Schematic illustration of several potential biases involved in the inference problem. Applied to the abundance determination problem, the observable space represents (notably) the oscillation frequency sets, while the unobservable space would represent the model abundances. Case a shows an ideal scenario, where for given constraints (blue circle) our modelling (in grey) provides unbiased abundances (red circle). Panel b: possible expression of physical processes that are missing or are poorly taken into account in the model. It thus results in physical biases although it has ideal observables. Case c provides an example of degeneracy, that is, parameters that can vary widely because they are too weakly constrained by the observables. The last panel d is a typical illustration of the solar problem, where no mixture seems to satisfy the constraints. In this case, inconsistencies in the model physics or much weaker constraints than actually provided by the oscillation frequencies must be considered in order to provide a solution.