[ad_1]
Heuristics are cognitive strategies used to make decisions when faced with data overload. However, they can lead to biases, such as overassigning probabilities to conjunctions and anchoring. Bayes’ rule can help make unbiased predictions, but it can be difficult to apply in everyday contexts.
Heuristics are “rules of thumb,” cognitive strategies people use to make selections when faced with data overload. For example, an employer might use the “long hair means the person is a flake” heuristic when making hiring decisions. As with the previous example, heuristics don’t always work effectively. Some heuristics lead to systematic errors that can be isolated experimentally, so they are labeled as biases.
The most common and illustrative example of a systematic bias is that of overassigning probabilities to conjunctions. Try the following:
Linda is 31 years old, single, outspoken and very bright. She majored in philosophy in college. As a student, she was deeply concerned about discrimination and other social issues and she participated in anti-nuclear demonstrations. Which statement is more likely?
a. Linda is a bank teller.
b. Linda is a bank teller and active in the feminist movement.
When combined with other options for throwing off the candidate, most people actually choose b, even though the probability of b (a conjunction) is definitely lower than the probability of a, which is a superset of b. But our mind automatically works this way. Various heuristics and biases appear to be built into the way our human mind works.
Try another:
Estimate the product of the series:
9x8x7x6x5x4x3x2x1 = ?
vs.
1x2x3x4x5x6x7x8x9 = ?
Experimental studies have confirmed that the estimates are heavily biased towards the first series. In one study that required participants to give their answers within five seconds, the mean estimate for the first set was 4,200 and for the second, only 500. The true answer is 40,000. Everyone has radically underestimated the real answer.
This bias is called anchoring: fixating on what comes first and undermatching as more data arrives. In a sales context, salespeople often show a customer a more expensive product, then gradually scale down. This makes all products seem cheaper and is a very effective sales strategy that exploits universal human heuristics and biases.
Bayes’ rule has often been cited as a way to make predictions mathematically and normatively, freeing decision makers from the threat of biased decisions. Unfortunately, applying Bayes’ rule in everyday contexts can be difficult for those who are not explicitly trained to do so.
[ad_2]