MeJennifer said:
I think there is a definite meaning for to go backwards in time.
Going backwards in time means going into the past while traveling forwards in time.
Let p be an event in spacetime, Event q is in the (chronological) past of p if there exists a future-directed timelilke curve from q to p.
Suppose that event p is on the worldline of an observer, and that there is an event q is in the past of p such that a future-directed timelike curve from p to q. Then, it is possible for an observer to travel into his own past.
Joining the future-directed timelike curve form p to to q with the future-directed timelike curve from q to p, shows that, as already mentioned by pervect, this is completely equivalent to the existence of a closed timelike curve.
Another question of course is if that is at all possible.
Its certainly allowed by general relativity, as there are numerous solutions to Einsten's equations that have closed timelike curves.
How does one deal with the paradoxes associated with time travel? Also as mentioned (in another thread), Matt Visser has written http://arxiv.org/abs/gr-qc/0204022" about this. He talks about four possibilies:
1. Radically rerwite physics from the ground up;
2. Permit time travel, but also invoke consistency constraints;
3. Quantum physics intervenes to prevent time travel;
4. the Boring Physics Conjecture, where we assume (until forced not) that our particular universe is globally hyperbolic, and thus doesn't have closed timelike curves.
In the past 4. was often assumed, but since global hyperbolicity is a very strong global condition and Einstein's equations are (local) differential equations, many physicists have moved to 2. and 3. Stephen Hawking likes 3., for example, and has formulated the Chronology Protection Conjecture, "It seems that there is a Chronology Protection Agency which prevents the appearance of closed timelike curves and so makes the universe safe for historians."
This roughly states that near a chronology horizon (horizon at which spacetime becomes causally ill-behaved), expectation values of stress-energy tensors for quantum fields blow up, thus preventing (by wall-of-fire barriers) physical objects from crossing chronology horizons. There seems to be some semi-classical evidence for this conjecture, but a http://arxiv.org/abs/gr-qc/9603012" by Kay, Radzikowski, and Wald muddies the picture a bit. Their analysis shows that the semi-classical stress-energy tensor is ill-defined, but not necessarily infinite, at a chronology horizon.
This may be just an indication that the semi-classical theory breaks down at chronology horizons, and that full quantum gravity is needed for definitive predictions.