It's natural to think in terms of explicit goals where players have to complete a particular task like dodging the fireballs or finding the key in order to succeed. The problem with explicit goals is that there is definitionally some possibility for failure built into the game. Maybe the player can't find the key or manage to dodge correctly and the narrative grinds to a screeching halt because they are unable to proceed.
As a brilliant example of an implicit goal, I submit the tutorial for Eternal Darkness: Sanity's Requiem. The game is genius in many regards, but I find the initial introduction particularly genius. Shortly after selecting “New Game,” the player finds themselves trapped in a room with an endlessly spawning horde of emaciated zombie-like creatures and armed with a shotgun. The player then discovers how to aim and discharge said shotgun into the mob for a short while until they awaken in a cold sweat to discover the combat tutorial was a dream all along!
This sequence is brilliant in terms of pacing in that it immediately grabs the player's attention for the following expository cutscene, rather than like most games where the player would impatiently sit through this cutscene waiting to be allowed to kill something. (It also does a good job of immediately humanizing the protagonist, which is also something a great many games struggle with.)
But what is pertinent here is the fact that shooting the zombies is a wholly implicit goal. There is no health bar introduced yet. It does not matter how effectively one deals with the zombies. The player could get completely mobbed by them, or she could dispatch them with ease and efficiency. But it doesn't matter. Mechanically speaking there are zero stakes; the heroine awakens in a cold sweat regardless. And furthermore it doesn't matter that it doesn't matter. The player isn't thinking “I don't see a health bar so I'll just let the zombies maul me,” she's thinking “Ahh! Zombies! Get away! Take that!”
Were there no consequences to zombie-mauling for the entire game, it would quickly lose its horror-aesthetics and feeling of challenge. But the purpose of the tutorial is not to provide challenge, it is to put across the basic mechanics in a way that also serves the narrative flow. (Also no player is going to replay the tutorial and be annoyed that they can't lose, because at some level they recognize what the tutorial's purposes are and aren't.)
Another example of this idea is the indie game Dys4ia, which attempts to put across the designer's experience of going through a sex change procedure via a series of often-abstract minigames which serve to convey various emotional circumstances. While there are some points at which an action must be performed by the player in order to proceed (making them explicit goals imposed by the mechanics), in each of these cases the action is both easy and glaringly obvious (e.g. press "up"). In any minigame in which there is actually some challenge posed, such as when the player deflects hurtful comments with their shield, there is not actually any penalty for failing to do so. Perhaps this is because we are all ingrained with games like Pong (which at one point maybe did require its failure penalties to teach the correct play dynamics), but it in no way detracts from the intended experience that the player cannot fail, and in fact having to retry a minigame would completely ruin the artistic message. The game isn't trying to tell you "if someone's comments hurt you, just keep listening to them over and over until you are able to deal with it." (Of course the unintended message of most games is "if you fail to defeat this world-ending threat, don't worry, you can just keep trying until you inevitably succeed," but that's another discussion.)
The point is that player behavior is guided by more than explicit goals, and additionally having an explicit goal can actually ruin the intended experience. Would it be a good experience for the player to fail the tutorial in Eternal Darkness and have to try again? No! It would be annoying, feel "gamey" and also vastly reduce the menace of the zombies when they are reintroduced much later. Luckily there is no need to make failure a possibility in this case because the player will behave as intended regardless, and the sequence loses no excitement or impact.
If we assume that players' behavior is guided only by what is mechanically defined as success and failure, we hamstring our ability to craft certain experiences. For example, escort missions are almost universally hated in games. This is because the goals of these sequence tend to be too explicit in the sense that if the character is not escorted properly, the player fails the sequence and is forced to go back and try again. On the opposite end is Telltale's The Walking Dead game (which in my opinion is one of the best games ever made) containing a couple of sequences in which you are trying to "protect" a character, but where failure is actually impossible (in one, you cannot make it to them in time, but another character steps in). These sequences work because it is unclear when the player is in danger of being penalized for failure, as well as being unclear what the ramifications will be (sometimes it means trying again, sometimes it means a beloved character is permanently dead). Secretly mixing between these three types of sequences (ones where you can't fail or it's almost impossible to fail, ones where you can try again, and ones where the effects of your actions are permanent) allows the game to ambiguate player assumptions in order to craft an experience (or rather, a set of possible experiences) which successfully puts across the intended emotions.
We can argue about whether The Walking Dead is really a game or an interactive story (it's both, and to varying degrees all games are also interactive stories), but the answer is irrelevant. What matters is that, through a combination of story and mechanics, The Walking Dead is successfully putting across unique emotions and experiences that no other game has managed, and so is Dys4ia, and they are doing so by subtly challenging longstanding conventions about how games, and gamers, work.