Dr Maxine Sherman
Subjective confidence, metacognition and mentalising
My research concentrates mostly on subjective decision confidence and metacognition – broadly, the ability to accurately reflect on one's own mental states. Studying confidence and metacognition for very simple decisions such as perceptual decisions (“is that moving left or right”) can be very useful for determining the computations underlying the construction of confidence judgements: stimuli are easy to control, and perceptual decisions are relatively easy to model.
However, we are also able to reflect on more decisions that are far more abstract. For example, we can evaluate our confidence in decisions about others’ mental states (i.e. confidence in mentalising). Successful interactions with others requires us to have at least some understanding of their beliefs and knowledge, or lack thereof ("I changed my plans and am no longer free for dinner; I should let my housemate know."), but how confidence is constructed for these kinds of inferences – inferences about what others are thinking - is far less studied.
Projects could centre around perceptual metacognition or metacognition for mentalising, or they could compare the two across individuals. There is scope to acquire skills in machine learning, computational modelling, visual psychophysics, and/or analysis of neuroimaging data (EEG, fMRI). Projects can be designed around your interests, but some possible research questions include:
• Are confidence ratings in perceptual tasks “Bayesian”? Do they incorporate expectations in a statistically optimal way?
• Can we predict perceptual confidence ratings from visual cortex activations alone, without any “higher” brain areas?
• How do we develop general beliefs about our ability (“I’m a good driver”), and to what extent are they task specific?
• How good are we at inferring the beliefs of others? And do we know how good we are (do we have good metacognition for mentalising)?
• To what extent does successful collaboration (e.g. in collaborative economic games) depend on having good metacognition for mentalising? Do we need to be able to appropriately evaluate the validity of our beleifs about our partner?