Lecture 5 Flashcards
Hindsight Bias
- Current bias towards previous probability and statistical judgements about past events
- Bias based on knowledge of actual outcomes
hindsight vs Foresight
Hindsight: Event has happened, Outcome is known
Foresight: Event has not happened, Outcome is not known
Judgements based on hindsight also known as:
Postdictive judgements
Judgements based on foresight also known as:
predictive judgements
Creeping determinism
“the relative view that an event is inevitable”
Another word for hindsight bias
creeping because this type of determinism is relative
determinism vs creeping determinism
Determinism:
* Perceived inevitability of
outcome is absolute
* recall of past prediction that
outcome will occur in terms
of a probability = 1
Creeping determinism:
- Perceived inevitability of
outcome is relative
* recall of past prediction that
outcome will occur in terms
of a probability: NOT 1.0
Judgements in hindsight produce:
Provide Higher likelihood for
actual outcome
Lower likelihood for alternative
outcomes
Memory Paradigm
*Estimate probability of possible outcomes for
a future event
*Event occurs
*Recall the probability that was estimated for
each possible outcome
Question:
Is there evidence for hindsight bias (aka
creeping determinism) [Fischoff & Blyth, 1975]
*There is evidence for hindsight bias
*Recalled probabilities were biased or
“creeping” toward the probability of the
actual outcome.
Selective Activation, Reconstruction and Anchoring (SARA)
- Selective activation of relevant information
- Outcome for the original prediction is also activated
- Reconstruction of original prediction process using activated
information - “anchored” toward probability of the known outcome
Probability Cue
A cue that might use based on how likely the
cue will lead to a correct or optimal decision
Cue Value
The value associated with each item being
compared based on the criteria set by the
probability cue
Take-The-Best
A prescriptive strategy
* Search for the probability cue that can differentiate between
the two items that are compared
* Start search based on how likely the cue can lead to a correct
decision
* Probability cue is found choose the item with the higher cue
value
Reconstruction After Feedback with Take-The-Best (RAFT)
- Apply take-the-best strategy
- Estimate the probability that the probability cue will lead to
the optimal decision - Consider feedback based on outcome of prediction
- Include feedback and re-estimate the probability that the
probability cue will lead to the optima decision