Inherent uncertainty therefore must be scaled into a domain-general unit (that could then facilitate both comparison and common reporting biases). But this seems to be something that emerges over development.
Next up is to understand how and why that happens!
For fellow dev MC researchers: some helpful correlational data + modelling with adult confidence reports to justify using RT as an implicit conf measure.
But note that while this corr is generally true in kids too, it's not perfect (West et al 2025; Leckey et al 2025)
Second, an experimental design. Participants compared which of two answers had higher confidence, allowing us to test whether the *unit* of confidence was the same across memory and perception.