A high spatial resolution in PET can be achieved by using small detector elements. To maintain good sensitivity these elements have to be quite long, thus introducing parallax error and making the spatial resolution non-uniform over the image volume. Uniformity of spatial resolution can be improved by utilizing depth-of-interaction (DOI) information to reduce the parallax error. In the present study we have focused on phoswich approaches based on interacting scintillators, that is, a phoswich combination in which one scintillator emits light in the excitation band of the other. We have looked at LaBr3:Ce and LaCl3:Ce and the interactions of those two scintillators with LSO:Ce, GSO:Ce and YSO:Ce. The reasons to use the two Lanthanum scintillators are twofold: light output is high and the two different emission wavelengths, 350 nm (LaCl3:Ce) and 380 nm (LaBr3:Ce) may produce different interactions with the three oxyorthosilicate scintillators. In addition a possible DOI detector comprising LuAG:Pr pixels with a thin LSO:Ce layer at one end has been evaluated. A Bollinger-Thomas set-up was used to measure luminescence rise and luminescence decay time characteristics in all cases. When using LaCl3:Ce, the phoswich combinations with YSO:Ce and GSO:Ce showed phoswich decay time characteristics as expected for a simple convolution of the decay times of the two phoswich components. A correction was needed, however, for the LaCl3:Ce-LSO:Ce phoswich due to the LSO:Ce intrinsic activity. For the LaBr3:Ce-LSO:Ce phoswich, corrections were needed for noninteracting LaBr3:Ce light in addition to the expected phoswich interaction.