Professional Documents
Culture Documents
Risk 3
Risk 3
HEURISTICS VS ALGORITHMS
Algorithm: A formalized set of steps or transformations intended to solve a specific problem. Algorthms are
correct by definition and if followed properly they should always lead to the correct answer.
A mathematical equation is an algorithm
Medical diagnosis based on symptoms and test results is an algorithm
Rigorous deductive reasoning is an algorithm
HEURISTICS VS ALGORITHMS
Heuristics are informal, ‚natural’ and quick forms of reasoning which often lead to correct conclusions, but may
not always be correct. They often focus on a single aspect of the problem, rather than a complex analysis
With old CRT TVs, hitting the TV often fixed issues. However the cause was unknown, and the TV may have broken more.
This is a heuristic-based action
So is resetting your router or computer when you lose connection
When selecting a beverage in a shop, choosing the one in the prettiest bottle is a heuristic-based action
Choosing IT as a field of study because „I heard IT sepcialists earn well” is based on heuristics (unless you actually
investigate earnings and career development of IT alumni, in which case you propbably use an algorithm)
HEURISTICS VS ALGORITHMS
The rule of maximizing (von Neuman & Morgenstern, 1944) assumed that we as decision makers choose options
which maximize (optimize) our gains, providing the highest benefits over losses – utility maximization based on
EUT/EVT models
Simon (1956) stated that maximization is impossible in the natural environment, as information processing
capabilities are limited, and access to a complete set of data is a rare occurence
Hence instead of maximizing, our goal is to „satisfice”
SATISFICING
If the agent has chosen an acceptable option, other options may appear
The agent will try to switch from the worse option to the better, leading towards maximization without assuming it as a
deliberate goal
Therefore satisficing explains both settling for the bare minimum, acting under an incomplete set of choices or information,
and still accounts for utility maximization as the ultimate goal under optimal conditions
THE CASE OF TOO MUCH CHOICE
In EUT/EVT maximizing, the complete set of choices is the best possible set
A rational agent should analyze all options and then choose the best
It is unthinkable that fewer options would be better (or is it?)
TOO MUCH CHOICE
TOO MUCH CHOICE
Iyengar & Lepper (1999, 2000) researched the willingness to buy a gourmet food item (exotic jam or chocolate),
selected from either:
6 items, or
30 items
Benson (2016) categorize 188 biases and heuristics into the Codex. We shall try to learn about a few of these.
AVAILABILITY HEURISTIC (BETTMAN, 1979; TVERSKY & KAHNEMAN,
1973)
Availability is the ease of access to a mental representation of an event (an outcome, a choice…)
How easily can you remember a situation?
How many such events can you easily recall?
The availability heuristic is basing one’s estimates that an event will/will not occur based on how mentally available
it is
What are the chances of being a victim of a terrorist attack? Are they higher than the chances of your spouse
killing you?
Is homicide more common than suicide?
Is it more likely to get hurt in a car accident or a plane crash?
AVAILABILITY HEURISTIC
Representativeness means a subjective fit of a situation (person, outcome, event, choice…) to a clear category
Is John a typical hooligan?
Is a company a typical scam?
The representativeness heuristic is usually employed to tasks when one has to judge the probability that an object
X belongs to class Y. (or an event X belongs to a class of events Y). The more typical the object or event seems,
the more probable it seems for us that it belongs to the given class.
It has significant consequences when the class in question is perceived as risky, dangerous, or opportune.
REPRESENTATIVENESS HEURISTIC
1 2 3
REPRESENTATIVENESS HEURISTIC
James Holmes, mass murderer Eric Calderon, MMA Fighter Ice Cube, rapper
REPRESENTATIVENESS: THE GAMBLER’S FALLACY
You are playing coin toss. The last four times it turned out Heads. Would you
A) bet on Heads
B) bet on Tails
GAMBLER’S FALLACY
If you were to play a lottery, which of these numbers would you prefer on your ticket?
A) 1, 6, 15, 22, 31, 45
B) 1, 2, 3, 4, 5, 6
GAMBLER’S FALLACY AND OBJECTIVE PROBABILITY
1
In a coin toss, the chance that Heads will turn out four out of four times is ( )4, which is 1 in 16.
2
The chance that it will now turn out Heads for the fifth time is 1 in 32
It is therefore correct to choose Tails, as the remaining chance is 31 in 32, right?
GAMBLER’S FALLACY AND OBJECTIVE PROBABILITY
Incorrect.
We tend to compare the chance that a coin toss yields Heads five times with „the rest”, ignoring the fact that four tosses
already happened and have no impact on the subsequent toss.
We need to compare the chance that Heads comes out 5 times in a row with the chance that Heads comes out 4 times in
a row and then Tails comes out. This chance is exactly 1 in 32
HOT HAND FALLACY
The hot hand fallacy is visible in casinos, where a gambler who won several times in a row assumes to have a ‚hot
hand’ or ‚luck on their side’, urging them to gamble more
Until the luck inevitably reverses
ANCHORING BIAS (TVERSKY & KAHNEMAN, 1974)
When a relevant value is available, we tend to „anchor” our predictions of probability (price, other numbers) to
this value. It is subsequently difficult to move away from the value.
Is the probability of getting hurt in a car accident more or less than 10%? How much is it exactly?
Do investments in stocks yield more than 5% profit annually? How much is it?
ANCHORING BIAS
Is this car worth more than $300 000? How much is it?
ANCHORING BIAS
In an auction (or a contingent valuation bidding game), the starting point (e.g. entry bid) acts as an anchor for
subsequent bids