First a caveat- this isn't encyclopedic. There are lots of ways to train: from the kihon-kata-kumite paradigm; basics versus fundamentals; technique-based or principles-based. Levels of resistance. Different types of randori. Attributes training. What I'm writing about here are things that can be build-ups to scenario training.
For me, a scenario involves roleplaying a specific scene- actors, props- trying to make things as real as possible. Somewhere in the middle there is practice for problem solving and free play.
One way to envision the difference is to start with a problem. For an example, you are pinned up against a wall. The threat has his left forearm across your chest, gripping your left collar in his left hand. He has a knife in his right hand against your throat.
For Drill training the instructor tells you exactly what to do and you practice it. Ideally it is a short, scripted and damned efficient set of motions that solve the problem. Even here there are elements of scenario training. One of the biggest, something that is always critical but more obvious in a scenario and often lost in other levels is to fight to the goal.
What the instructor drills will be heavily influenced by his or her own trainings and beliefs. Ergo, if the goal is to escape, the response practiced may be very different than if the goal is to neutralize the threat. It will get even messier if the instructor confuses strategic goals (escape or neutralize) with tactical goals (control the weapon).
Anyway, drills are set and scripted. They can be invaluable for instilling quick responses to specific situations. Properly evaluated, they can give quick and accurate insights on efficiency at all levels (are the goals realistic? Do the body mechanics actually work? Are the motions enhancing or inhibiting each other?*) It's especially important to practice drills at variable levels of resistance or with strangers. Often students have been subtly programmed to help things along that simply don't work on other people.
Somewhere in the middle between drilling and scenario is free play- put the student in the same situation and it is the student's responsibility to figure a way out of it. This is interesting because a lot of factors come into play here. This is where the student gets the first taste, often, of applying principles instead of techniques. This is where they learn to adapt things to fit their own size, strength and temperament.
One of the big glitches comes up here as well: you see the students eyes dart as they try to remember what they are "supposed" to do. The best thing, IMO, an instructor to do is 1) to give the student permission to be sloppy and 2) put a very short time limit so that the student doesn't have time to remember. Or even think.
It's easy to train a student to do something faster than thought. It's much harder and infinitely more valuable to teach a student to improvise faster than conscious thought. And that is really what surviving in a dangerous encounter is all about.
If you ever, as an instructor, catch yourself telling a student they did something incorrectly that worked, give your ego a double-check.
Freeplay can be expanded a lot. Sometimes subtle differences in the situation require more reliance on principles and lest on rote technique. Giving permission to deal with the situation at a different point in time, early, can give the student a chance to deal with force dynamics rather than static position- another critical skill that both makes things harder (more complicated) and much, much easier (the more variables you can exploit the more you can accomplish with less effort.)
Then the scenario. Good scenario training requires props and roleplayers. Further, the roleplayers must know how people (both real threats and real bystanders) act under stress. They have to give realistic feeds and precursors to the student and respond realistically to what the student does.
- If the roleplayers don't act like real threats, the student gains little valuable from scenario training. They don't learn how to spot a potential threat or do a threat assessment or scale force.
- If the facilitator is constantly trying to 'trick' the student by not having the bad guys act realistically or setting up ambushes from people deliberately scripted to appear 'safe' it can create either a learned helplessness** in the student or a student who over-reacts to everything. It decreases the student's potential for survival.
- If the roleplayers don't respond realistically, if they stay on their script despite what the student does, the student misses polishing other skills, like defusing and avoidance. It goes back quickly to learned helplessness.
Like a lot of things, bad scenario training can be more damaging than no scenario training at all. Good scenario training allows a student to develop as not just a fighter, but a tactician, a strategist, a diagnostician of violence, an observer... It also forces the student to practice in tandem with laws on force and realities of legal violence.
Scenario- You're watching a pool game in a bar. Several people including the players are walking between you and the table, so they get quite close. Some are talking among themselves. A group of three men and a woman were giving you flack about looking at the woman earlier in the evening. The bouncer quieted the situation down.
So you have obstacles (pool table, bar stools) weapons and potential weapons (cue sticks, pool balls, beer steins and potentially concealed knives and guns), distractions (extraneous people, music, lights) witnesses and legal and ethical issues. The instructor has already given one of the people involved (who has a motive, it does no good to try to get the student to expect to be blindsided, the goal is to raise student awareness so successful ambushes become less likely) a practice knife with instructions to try to get the student pinned against the wall as in the drill.
The student gets to practice seeing things coming. Winnowing important information from a mass of detail. Identifying attack pre-cursors.
The instructor will go over it later:
Instructor: Why did you stay on the wall?
Student: Because you told me... oh. That was stupid.
Instructor: When did you know it was going to go bad?
Student: When those three people were arguing earlier.
Instructor: So why did you stay?
.... On and on. Were you fighting to win or fighting to leave? Was the one who attacked working solo or with the other two? How many people in the scenario were armed? Why did you (or not)*** use an improvised weapon?
The student has an opportunity to handle it strategically e.g. by leaving; socially e.g. by talking threats down or calling on allies; tactically with or without force e.g. by accessing a superior position and weapon; fighting the initial energy; spiking the pre-assault;....
And maybe that's one of the reasons that scenario training is hard. Maybe most instructors aren't ready to coach a multi-leveled reality.
*Sometimes footwork and handwork complement each other; sometimes they detract. That's why it is hard, but not impossible, to hit hard while retreating- the feet are subtracting power from the arms. Sometimes people are trained to evade with the body while trying to shock with the hands- you have to be careful that the hands aren't slowing the evasion and the evasion isn't weakening the strike...
** You can train a student, a child or even a lab rat to be helpless. By putting them not in no-win scenarios, but in scenarios where winning and losing are arbitrary, decided by an outside factor, you create a personality type that becomes progressively fatalistic and quits trying.
***Another thing- a lot of stuff that the student will do isn't right or wrong. Bad debriefers try to tell the student what he did wrong. Good debriefers get the student to examine his own reasoning and especially his or her subconscious reactions.