Getting rid of bias in decision making
- Datum: 04.07.2022
- Uhrzeit: 14:00 - 15:00
- Vortragende(r): Chris Donkin
- LMU Munich Computational Modelling in Psychology
- Ort: Zoom
- Gastgeber: Eric Schulz (Mirko Thalmann & Alexander Kipnis)
Even in simple perceptual decision-making tasks, it is easy to provoke participants into biased responding. For example, if one of two responses should be made more often, people tend to over-adjust, and end up 'biased'. At the start of this project, we set out in search of general mechanistic accounts of bias, things like: people accumulate a combination of bias and stimulus-based evidence weighted by reliability, or people have prior biases and they interact with internal timeout signals. However, it became clear over a series of experiments that bias is an epiphenomenon, and so we should expect no general theory of bias itself. Rather, any bias we do observe simply reflects the difference between what experimenters and participants think, based on their respective understanding of the situation, is the right thing to do.