Fig 1.
Schematic illustration of Crollen et al.’s ([4], p. 41) bi-directional mapping processes (in gray arrows): Panel (A) shows the assumed estimation process in a perception task requiring estimation from the logarithmically compressed representation of non-symbolic numerosity onto the linearly spaced representation of symbolic number magnitude which leads to underestimation. Panel (B) depicts the estimation process in the production task, in which linearly spaced symbolic number magnitude representations have to be mapped onto the compressed representation of non-symbolic numerosity, hence, leading to overestimation.
Fig 2.
Expected error patterns in the three different types of estimation tasks.
We hypothesized that error patterns of numerosity estimation and unbounded number line estimation should be identical, whereas the error pattern for bounded number line estimation should be reversed.
Fig 3.
Schematic illustration of an example of the (A) unbounded and (B) bounded number line estimation. The production tasks are displayed in the upper row, the perception tasks in the lower row.
Fig 4.
Estimation patterns (mean estimates across all participants, left charts) and estimation error variability (SD of REE, right charts) for numerosity estimation (Panels A + B), unbounded number line estimation (Panels C + D) and bounded number line estimation (Panels E + F).
Fig 5.
Mean percent REE for the three magnitude estimation tasks separated for perception and production task type.
Error bars reflect 1 Standard Error of the Mean (SEM).