Skip to content
AP Psychology · Cognition

Cognitive biases

Cognitive biases appears in 13 reviewed AP Psychology questions and sits in Cognition. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

Cognitive biases are the family of predictable thinking errors the exam asks you to tell apart. The most tested ones include hindsight bias, the sunk cost fallacy, overconfidence, confirmation bias, framing and anchoring.

Why it matters

Each bias has its own trigger, so naming the right one shows you understand the cause. Hindsight bias rewrites your memory of past predictions after you learn the outcome. The sunk cost fallacy lets money or time already lost drive future choices. Overconfidence means being more sure of your judgments than your accuracy justifies. In studies, people who say they are 90% sure are often right only about 70-75% of the time.

How the exam tests it

Stems quote a person, such as "I knew it all along" (hindsight) or "I've invested too much to quit" (sunk cost). Overconfidence items may ask what the finding says about metacognition: our sense of knowing is poorly matched to what we actually know.

Don’t confuse it with

Students mix up hindsight bias and confirmation bias, but hindsight bias is about feeling an outcome was predictable after the fact, while confirmation bias is about favoring evidence for a belief you already hold.

Cognitive biases in real questions

Example 1Easier
After the results of a political election are announced, many voters claim they 'knew all along' who would win, even though they did not make such a prediction before the election. This cognitive bias is best described as:
  1. AOverconfidence bias, because voters are expressing excessive certainty about their retrospective knowledge of the outcome
  2. BAvailability heuristic, because the winning outcome is more available in memory after the election, making it seem predictable in retrospect
  3. CConfirmation bias, because voters are selectively remembering evidence that supported the winner before the election
  4. DHindsight bias, because after learning the outcome, people revise their memories of their prior predictions to make them seem more accurate than they wereCorrect

Why D is right

Hindsight bias (the 'I knew it all along' effect) is the tendency to perceive past events as having been more predictable than they actually were. After learning an outcome, people reconstruct their prior beliefs to align with what happened, making them feel (often incorrectly) that they had predicted the outcome. This is a form of reconstructive memory distortion — the outcome knowledge retrospectively changes how people remember their prior beliefs.
Why the other 3 choices miss
A
Incorrect because overconfidence bias refers to excessive certainty about current knowledge or predictions (e.g., being 90% confident when you are only 70% accurate). Hindsight bias specifically involves the retrospective claim of prior knowledge after learning the outcome, not general overconfidence in current knowledge.
B
Incorrect because the availability heuristic involves judging likelihood based on ease of retrieval. Hindsight bias involves restructuring past beliefs to match known outcomes, not judging probability based on memory retrieval ease.
C
Incorrect because confirmation bias involves selectively seeking or interpreting information to support existing beliefs before or during a decision. Hindsight bias specifically involves the retrospective revision of prior beliefs after learning an outcome — it is a post-hoc distortion, not a prospective attention bias.
Example 2Easier
An investor has lost $50,000 in a struggling business venture. Despite all analysis suggesting the business will continue to fail, the investor continues to pour additional money into it, saying 'I've already invested so much, I can't stop now.' This reasoning represents which cognitive error?
  1. AFraming effect, because the investor is framing continued investment as avoiding a loss rather than as accepting a definite loss
  2. BAvailability heuristic, because memorable stories of business turnarounds make the investor overestimate the probability of recovery
  3. CSunk cost fallacy, because the investor is allowing past, irrecoverable losses to irrationally influence future investment decisionsCorrect
  4. DConfirmation bias, because the investor is selectively attending to positive business indicators while ignoring evidence of failure

Why C is right

The sunk cost fallacy involves allowing past, irrecoverable costs (money, time, effort already spent) to influence future decisions. Rational decision-making should consider only future costs and benefits — the $50,000 already lost cannot be recovered regardless of future decisions. Continuing to invest to 'recover' the sunk cost is irrational if the expected future return doesn't justify the new investment. The sunk cost fallacy is driven by loss aversion and commitment psychology.
Why the other 3 choices miss
A
Incorrect because the framing effect involves the same information presented as gains vs. losses affecting decisions. While loss framing may contribute to the investor's psychology, the specific cognitive error of being influenced by irrecoverable past investment is the sunk cost fallacy, not the framing effect.
B
Incorrect because the availability heuristic involves judging likelihood based on ease of recall (e.g., memorable turnaround stories). While this might also influence the investor, the explicit reasoning given ('I've invested so much') directly describes the sunk cost fallacy, not availability-based probability estimation.
D
Incorrect because confirmation bias involves selectively seeking confirming evidence. While the investor may also exhibit confirmation bias, the specific error stated ('I've already invested so much, I can't stop now') is explicitly about past investment influencing current decisions — the definition of the sunk cost fallacy.
Example 3Medium
Research on overconfidence bias has shown that when people rate their confidence in their answers as being 'about 90% sure,' their actual accuracy rate is typically closer to 70-75%. What does this finding most directly imply about metacognition?
  1. AConfidence and accuracy are completely uncorrelated — feeling certain about an answer provides no information about whether the answer is actually correct
  2. BPeople consistently underestimate their own cognitive abilities and should trust their knowledge more, because actual accuracy is typically higher than perceived accuracy
  3. COverconfidence only occurs on expert-level knowledge questions, because novices appropriately feel uncertain about difficult material
  4. DPeople's subjective sense of knowing is often imperfectly calibrated with their actual knowledge, leading to systematic overconfidence in the accuracy of their beliefsCorrect

Why D is right

Overconfidence bias demonstrates that people's metacognitive assessments of their own knowledge (their confidence in specific beliefs) are systematically inaccurate in the direction of overconfidence. When someone is 90% confident, they are often only 70-75% accurate, meaning their subjective certainty exceeds their actual accuracy. This has important implications for eyewitness testimony (confident but inaccurate witnesses), medical diagnosis, and decision-making in high-stakes contexts.
Why the other 3 choices miss
A
Incorrect because confidence and accuracy are correlated (higher confidence does predict somewhat higher accuracy), but the correlation is imperfect. Overconfidence means confidence systematically exceeds accuracy, not that they are completely uncorrelated. Extremely high confidence is somewhat diagnostic but consistently overestimates accuracy.
B
Incorrect because overconfidence bias is a systematic overestimation of accuracy, not underestimation. The finding that confidence (90%) exceeds accuracy (70-75%) shows that people are overconfident, not underconfident. Trusting their knowledge more would worsen calibration.
C
Incorrect because overconfidence bias is actually more pronounced among non-experts than experts in many domains. Dunning-Kruger research suggests that those with limited knowledge are often the most overconfident, while true experts may even show slight underconfidence (calibrated humility). Overconfidence is not limited to expert-level questions.

Studied alongside

Other concepts from Cognition.