Decision making is an important aspect of our lives. Cognitive psychology is concerned with how people make decisions and what these processes reveal about the underlying cognitive processes. Decision making is extremely complex because there are so many options and dimensions to consider, and each decision has a major impact on our lives.
Decision making approaches:
- Consider pros and cons of each alternative. Attractive dimensions may compensate for unattractive dimensions
- Problems: labor intensive, memory intensive (probably only done for major life choices that have relatively few alternatives for them)
Elimination by aspects
- Decision is made by gradually eliminating less attractive choices. Seen in situations where one is overwhelmed by choices; used by committees, graduate school admissions, GPA, GRE scores, letters of rec
- Problem: people aren’t frequently overwhelmed by choices (four marriage proposals in one week), so people do eliminate by aspects, but probably not frequently
- Decision is made by multiplying the value of events by their probability of occurring
You can buy a ticket to participate in one of the following:
- 20% chance of winning 200 = EV 40
- 50% chance of winning 100 = EV 50
- 33% chance of winning 300 = EV 100
- 10% chance of winning 1000 = EV 100
- Which would you choose? People should break evenly with 3 and 4. Something about #2 makes it seem attractive for people for some reason. It does make very clear predictions about decision making allowing it to be tested. Usually results don’t support the theory.
- Used to calculate the probability of intelligent life existing elsewhere.
- Problems: difficult to estimate the probability of uncertain events, computationally complex, consistently fails to predict behavioral data
- People seem to like alternative 2 in the example (50% chance of winning 100) because it provides the greatest chance of winning something. But the expected value model doesn’t take this into account. Could it be revised to do so?
Subjected Expected Utility (SEU)
- Modified version of expected value model: subjective values replace actual values, subjective probabilities replace actual probabilities.
- So the 50% chance of winning would receive more weight than the 10% chance
- Problems: still difficult to estimate the probability of events, still computationally complex
- These four models all assume that people make rational decisions about the evidence at hand. Maybe this isn’t true.
Led to Nobel Prize in economics for Kahneman in early 2000′s. You have to be living in order to receive the prize and there is not a Nobel Prize in Psychology.
- People don’t think about probability or statistics when making decisions under uncertainty. Rules have never been learned in the first place or rules are known, but not applied (computationally complex).
- Instead people rely on simple mental shortcuts, called a heuristic. These heuristics lead to predictable biases in the decision making process.
Review several of these heuristics, and the biases they cause.
- Making judgments about probability: (It rained yesterday. What is the probability that it will rain today? What is the probability that someone is republican?)
- Heuristic: Does A resemble B?
- Representativeness is not influenced by several factors that should affect judgments of probability: insensitivity to prior probability of outcomes (people fail to take base rates into consideration).
Insensitivity to Sample Size
- Small samples contain more extreme values than large samples. Called the law of large numbers. As sample size increases, its mean will move closer to the population from which it is drawn (called the central limits theorem).
- Picture below illustrates the central limits theorem.
- Most people aren’t formally taught the principles of probability….and if you learn them in college, the Heuristic of Representativeness is already deeply ingrained. Even research psychologists show the same biases as those without training in probability.
- Local representativeness is the belief that a sequence should retain its essential characteristics, even when it is short. Leads to gambler’s fallacy (viewing chance as a self-correcting process).
- People are very bad at generating strings of random numbers. They avoid runs (the same thing happening more than twice in a row) where the same number appears consecutively. 39451777723. Runs are common even in short random sequences.
In image above, the output of flipping a coin and landing on either blue or red are shown. You can see how at the beginning the proportion of blue to red is very extreme, but it balances out as sample size increases.