Proxy data (e.g. lacustrine) can be used to generate proxy temperature and precip. data for past periods, and they can conceivably be compared with historical simulations by climate models for validation purposes, especially for times before ~1950, i.e., before good observational networks were established to provide more direct validation. Those lacustrine records, however, don't have much to say about future climate. Any validation would be over historical times, and if the models perform okay for past climate, then confidence should be increased in their future projections.
See Article PRYSM v2.0 : A Proxy System Model for Lacustrine Archives: P...
for info. on how to calibrate your lacustrine data. It looks like there are lots of tricky subtleties involved!