memory. Source amnesia is a related error, where you keep the information but forget where it came from. Imagination inflation is another. Picturing an event over and over makes people more confident that it really happened, even when it did not.

Flashbulb memories feel like an exception, but they are not. After the Challenger explosion in 1986, Ulric Neisser had students write down how they heard the news. Years later their accounts had often changed, yet they felt just as sure. Vivid feeling and high confidence are the special features, not accuracy.

The serial position effect also shows how storage works. In free recall of a list, the first and last items are remembered best. The primacy effect happens because early words get the most rehearsal and reach long-term memory. The recency effect happens because the last words are still in working memory when recall starts. Add a 30-second distractor task before recall and recency disappears, while primacy survives, which shows the two effects come from different stores.

7. Problem solving, heuristics and cognitive biases

Key idea: Name the exact shortcut: easy-to-recall examples is availability, fits-the-type is representativeness, a streak of chance is the gambler's fallacy, already-paid is sunk cost.

Thinking runs on concepts, which are mental groupings of similar things, events or people. You often sort new items by comparing them to a prototype, the best example of a category. A robin is a prototype bird, so you agree that a robin is a bird faster than you agree that a penguin is one. Executive functions are the control skills that manage thinking, such as planning, holding back an impulse and switching between tasks. They depend heavily on the prefrontal cortex.

An algorithm is a step-by-step procedure that guarantees a solution, like trying every combination on a lock. A heuristic is a mental shortcut that is faster but can fail. Insight is a sudden realization of a problem's solution, an 'aha' moment, rather than a step-by-step solution. It can strike while you are still working on the problem. A solution that comes after you take a break is the related idea of incubation. Two obstacles block problem solving. Mental set is sticking with a method that worked before, even when a simpler one exists. Functional fixedness is seeing an object only in its usual role, so you miss that a coin can work as a screwdriver.

Creativity is the ability to produce ideas that are new and useful. Convergent thinking narrows down to one correct answer, which is what most multiple-choice tests reward. Divergent thinking spreads out to many possible answers, as when you list every use you can think of for a brick. Creativity tests measure divergent thinking.

Daniel Kahneman and Amos Tversky studied heuristics in the 1970s and showed that they produce predictable errors. The availability heuristic judges how likely something is by how easily examples come to mind. Plane crashes get heavy news coverage, so people overestimate their frequency compared with car crashes. The representativeness heuristic judges by how well something fits a prototype, which leads people to ignore base rates. A shy, tidy man seems like a librarian, though salespeople vastly outnumber librarians. Kahneman later described two modes of thinking: fast, intuitive System 1 and slow, deliberate System 2. Heuristics belong to System 1.

Cognitive biases are systematic errors in judgment. Confirmation bias is looking for evidence that supports what you already believe. Belief perseverance is clinging to a belief after it has been discredited. Hindsight bias is the 'I knew it all along' feeling. Overconfidence is being more sure than your accuracy justifies. Framing means the wording of a choice changes the decision, so '90 percent survive' sounds better than '10 percent die'.

Two fallacies are easy to mix up with the heuristics. The gambler's fallacy is believing that past results change the odds of an independent random event. After five reds in a row at roulette, black feels 'due', but the odds are the same on every spin. Pick it when a stem describes a streak of chance outcomes. Pick availability when the stem is about how easily examples come to mind, and representativeness when it is about fitting a type. The sunk-cost fallacy is sticking with something because of what you have already put into it. You sit through a terrible movie because the ticket cost $15, even though the money is gone either way. It is about past investment, not about judging how likely something is.

8. Language acquisition

Key idea: 'Goed' is overregularization, evidence for rule learning and the nativist view, not for imitation or conditioning; Chomsky's core argument was the poverty of the stimulus.

The current Course and Exam Description files language development under Unit 3, Development and Learning (topic 3.5), so the exam may test it in that unit's context. Language has layers. Phonemes are the smallest units of sound, morphemes are the smallest units of meaning, and grammar is the system of rules that combines them. Syntax covers word order and semantics covers meaning. Children move through a predictable sequence. They begin cooing, then babbling at around 4 to 6 months, which early on includes sounds from many languages. They say single words near their first birthday. By about age 2 they produce two-word, telegraphic speech, like 'want cookie', which keeps the content words and drops the rest.

B. F. Skinner argued that children learn language through imitation and reinforcement. Noam Chomsky attacked that view in 1959. He proposed that humans have an inborn language acquisition device and that all languages share a universal grammar. His central argument was the poverty of the stimulus. Children master grammar from input that is far too limited and messy to teach it, so some of the knowledge must be built in. Overregularization is strong evidence often cited for the nativist view. A child who once said 'went' starts saying 'goed' after learning the add-ed rule. No adult models 'goed', and correcting the child often does not stop it, so imitation and reinforcement cannot explain it. The error shows the child is applying a rule. A sensitive period for language supports the nativist side too. Children exposed to language late, like Genie, never fully master grammar, and adults learning a second language rarely lose their accent. Most psychologists now take an interactionist view: the capacity is inborn, but it needs social input to develop.

9. Intelligence and intelligence testing

Key idea: Reliable means consistent, valid means accurate; Sternberg's three are analytical, creative and practical; fluid declines with age while crystallized holds or rises.

Charles Spearman noticed that people who do well on one mental test tend to do well on others. He called the shared factor general intelligence, or g. Howard Gardner argued instead for multiple intelligences, including linguistic, spatial, musical and interpersonal. Robert Sternberg's triarchic theory names three kinds. Analytical intelligence is solving well-defined academic problems, which IQ tests measure. Creative intelligence is handling new situations. Practical intelligence is street smarts, the ability to adapt to everyday life.

Raymond Cattell split intelligence a different way. Fluid intelligence is reasoning quickly and abstractly about new problems, like solving an unfamiliar puzzle. Crystallized intelligence is the knowledge and vocabulary you have built up. Fluid intelligence declines from early adulthood on, while crystallized intelligence stays stable or even rises with age. That is why an older adult can beat a teenager at a crossword but lose at a fast logic puzzle.

Alfred Binet built the first practical test to spot French children needing extra help, using the idea of mental age. Lewis Terman revised it at Stanford, and IQ was first computed as mental age divided by chronological age, times 100. Modern tests like David Wechsler's WAIS use a deviation IQ instead. Scores are set on a normal curve with a mean of 100 and a standard deviation of 15, so about 68 percent of people score between 85 and 115. James Flynn found that average raw scores rose across generations in many countries, by about 3 IQ points per decade in the 20th century. This Flynn effect is why tests must be re-normed, or the average would drift above 100.

A test is only useful if it has three properties. Standardization means it is given the same way to everyone and compared against a large norm group. Reliability means it gives consistent results, for example on a retest. Validity means it measures what it claims and predicts what it should. A test can be reliable but not valid, like a scale that always reads five pounds heavy. Achievement tests measure what you have learned, while aptitude tests predict what you can learn.

Test bias and stereotype threat are different problems. A biased test is flawed itself, because it predicts performance less accurately for one group than another. Stereotype threat is a problem with the situation. Fear of confirming a stereotype about your group lowers your score, even on a fair test. Beliefs about ability matter too. Carol Dweck found that students with a growth mindset, who see ability as something effort can build, keep working after failure. Students with a fixed mindset see ability as set from birth, so they are more likely to give up.

Common mistakes

Drawn from the wrong answers students actually pick on these questions.

Choosing re-reading, highlighting, colorful summary sheets or a long cramming session as the best way to retain material for an exam.

Re-reading builds a feeling of familiarity, not the ability to retrieve, and cramming fades fast after the test. The answer the exam rewards combines retrieval practice (self-testing) with distributed practice (sessions spread over days). Neither primacy nor recency is a reason to cram.

Explaining a child's 'I goed to the store' as operant conditioning, latent learning or imprinting.

It is overregularization: the child has learned the add-ed rule and applied it to an irregular verb. Conditioning fails because nobody reinforces 'goed' and correction rarely stops it. Latent learning is Tolman's unreinforced learning, and imprinting is Lorenz's animal attachment, so neither fits.

Thinking Sperling's key finding was that people could recall only 4 or 5 letters from the grid.

That was the whole report condition, and it shows how little can be transferred out before the image fades. The finding that proves iconic memory is large is the partial report: cued to any row right away, people reported 3 or 4 of its letters. Delaying the cue killed the advantage, which shows the decay.

Treating Miller's 7 plus or minus 2 as a time limit, as exactly 7, or as a count of single letters and digits.

Miller described capacity, measured in chunks, with a range of about 5 to 9. Duration is a separate property of roughly 15 to 30 seconds without rehearsal. Rehearsal keeps items alive but does not raise the number of chunks you can hold.

Calling a same-room study effect state-dependent memory, or picking the broad term encoding specificity when the scenario is about a place.

An external setting matching study is context-dependent memory, and a matching internal state like caffeine or mood is state-dependent memory. Both are cases of encoding specificity, but when a stem describes one of them, choose the specific term. Also note that phonemic encoding is only intermediate depth, and semantic encoding is deepest.

Saying the first words of a list are remembered best because they are still in short-term memory, or because they stand out.

That reasoning describes the recency effect, which applies to the last words. The first words win because they got the most rehearsal and moved into long-term memory. That is why a distractor task before recall wipes out recency but leaves primacy intact.

Labeling 'black is due after five reds' as the availability or representativeness heuristic.

That is the gambler's fallacy, the belief that past results change the odds of an independent random event. Availability is about how easily examples come to mind, and representativeness is about how well something fits a type. Staying in a bad choice because of money already spent is the sunk-cost fallacy.

How this unit shows up on the exam

Cognition is about a fifth of the multiple-choice section, and most of its questions are applied scenarios rather than definitions. A typical stem describes a student, a patient or a participant and asks which concept best explains the result. A diver recalls better underwater, a patient improves at a game he cannot remember, a child says 'goed', or a shopper overrates a risk after seeing news coverage. The wrong options are usually real terms from the same unit, so you win by knowing the boundary between neighbours. That means context-dependent versus state-dependent, proactive versus retroactive, anterograde versus retrograde, availability versus representativeness, and primacy versus recency.

A second style names a study and asks what it showed or why. Expect Sperling, Miller, Craik and Lockhart, Loftus and Palmer, Ebbinghaus, H.M., Kahneman and Tversky, Chomsky and Sternberg. Know the method well enough to rule out options that describe a procedure the researchers never used. A third style asks you to judge a test or a strategy. You might decide whether a test is reliable, valid or both, or pick the study plan with the best research support.

This unit also feeds both free-response questions. The Article Analysis Question and the Evidence-Based Question can be built around a memory or decision-making study, since those are simple to run and measure. You may need to identify the variables in a levels-of-processing experiment or explain how a misinformation finding supports reconstructive memory. Readers reward a correct term applied to the specific detail in the prompt, not a definition on its own. Perception is part of the official unit too, so expect stems that test top-down versus bottom-up processing, perceptual set, Gestalt grouping, inattentional blindness and depth cues.

How the exam actually asks it

Real items from the AP Psychology bank, easier first. The answer is marked, and every wrong choice is explained — because knowing why a distractor is tempting is most of the work.

Worked example 1Easier
Top-down processing in perception refers to which of the following?
  1. AThe processing of sensory information from the raw physical stimulus upward to higher brain areas for interpretation
  2. BThe use of prior knowledge, expectations, and context to interpret sensory information, influencing what we perceiveCorrect
  3. CThe ability of the visual system to process the top portion of a visual scene before the lower portion
  4. DThe hierarchical organization of the visual cortex, in which simpler features are detected at lower levels and complex features at higher levels

Why B is right

Top-down processing refers to perception driven by higher-level cognitive processes — expectations, prior knowledge, context, language, and mental schemas — that influence how we interpret incoming sensory information. For example, seeing a blurry shape in context makes us perceive it differently than seeing the same shape out of context (as in reading: we perceive ambiguous letters in words based on semantic context). Top-down processing contrasts with bottom-up processing, which is driven purely by the physical properties of the stimulus.
Why the other 3 choices miss
A
Incorrect because processing from raw physical stimulus upward to higher brain areas describes BOTTOM-UP processing (data-driven); top-down is the opposite — higher-level knowledge comes 'down' to influence the interpretation of lower-level sensory data.
C
Incorrect because top-down processing does not refer to the spatial sequence of visual scanning (top of scene before bottom); it is a theoretical concept about the direction of information flow in the perceptual hierarchy — from higher cognition down to sensory processing.
D
Incorrect because the hierarchical organization of the visual cortex from simple to complex feature detection describes the visual processing hierarchy, which is more related to the physical architecture of the visual system; top-down processing specifically refers to cognitive factors (knowledge, expectation) influencing perception.
Worked example 2Easier
Perceptual constancy allows us to perceive stable objects in a changing environment. Which of the following best illustrates SIZE constancy?
  1. ASeeing a departing car as keeping its size even as its retinal image shrinksCorrect
  2. BPerceiving a coin as circular even when tilted and projecting an elliptical image on the retina
  3. CRecognizing that a friend is still the same person even when seeing them in different clothing or hairstyles
  4. DPerceiving a white piece of paper as white in both bright sunlight and dim indoor lighting despite different light levels

Why A is right

Size constancy is the tendency to perceive an object as the same size regardless of changes in the size of its retinal image as it moves closer or farther away. As a car drives away, its retinal image becomes smaller, but we do not perceive it as actually shrinking — we perceive it as the same car, now simply farther away. The brain uses distance cues to compensate for the smaller retinal image and maintains a stable perception of the car's size.
Why the other 3 choices miss
B
Incorrect because perceiving a tilted coin as circular despite its elliptical retinal image illustrates SHAPE constancy, not size constancy; shape constancy allows us to see objects as having stable shapes despite viewing angle changes.
C
Incorrect because recognizing the same person in different clothing illustrates object constancy or identity constancy, not specifically size constancy; the clothing change does not alter perceived size — it alters surface appearance.
D
Incorrect because perceiving a white paper as white under different lighting conditions illustrates LIGHTNESS (brightness) constancy; this is about perceived surface reflectance remaining stable across lighting changes, not about perceived size.
Worked example 3Easier
The Gestalt principle of similarity refers to which perceptual tendency?
  1. APerceiving elements that appear close together as belonging to the same group
  2. BPerceiving elements that share similar features (color, shape, size) as belonging to the same groupCorrect
  3. CPerceiving a figure as a distinct entity separated from its surrounding background
  4. DPerceiving parts of a figure as following a smooth, continuous path rather than an abrupt change in direction

Why B is right

The Gestalt principle of similarity states that we tend to group elements that share similar visual properties — such as color, shape, size, texture, or orientation — as belonging to the same perceptual unit. For example, in a grid of circles and squares, we naturally perceive all the circles as one group and all the squares as another group, even if they are physically intermixed, because similar shapes are grouped together.
Why the other 3 choices miss
A
Incorrect because grouping elements based on closeness in space describes the proximity principle, not similarity; proximity is about spatial distance between elements, while similarity is about shared visual properties.
C
Incorrect because perceiving a figure as distinct from its background describes the figure-ground principle, not similarity; figure-ground organization is about foreground/background segregation, not about grouping similar elements.
D
Incorrect because perceiving parts following a smooth, continuous path describes the continuity (good continuation) principle, which involves perceiving lines or contours that follow the smoothest path; this is distinct from similarity, which groups elements with shared features.
Worked example 4Easier
According to George Miller's classic research on short-term memory capacity, which statement most accurately describes STM's limitations?
  1. AShort-term memory holds information for approximately 7 seconds before decay, with capacity limited only by processing speed
  2. BShort-term memory can hold unlimited items if rehearsal prevents decay, with the 7-item limit applying only to visual information
  3. CShort-term memory can hold exactly 7 items with no variability, provided each item is a single letter or digit
  4. DShort-term memory has a capacity of 7 plus or minus 2 items (5-9 chunks), where a 'chunk' can be any meaningful unit regardless of sizeCorrect

Why D is right

Miller's 'magical number 7, plus or minus 2' describes short-term memory capacity as approximately 5-9 chunks, where a chunk is any meaningful unit of information. The critical insight was that chunks (not individual items) determine capacity — so a 7-letter acronym 'FBI' takes only 1 chunk of STM, not 3. Chunking allows people to expand the effective information held in STM, even though the number of chunks remains limited to about 7±2.
Why the other 3 choices miss
A
Incorrect because Miller's finding was about number of chunks (capacity), not duration of storage. Duration of STM without rehearsal is approximately 15-30 seconds (Peterson & Peterson), not 7 seconds. These are separate properties of STM.
B
Incorrect because the 7±2 limit applies to all modalities (visual, auditory, etc.), not just visual information. Also, STM capacity cannot hold unlimited items even with rehearsal — rehearsal prevents decay but does not increase the number of items that can be actively held.
C
Incorrect because Miller specified '7 plus or minus 2' (5-9), not exactly 7. More importantly, capacity is measured in chunks, not individual letters or digits. A phone number can be chunked as 3 chunks rather than 7-10 individual digits.

Can you do all of these?

Tick what you can do without looking. Anything left unticked is where to go back to.

0 of 12 ticked

Ready to test yourself?

Open practice pulls from the 44 free questions in this unit. The other 208 come with the one-time AP purchase, which unlocks every AP course.

Official topic list

The official course framework breaks this unit into 8 numbered topics.

  1. 2.1Perception
  2. 2.2Thinking, Problem-Solving, Judgments, and Decision-Making
  3. 2.3Introduction to Memory
  4. 2.4Encoding Memories
  5. 2.5Storing Memories
  6. 2.6Retrieving Memories
  7. 2.7Forgetting and Other Memory Challenges
  8. 2.8Intelligence and Achievement

Key terms in Cognition

Ordered by how often our reviewed questions test them. Open any term to see the questions that use it.

Perception27 questionsMemory21 questionsCognitive biases13 questionsHippocampus13 questionsWorking memory13 questionsDecision making12 questionsDepth perception12 questionsIntelligence12 questionsProblem solving11 questionsTop down processing11 questionsInferential statistics10 questionsRetrieval cues10 questionsAvailability heuristic9 questionsCognitive bias9 questionsEncoding specificity9 questionsLevels of processing9 questionsStandard deviation9 questionsStudy strategies9 questionsCognitive perspective8 questionsDescriptive statistics8 questionsLong term memory8 questionsNormal distribution8 questionsRetrieval practice8 questionsExpertise7 questionsForgetting7 questionsGestalt principles7 questionsIntelligence testing7 questionsKahneman Tversky7 questionsReconstructive memory7 questionsAnterograde amnesia6 questionsExpectation6 questionsExplicit memory6 questionsGestalt6 questionsImplicit memory6 questionsMental set6 questionsPerceptual learning6 questionsPriming6 questionsProcedural memory6 questionsResearch ethics6 questionsSensory memory6 questionsSternberg6 questionsVisual organization6 questionsAttention5 questionsCentral tendency5 questionsCognitive psychology5 questionsCognitive revolution5 questionsContext dependent memory5 questionsCorrelation coefficient5 questionsCraik Lockhart5 questionsCreativity5 questionsDeception5 questionsEpisodic memory5 questionsIconic memory5 questionsMonocular depth cues5 questionsRetrieval5 questionsSampling bias5 questionsSpacing effect5 questionsVariables5 questionsBaddeley4 questionsCentral executive4 questionsEmotional memory4 questionsMental age4 questionsPerceptual constancy4 questionsPerceptual grouping4 questionsPerceptual set4 questionsProactive interference4 questionsProbability judgment4 questionsRetrieval failure4 questionsSelf selection bias4 questionsSemantic encoding4 questionsSperling4 questionsState dependent memory4 questionsSurvey research4 questionsVisuospatial sketchpad4 questionsBinet3 questionsChunking3 questionsConfirmation bias3 questionsConstructive perception3 questionsCross modal perception3 questionsDistributed practice3 questionsDivergent thinking3 questionsEbbinghaus3 questionsEinstellung3 questionsElaborative encoding3 questionsEmpiricism3 questionsG factor3 questionsGibson Walk3 questionsKahneman3 questionsLoftus3 questionsMetacognition3 questionsMisinformation effect3 questionsMnemonics3 questionsMultiple intelligences3 questionsNativism3 questionsParallel processing3 questionsPartial report3 questionsPrimacy effect3 questionsPsychometrics3 questionsQuasi experimental3 questionsRadical behaviorism3 questionsRecency effect3 questionsReliability3 questionsRepresentativeness3 questionsRetroactive interference3 questionsRisk perception3 questionsSerial position effect3 questionsShort term memory3 questionsSize constancy3 questionsSocial desirability bias3 questionsSpearman3 questionsSunk cost fallacy3 questionsTesting effect3 questionsTriarchic theory3 questionsTulving3 questionsValidity3 questionsVisual cliff3 questions

The rest of AP Psychology

Every unit with its share of the exam. You are on Unit 2.

  1. 1Biological Bases of Behavior20% of exam
  2. 2Cognition20% of exam
  3. 3Development and Learning20% of exam
  4. 4Social Psychology and Personality20% of exam
  5. 5Mental and Physical Health20% of exam