AP Psychology Unit 2 Practice Test: Cognition MCQs (Free PDF)

Here's a free AP Psychology Unit 2 practice test: 20 AP-style MCQs on cognition, five scenario and data stimuli, and a full rationale for every answer, all aligned to the 2024-25 CED.

Below, three mistakes I see students make on Unit 2 and how to avoid them, plus one worked example so you can see the reasoning.

Unit 2 is 15-25% of the exam (every unit is weighted equally under the redesigned CED, so no single unit is "the biggest"). What makes Unit 2 different is that it stitches together perception, judgment, and the three big memory stages (encoding, storage, retrieval), and it's where the CED loads up on cognitive biases as a first-class skill. Students often lose marks not because they haven't studied, but because they mix up neighboring memory terms or answer a decision-making stem with the wrong bias label. For the full 5-unit map, see our AP Psychology units guide.

AP Psych MCQs mostly don't ask you to recall a definition in isolation. The 75-question multiple-choice section leans stimulus-heavy: many items sit on a scenario, a data table, a research vignette or a diagram, and you're expected to apply the concept, not just name it.

One framing point matters more than any other for Unit 2: cognitive biases show up as first-class content under Topic 2.2 and Practice 1.B. In practice that means a stem describing a choice or a judgment usually isn't asking you to remember what a bias is called; it's asking you to identify which bias fits the pattern of behavior in front of you. Confirmation bias, hindsight bias, overconfidence, framing, anchoring, availability and representativeness are the ones you'll see most often. For the wider research-skill lens, our AP Psychology research methods guide covers the study designs Unit 2 loves to test.

Common stimulus formats on Unit 2 items include:

• Perception scenarios with depth cues or Gestalt grouping (a bridge at dusk, a radiologist reading a scan)
• Framing and anchoring vignettes with a two-condition table
• Encoding vignettes contrasting shallow rehearsal with elaborative chunking
• Multi-store model prompts that test which memory system holds what
• Misinformation effect studies with a leading-word manipulation
• Interference examples (older learning disrupting newer, and vice versa)
• Psychometric evidence for a new intelligence test (reliability and validity)

The practice test in the PDF mirrors that mix. Each item is tagged to a Unit 2 topic (2.1 through 2.8) and to a named AP Psych Science Practice.

Every year I see students walk into Unit 2 MCQs making the same three moves. Catching yourself doing any of them buys you marks with no extra content revision.

The misinformation effect is the specific process in which information introduced after an event (a leading question, a rumor, a follow-up news story) gets folded into the original memory of that event. Students routinely describe it as "just forgetting" or as "remembering wrong," which loses the mechanism entirely. The misinformation effect was tested on the 2025 AP Psychology free-response item, so it's fair game and worth naming precisely.

The right framing on any Loftus-style stem: the exposure happens first, misleading post-event information arrives next, and the reconstructed memory later includes details that were never present. If a stem gives you a leading word ("smashed" versus "hit"), it's testing the misinformation effect, not proactive or retroactive interference. Interference is competition between two separate memory sets. Misinformation is distortion of one memory by later information.

Encoding is getting information in. Storage is keeping it. Retrieval is pulling it back out. Students commonly blur these three stages, so a stem about failing to "remember" a phone number gets labeled as an encoding problem when it's actually about short-term storage capacity, or a stem about a study environment gets labeled as encoding when it's really about a retrieval cue. This is one of the errors I see most often on mock papers: a student names the right concept but attaches it to the wrong stage of the memory system.

A clean rule of thumb: if the stem asks how information is first taken in (rehearsal type, chunking, imagery, self-reference, depth of processing), it's about encoding (Topic 2.4). If it asks about a memory system holding information (sensory versus short-term versus long-term, explicit versus implicit, semantic versus episodic versus procedural), it's about storage (Topic 2.5). If it asks about pulling information back out (context, state, mood, primacy or recency in a study list), it's about retrieval (Topic 2.6). Match the stage before you pick the term.

Reliability is about consistency: does the test produce the same score on repeated administration or across parallel halves? Validity is about accuracy: does the test actually measure what it claims to measure, and does it predict what it should predict? A test can be highly reliable without being valid (a bathroom scale that always reads two kilograms heavy is perfectly consistent, but not measuring your true weight), and validity depends on reliability without being the same property. On mock papers I regularly see students use "reliable" and "valid" as if they mean the same thing, which loses easy marks on Topic 2.8 items.

On any Topic 2.8 psychometrics stem, name the subtype before you commit. Reliability comes in test-retest (same test, two testings), split-half (correlation between halves of one testing), and inter-rater (agreement between scorers). Validity comes in content (does it cover the domain?), construct (does it measure the underlying trait?), and predictive (does it forecast a real-world outcome?). Two identical scores on two weekends is test-retest reliability. A strong correlation with later grade point average is predictive validity. Name the property, then name the subtype.

Here's a question straight from the Unit 2 pack. The stimulus: Students watch a short video of a two-car collision at an intersection. Students are randomly assigned to Group 1 or Group 2. One week later, Group 1 is asked, "About how fast were the cars going when they smashed into each other?" while Group 2 is asked, "About how fast were the cars going when they hit each other?" A week after that, both groups return and are asked whether they remember seeing any broken glass in the video, even though no broken glass was ever shown.

The stem: "Participants in the smashed group later report seeing broken glass in the video more often than participants in the hit group, even though no glass was shown. Which memory process best explains this result?"

• (A) Retroactive interference, in which newer learning disrupts the accurate recall of older material stored earlier.
• (B) Source amnesia, an inability to remember where a specific fact or claim was first learned by the person.
• (C) The misinformation effect, in which post-event information becomes incorporated into the memory of the event.
• (D) Proactive interference, in which older learning disrupts the accurate encoding of newer material learned later.

Answer: C. The misinformation effect describes the incorporation of misleading post-event information into a memory of the original event, which is exactly what a suggestive word like smashed produces on later recall. Interference effects, both proactive and retroactive, involve competition between two separate memory sets rather than distortion of a single event, so A and D don't fit. Source amnesia is about forgetting where information came from, not about false details being added to what you remember seeing, so B is out. The CED names Loftus as the illustrative researcher for this effect in Unit 2.

The Unit 2 memory topics are organized around three stages: encoding (2.4), storage (2.5), and retrieval (2.6). Most memory MCQs map onto one of the three, and the wrong-answer options often pull the correct term from a neighboring stage.

Before you read the options, decide which stage the stem is asking about. If the stem describes how information is first taken in (chunking, imagery, elaborative rehearsal, levels of processing), you're in encoding. If the stem describes a memory system holding information (sensory memory span, short-term or working memory capacity, long-term subtypes such as semantic, episodic, and procedural), you're in storage. If the stem describes pulling information back out (context cues, mood cues, position in a study list, tip-of-the-tongue), you're in retrieval.

Once you've named the stage, options that belong to a different stage move to the bottom of your list. They can occasionally still be right (stems that span stages happen), but the stage you named is where you should look first.

Source: AP Psychology CED, Topics 2.4-2.6.

Under Practice 1.B, cognitive biases are treated as a skill you apply to scenarios rather than a list you recite. The biases most commonly tested on Unit 2 MCQs are confirmation bias, hindsight bias, overconfidence, the framing effect, anchoring, the availability heuristic, and the representativeness heuristic.

Learn them as a set, not one at a time, because the distractors are usually other biases in the same family. A useful drill: for each bias, write down one sentence with the trigger (what the stem will describe) and one sentence with the tell (the giveaway phrase in the option). Anchoring's tell is a numeric estimate pulled toward an earlier reference point. Framing's tell is two equivalent descriptions of the same options producing different choices. Availability's tell is a probability judgment inflated by how easily an example comes to mind. Representativeness's tell is a likelihood judgment based on how well a case matches a prototype.

On test day, if the stem describes a choice or a judgment, expect a bias option to be correct and expect at least two other bias options as distractors.

Topic 2.8 psychometrics stems reward students who split reliability and validity before reading the choices, and they punish students who treat them as interchangeable.

Reliability is consistency of a score. Test-retest reliability is same test, same person, two administrations. Split-half reliability is correlation between two halves of a single administration. Inter-rater reliability is agreement between two scorers on the same responses.

Validity is accuracy of what the score means. Content validity is coverage of the domain the test claims to measure. Construct validity is the extent to which the test measures the underlying trait. Predictive validity is the correlation between the score and a real-world outcome (grades, job performance, treatment response) that the score is supposed to forecast.

On a psychometrics stem, decide first which one the stem describes (score stability = reliability; correlation with an outcome = validity), then name the subtype. This is one of the few Unit 2 topics where the correct option can usually be identified by process of elimination on structure alone.

I recommend a three-pass method.

• First, an untimed pass: no clock, no pressure, roughly 90 seconds per question. The goal is diagnostic. You want to see whether you can reason from the stimulus at all, and to flag any Unit 2 topics you keep missing.
• Second, a timed pass one week later: 24 minutes for 20 questions, which matches the AP exam pace of about 72 seconds per MCQ (90 minutes across 75 questions).
• Third, a review pass: check the answer key, then revise weak topics on Cognito's AP Psychology notes before you retry the missed questions a week after that.

The gap between passes is where the learning sticks. For general study habits across the whole course, see how to study for AP Psychology.