Interesting to think of this. I myself would assume that if thoughts are entirely physical (which I believe them to be), and if they are each a response to various stimuli (which I believe them to be); then there is a
possibility that thoughts (and their resulting actions) could be entirely predicted.
The problem with this, is that each stimulus may have a limited number of responses. But if each of those responses have an equal chance of occuring; then there is no way of predicting with 100% certainty an outcome from a certain stimulus. Only a general prediction can be made of the various responses.
We do this naturally each day when we anticipate the reactions of those around us. Think of a series of roads which all split off from one another. For example: when I gave my ex-girlfriend flowers (stimulus), I predicted her "emotions" (resonse) would take a road of general happiness. Now down this road are many other off-shooting roads (each represents happiness, but different specific reactions). If we can predict one of the roads (the main one), then there is no reason why we cannot predict the other, more specific, roads. That is to say, that some roads (responses) are more likely to occur than others, allowing for predictions. Our brains may not be able to do such predictions, but there is no reason that a higher "brain", such as a computer, could not.
I believe that with proper technology and know-how, it is possible to predict EVERY action of EVERY human being simply based on environmental stimuli (including that from other humans).
Trék
In considering the Origin of Species, it is quite conceivable that a naturalist... might come to the conclusion that each species had not been independently created, but had descended, like varieties, from other species. - Charles Darwin
On the Origin of Species
™ Version 4.2