Explanation of the normative model of reasoning
It would probably be a good idea to briefly explain the term 'normative model of reasoning' 22.214.171.124 08:16, 24 April 2022 (UTC)
- Adding to that: how do we know what the correct normative model is? Bayes vs. Frequentist for example is still an open debate for many 126.96.36.199 08:18, 24 April 2022 (UTC)
The content of the article at the moment is very good. I think it could be presented more clearly and succinctly.
I would propose the following structure for the article:
- introduction sentences: In principle just as they are. Maybe they could be phrased in a way that is easier to understand? E.g. something like "Cognitive biases are a systematic deviation from rational thinking, often leading to errors in judgement as information and evidence are interpreted in an inappropriate way." as a first sentence. (I'm not entirely sure, maybe the term "normative model of reasoning" should have it's own article and maybe it even makes the article more readable to omit it entirely?).
- Section: Causes I like the part about people using faulty heuristics. I think it's also good to mention that people may choose different cognitive models and that some of them are just wrong. But personally I would omit the Frequentist vs. Bayesian debate (or make it very clear what we mean) and also the part where people accuse each other of different biases. If we want to have it in the article it should probably get its own section.
- Section: Examples of cognitive biases could list the examples here
- Section" Effects of cognitive biases (or maybe just effects and examples above). First paragraph of the section could state that biases do two things: they increase bias and they increase noise (what's the relationship between noise and variance exactly?). Second paragraph could explain that being biased doesn't mean that one is wrong (probably good to have one sentence in the opening paragraph right at the top to that effect and explain in. more detail here). Third paragraph could talk about what happens if you have not only a single forecaster, but you combine different forecasts (if you wanted to include that). I think the bias-variance-tradeoff would need to be explained in more detail if we want it in here