Specifically it is erroneous to consider that choices are evaluated and determined solely at the point of action, but rather, default states may well be set within the brain based on our training and indoctrination. It is these default states that represent the possibility of choices that we can base a decision on.
The shorter the time interval to make a decision, the more established the choice must be to be considered as a default action. This is one reason why training becomes so important in ensuring that proper and valid choices are used in circumstances that may require split-second decisions.
There is a strong argument that our choices are governed by rationalizing responses to our pain/pleasure centers in the brain. In the end, it doesn't really matter what motivations or rationalizations occur, but whatever is determined to be the most desirable or appealing aspects of a choice will tend to be those that are the most likely to be used.
Some may argue that there are far too many events for which we have no specific knowledge or experience for this to be a plausible explanation, however careful examination will show that in the absence of any particular response the default of "fight, flight, or freeze" will invariably occur. Any other response must originate from some cause, therefore we must consider that the "cause" has already been processed and accepted into the brain as a possible solution at some previous point in time. Anything else would require that we accept an indeterministic or random solution having originated, against which we have absolutely no evidence that such a circumstance has ever, or would ever occur.
Similarly when we consider acts that we consider morally reprehensible, it is inconceivable that such acts originated without prior thought or consideration. In fact, there is strong evidence to suggest that before such crimes occur, there is a considerable amount of thought and time invested in reviewing the particulars of such a choice. This lends credence to the idea that such a choice is used to "grant permission" for use as a decision, and that such a choice also carries with it the moral responsibility for considering it as a possibility.
Once a decision has been made, then other choices may no longer be relevant since this can send the individual down an irrevocable path that cannot be undone. This is often the case with addictions where the early decisions may be made thinking that there will be no consequences, however the choice must be considered as an acceptable option for this to be an issue. When someone excludes a choice from possible use (either morally or through training), then it is exceedingly unlikely that it would ever manifest in an individual's actions.
This also lends support to the role of self-indulgent behaviors, especially those that reinforce anger and even "mob" behaviors are created because we effectively create an environment in which radical responses are considered acceptable. For example, if we indulge in unconstrained anger as a response, we may find that it escalates to violence and perhaps killing. While it was never intended that an individual be killed, the absence of a constraint against anger effectively allows the brain to escalate it to whatever degree is needed for the circumstance since it is considered a viable solution. In the mob, the individual decisions may be superceded by the consensus of a group "choice" so that it lends justification to the action being perpetrated. If an individual's choices have not been well-formed or considered, it is entirely possible that they will be subsumed to the authority of the mob.
It is precisely behaviors with unintended consequences that lend credence to the idea that once a choice is in progress it simply isn't possible to intervene without alternative choices ready. To do so would require considerable training and a range of alternative choices, so while it may be possible for a martial artist to "pull a punch", it isn't likely with an untrained fighter responding with adrenaline or panic. This is also a reason why people in these circumstances invariably respond with statements like, "I don't know, I just didn't think".
Invariably when a decision to act is made that results in something we consider negative or detrimental, the argument that assigns moral responsibility is based on the presumption that one of the available choices was to preclude that particular action. As we indicated before, that there must be a cause to perform that action, one is forced to conclude that to retain an undesirable option as a choice requires a decision wherein the actor has concluded that it (a) isn't a bad thing, or (b) they don't care that it is a bad thing. The only exemption from this reasoning is if they truly didn't know that it was bad, in which case they couldn't be held morally responsible for the choice. This is also why even our legal system attempts to determine the "motive" for a crime, since it represents the sequence of events and possible rationalization that could explain what happened and how/why it was chosen. We are suspicious of crimes without motive, because that suggests either psychopathic behavior, or a falsely accused individual.
While we may not have free will in the sense of dynamically being able to modify our behavior, there is sufficient reason to believe that we are morally culpable in the thoughts we process and accept, long before any action is decided upon. As a result, there is some merit to the notion that a crime or action must be committed in one's mind before it can ever manifest in the real world.