Seminar 4 Cognitive biases

In this seminar, we further discussed the lecture content, especially around the identifiable victim effect. We also discussed why you estimated, on average, to be more trustworthy than 66% of St Mary’s psychology students - three explanations might be worth keeping considering:

  • We overestimate our positive attributes as that makes us feel good (the general Optimism Bias).
  • We have an excessively pessimistic perception of others, having been trained to accept a negative view of humans (the Humankind argument).
  • People are most prone to overestimate their relative performance/skill when they are bad at something - the so called Dunning-Kruger effect.

We then discussed the example of Linda - see slides - and I highlighted the importance of avoiding the Conjunction Fallacy, i.e. of recognising that a more specific criterion is always less likely to apply.

Then we talked about Covid mass-testing ideas to explore the importance of considering base rates. Wikipedia has a clear example of that here. Given that there are many more accountants than investigative journalists, this is important to the Linda example as well.

If you have time for some further reading, have a look at the paper by Schleifer below - it discusses the Linda example further (split into 2, once called Steve with regard to base rates, and then Linda only with regard to the conjunction).

4.1 Seminar materials

The seminar slides are here - but they are not self-explanatory, if you missed the seminar make sure to also read the above.