How can an action be dependent on both its past and future?

Is it true that whenever an action takes place it is dependent on both its past and future? I mean if we already know that whatever we are doing is dependent on future as much as it is dependent on the past. If that is the case how can you not tell what I will be typing next?

-
What makes you think I can't tell what you will be typing next? –  WillO Sep 8 '12 at 12:50
If you formulate this question properly, by logical positivism, there is no problem. This is not a meaningful contradiction, your brain is only fooling you into thinking it is. –  Ron Maimon Sep 8 '12 at 15:49
Comment to the question(v1): If OP by the word action means $S=\int_{t_i}^{t_f}\!dt~L$, then it seems that OP's question is related to this Phys.SE post. –  Qmechanic Jan 24 '13 at 14:23

If you leave aside quantum mechanics, the laws of Physics are deterministic. That means if you have a perfect description of a system you can calculate the properties of the system arbitrarily far into it's past or future. So if you know the properties of the system at some time you can tell what the system was like in the past. However just because we are calculating past properties from future properties I wouldn't say that makes the properties of the system dependant on it's future.

Which is all very well, but many interesting systems are chaotic and since it's impossible to characterise even a classical system perfectly we can't calculate the systems properties for more than a short distance into the future.

And of course in quantum mechanics we have the collapse of the wavefunction, which means we can't calculate the results of future observations. The best we can do is assign them a probability.

-

Coexistence of Past and Future Measurements’ Effects... http://arxiv.org/ftp/arxiv/papers/1207/1207.0667.pdf

Can a Future Choice Affect a Past Measurement's Outcome? http://arxiv.org/ftp/arxiv/papers/1206/1206.6224.pdf

-