We investigate the gender specific intertemporal allocation behavior of spouses with different deterministic life expectations in an experiment where the gen- der of one's partner is known. In each period of their life both partners propose a consumption level one of which is then randomly implemented. To allow for learning one experiences many "lives". Participants achieve a rather high degree of optimality that does not change over time. Indepen- dent of the own gender a participant is nicer to women and acts more selfishly if the partner is a man. Participants are not aware of their discriminating behavior.