Abstract
We investigate the robustness of behavioural type classifications across related public goods games.
We study four games in which contributions to the public good have different, nonlinear opportunity costs in private income.
The classification of a player by their behaviour in the baseline linear case predicts their behaviour in other games.
In particular, strong conditional cooperators, who match contributions one-for-one when payoffs are linear, do so in nonlinear games only when such matching improves social welfare.
This mode of conditional cooperation therefore is underpinned by a sophisticated understanding and assessment of the financial incentives presented by the game.