This is the method of tactical thinking I was taught in fencing coaches college, augmented with some other things I've learned since. It divides an encounter into beginning and end, with each being foreseen or unforeseen. In order for something to be foreseen, you should have at least two of three pieces of knowledge: when, what/how, where. Sometimes the divisions get a little murky if you only have two pieces, and only rarely could count as foreseen if you only have one. So, three possibilities:
Foreseen beginning - Foreseen end
This category includes all first and second intention actions. For example, a first intention attack with a jump-lunge that successfully lands. You begin the action, thus know when/what/where, and it goes according to plan. A second intention action is one in which the first action is intended to fail, but the followup is known. Example: you attack with a lunge, knowing the opponent will parry strongly with a retreat and hold it for too long, and then redouble and remise with a disengage.
Attacks in response to an opening could be placed in the first intention category, but the beginning knowledge may be incomplete. For example, you know that at some point your opponent will relax and lower their guard, opening an arm target, but you do not know when. You have what and where, but not when. Or, you know that after two quick steps, your opponent will prepare and leave either a high or low opening. You know when (the two quick steps giving ample warning), you know what (you're going to attack the preparation), but you don't know where you are attacking.
1st and 2nd intention actions can be some of the fastest, executed at full speed with no need to make decisions. For that reason, the glitches and OODA loop interruptions are most helpful here. However, they are obviously dependent on knowledge of your opponents behavior. This can be gleaned from past experience, either previous encounters or touches, or observation of reflex reactions or repeated arm movements. Or it can be formed by forcing shaping the opponent's behavior. One of my favorite three touch patterns comes from a combination of experience and observation. I observe that my opponent is not fully aware and active- maybe it's his first bout of the day. As soon as the referee starts the action, I immediately launch a fast fleche action. If it scores, my opponent will be slightly shaken, and trying to wake up. The next touch will be a fleche that is even faster. The opponent was prepared for the speed of the first one- it seemed like the fast thing I could possibly do, and the extra speed is surprising. Now he's going to overly aware to the point of being twitchy. The third touch is going to be a stutter, which will immediately draw a panicked parry, then a full fleche as they relax.
Unforeseen beginning - Foreseen end
This category includes a lot of counter-attacks and "whatever happens, I'm gonna do this" kind of things. A lot of reality-based martial arts go for these kind of actions, where they try to simplify down to a couple of a responses that handle most things, so when you're surprised, you just go. It can be very effective, and definitely useful to have some of these prepared. In bouting, a lot of opponent initiated actions go this way, where the initial action fails/turns into a struggle, and one side eventual manages to transform it into their favorite action. I try to put one of these in my mental rack, which probably I'll write about another time.
Foreseen beginning - Unforeseen end
This category is referred to as "open eyes". You start the action, watch the response, and then deal with it. This is one of the most trainable types of tactical thinking. The methodology goes like this: teach an opening, say advance and press the opponents blade. Then teach individual responses: you try to press the blade, the opponent deceives and you don't find it, you close the line the other way and finish. You try to press the blade, find it, finish. You try to press the blade, the opponent presses back, disengage and finish. Then you give random responses and have the student respond appropriately. It obviously works best if the potential responses are limited and mappable, and when you can force the timing, leaving only what and where as the unknowns. I try to always add an additional "if you get a nonsense response, or something you don't understand, break distance and start over" for my fencers, just to prevent locking up or forcing an inappropriate action.
Unforeseen beginning - Unforeseen end
The worst and sometimes the best. The worst when you don't have a plan, you're trying to understand what's happening, and deciding how to respond in the moment. Basically disastrous, and where a lot of thinking beginners live and die. Purely instinctive fighters fair better, as they do whatever they feel, which can be reckless, but, as it cuts out the conscious mind, done very, very fast. And the best, in the zen no-mind state, where your training is good and flows out all on its own. Flow state. Experience, training, mindset.