likelihood
The probability of obtaining certain data given that a particular proposition is true; in Bayesian inference these are not subjective in the way priors are.
As defined in AMA | March 2022
Where it comes up · 2
- AMA | March 20222 claims
- 320 - Constructing Self And World1 claim
Most important claims · 3 of 3Search all claims for “likelihood” →
- Bayes' theorem is a mathematical theorem you cannot disagree with given the axioms, but Bayesian epistemology is a broader claim that updating credences from priors is how we should gain knowledge; new insights, not just new empirical data, count as new data that can change your credences.AMA | March 2022
- Everyone is entitled to their own priors but no one is entitled to their own likelihoods; priors come from human brains embedded in the physical world that simply have opinions, and the hope of Bayesian epistemology is that enough data collection washes out differing priors.AMA | March 2022
- When sensory data is low-precision (e.g. a foggy day), the brain weights priors more heavily, so a higher-prior dog is perceived even if the raw data slightly favors a lion; but with high-precision clear data, the likelihood term overrules the prior.320 - Constructing Self And World