“Roll A Will Save:” From Edmund Gettier to Skip Williams

[bekeleven]

I'm going to begin by reproducing a larger-than-normal amount of text. I apologize. You can skip this for now, and refer back to it if you like; I didn’t want to keep making points and then going “that would be true, EXCEPT…” and then inserting the next sentence, so I’m giving you everything I can in context.

This text is taken from the player's handbook, Page 173-174. I've bolded the sections that made it into the SRD. (This text was also published on the Wizards website in one of the articles I’m linking below.)

Creatures encountering an illusion usually do not receive saving throws to recognize it as illusory until they study it carefully or interact with it in some fashion. For example, if a party encounters a section of illusory floor, the character in the lead would receive a saving throw if she stopped and studied the floor or if she probed the floor.

A successful saving throw against an illusion reveals it to be false, but a figment or phantasm remains as a translucent outline. For examples, a character making a successful saving throw against a figment of an illusory section of floor knows the “floor” isn’t safe to walk on and can see what lies below (light permitting), but he or she can still note where the figment lies.

A failed saving throw indicates that a character fails to notice something is amiss. A character faced with proof that an illusion isn’t real needs no saving throw. A character who falls through a section of illusory floor into a pit knows something is amiss, as does one who spends a few rounds poking at the same illusion. If any viewer successfully disbelieves an illusion and communicates this fact to others, each such viewer gains a saving throw with a +4 bonus.
We can supplement this with excerpts from the 4-part "Rules of the Game: All About Illusions," written by PHB co-writer Skip Williams. There’s a lot of interesting stuff in these, but I’m considering it a secondary source. Still, feel free to read up. Part 1 is useless, but included for completeness: 1 (http://rpg.nobl.ca/archive.php?x=dnd/rg/20060207a) 2 (http://rpg.nobl.ca/archive.php?x=dnd/rg/20060214a) 3 (http://rpg.nobl.ca/archive.php?x=dnd/rg/20060221a) 4 (http://rpg.nobl.ca/archive.php?x=dnd/rg/20060228a).

Interaction

The first, and hopefully smaller, issue we face is that we have 3 tiers of potential interplay with illusions: Passive (during which the character is not allowed a save), interaction (which grants a save to see through the illusion), and proof (which disbelieves with no save, collapsing the illusion into transparency). We need to determine exactly what delineates these three tiers of interplay.

For instance, just viewing an illusionary wall or floor grants no save, if you’re just passing by it, but you can still get a save if you’re inching (”https://youtu.be/oSynJyq2RRo?t=2029”) down the hall, “studying” it carefully enough. What about direct contact? Well, you can interact with a floor if you probe it, but don’t even need a save if you poke it.* Probing a finger through a fake wall still needs a save, put poking one doesn’t.

Proof

Since Plato, many epistemologists defined knowledge as “justified true belief.” That is, if you believed something, and you had a reason to believe it, and it was true, you knew it. Then this absolute baller Edmund Gettier comes in swinging with this tiny little 3-page paper that says “What if you hold a belief of something, and you have a reason to believe it, and that thing is true, but not for the reason you think? What if you think 'Plato is old, and therefore Plato is wrong.' And then it turns out that Plato is wrong, but it’s just because Edmund Gettier is an OG. Your belief is justified and it’s true, but can you be said to know the idea in question?” And then everybody did the 1963 philosophy professor equivalent of playing clips of Phil Swift saying “that’s a lot of damage” and forgetting that Plato didn’t even really say that.

I know this sounds redundant, but D&D generally takes place in worlds with magic. In fact, worlds in which we have to determine whether characters believe in illusions? Those ones just about 100% have magic in them.

This vacuous statement needs to be said because we’re coming up on the worst part. Simultaneously pedantically insignificant and crucial, the question we have to answer now is: What is proof that an illusion is an illusion?

I’ll begin by running down 4 scenarios. After each scenario, consider whether the interplay of any potential illusions was passive, interactive, or proven.


Alice is greeted at the gate by a tall guard in burnished mail. She asks to be let in, and he tells her that the count is accepting no visitors. She tells him that it's urgent, and he says that the count is accepting no visitors. She asks for his name, and he tells her that the council is accepting no visitors. Frustrated, she pokes at him with her quarterstaff, but the staff somehow penetrates (and then exits) his midsection with no resistance. She repeats this trick twice more with the same result.
Bob receives a missive from his old sorceress friend that was training under a dragon. It looks a lot like the woman herself, actually. “Hey, Bob!” it yells as it runs into the room. “I figured out long-distance imaging! I’m still in the Magis Arcanis, but check out this figment projection!”
Clair tells her party wizard that she hears goblins moving in a passage ahead. "We'll lay an ambush," the wizard declares. "Everybody back into this nook. I'm going to make an illusion that it's solid rock." He casts a spell. She uses spellcraft to identify it as Silent Image.
Darryl approaches a possibly-locked door with trepidation. But his party’s not big on the lockpicking type, so he swings his axe into the door near the handle to try and break it. Somehow, the axe goes through with no resistance. The door smells of fine mahogany.


Let’s dig in. First, Alice and the guard.

The only think we have 100% determined as proof that something is an illusion is repeatedly poking it. It’s an example found in the player’s handbook and the Rules of the Game article. Poking through something that appears to be corporeal is proof that it’s an illusion, and lets you disbelieve it with no save, which reverts it to a translucent outline.

For the second scenario, with Bob, I’m going to quickly refer to the Rules of the Game articles. I didn’t want to quote long passages, and I do only plan to use them for divining designer intent and clearing up ambiguities, but this is something I consider ambiguous:

Characters hiding behind or under the illusions here need to make saving throws to successfully disbelieve them (assuming they want to do so). The caster, however, knows the illusions aren't real. If the caster points out the illusions, the characters get a +4 bonus on their saves; in this case, the DM might want to waive the saving throws and assume disbelief to save time.
In perhaps the second clearest example presented, the caster telling you that an illusion is illusory is also —sometimes, with DM approval— allowed disbelief with no save. This passage is a little weird, but for now let’s accept it. That means that our first two scenarios are both “Disbelief, no save, see translucent outlines.”

Do you see the flaw? You see it, don’t you.

What if I told you that in the first scenario, one of the following things happened:

A wizard made the tip of Alice’s quarterstaff ethereal.
Alice is ethereal, and forgot this fact.
A wizard cast Modify Memory on Alice to make her believe she had poked the guard several times when she had not.
The guard’s appearance is due to a nonmagical optical illusion or hologram.
Any other explanation (left as an exercise for the reader).

And in the second situation, Bob’s friend just teleported home for a visit and wanted to screw with him. In neither case are there Illusion (Figment) spells in play, and thus, in neither case is there a translucent background to see.

But the weirdest thing just happened in these two examples. Both of these apparent illusions got studied or interacted with, and yet regardless of whether they were illusions, no will saves were rolled. One situation in both cases was an evidence-based disbelief, and in the second case, there was no will to save because the guard and the sorceress were both real. It’s possible for the burden of proof to be met to allow for evidence-based disbelief in situations where there might not even be anything to disbelieve.

Let’s look at further examples.

In Clair’s case, the main thing I wanted to point out is that the above passage from Skip Williams, it suggests that the caster pointing out the spell he casts doesn’t necessarily void the will save, but could be considered as a time-saving measure (you know, because all saves rolled at +4 automatically succeed). This means that it’s possible for Clair’s wizard party member to say “I’m casting Silent Image,” then she Senses his Motive and believes he’s telling the truth, then spellcrafts his casting and identifies it as silent image, then rock appears around her, and if she rolls a bad will save, she thinks she’s encased in stone.

A suggestion I’ve seen to remedy this specific case is that characters can be “of two minds” about an illusion: What they believe can be separate from what they perceive. In other words, just like somebody on a roller coaster might have to shut their eyes and repeat “it’s safe it’s safe it’s safe,” maybe Clair has to shut he eyes and repeat “I can breathe, I’m not encased in rock” when she fails her save. I would consider this a partial fix, but be wary: It comes with a whole new axis of fuzziness grafted onto the existing fuzziness, since we’re tracking two different forms of consideration. And while it might make intellectual sense, it still means that Clair is the only member of the party that won’t see the goblins arrive.

As far Darryl, this is just a piece of trivia about figments. The key is in the final sentence. Swinging an axe straight through a door meets our standards for proof of disbelief, but there’s a twist. You see, smelling the door means that it can’t be made using Silent Image. Instead, it was most likely made with Major image, or one of the P images (Permanent, Persistent, or Programmed). Now, all of the P images act like major image, but they all inherit from Silent image, meaning they duplicate functionality, and do so imperfectly. Major Image has the following text: “The image disappears when struck by an opponent unless you cause the illusion to react appropriately.” This text is absent in silent image and all of its other children. So due to poor OO practices (https://en.wikipedia.org/wiki/Inheritance_(object-oriented_programming)), we know a bit more about the spell in question.

Conclusion
Illusion spells are hard to adjudicate, and anybody pitching an easy solution is selling something. The more these spells come up in a game, the more the DM is just going to have to go with their gut and try not to let any players feel either omnipotent or cheated.

I hope that, at best, readers come away from this more skeptical of the idea of “proof.” 2022 is a post-truth society and we don’t have access to literal actual memory- and sense-altering magic cast by literal wizards.

The logic of “saving against an illusion lets you see through it” is commonly cited by optimizers (See: the ol’ “Shadow Evocation Deeper Darkness” trick), but frankly, I think it causes as many problems as it solves in typical play. Give it a thought and judge for your own table.

*Ok, technically if you “spend a few rounds poking” it. I stand by my conclusion that this text does not adequately give DMs or players tools to run or play games.

[SquidFighter]

First off, great read ! It is an interesting foray into the logical implications of RAW in an actual world.

But primarily it underlines a very important idea we tend to forget around these parts : You can't actually play RAW. D&D, notably 3.X, can't run as a computer program. At least it doesn't remotely make any lick of sense if you try. (The many tomes of the dysfunctional handbooks are a lovely indicator, I think).

With regards to illusions, I think it is important to differentiate between DM illusions and player illusions. DM illusions are basically puzzles (of which you gave us plenty of fine magical details that can thicken it !). Player illusions are tools, which the disbeleive rules serve as guidelines to adjucate the NPC's reactions, but aren't limited by them. At that point, I think it's unreasonable to try to Optimize illusions by appealing to RAW.

[Duke of Urrel]

Concerning the problem of proof: Absolute proof is a rare thing, even in the real world. Karl Popper said that science doesn't work by validating any true theory once and for all, but rather makes progress by falsifying false theories. We can be sure that certain theories are false, but we can never be sure that any theory is absolutely true. We just have to keep testing each theory that we believe to be true by trying to disprove it, again and again.

So we cannot require absolute proof as necessary to disbelieve an illusion automatically, without having to make a saving throw. The closest that we can get is something like a reasonable assumption guided by Occam's Razor. In other words, which assumption is the most likely for Alice: that this guard, which she has just poked through with her staff, is an illusion, or that part of her staff has somehow just gone ethereal? When you have competing theories, you have to go with the one that is the least far-fetched. I think we should apply the Occam's Razor standard to determine when we can automatically disbelieve an illusion.

As a dungeon master, I add the +4 bonus to your disbelief Will save only if I want to know whether you have the power to see through a visual illusion or not. If your ally creates an illusion and your ally tells you that they're creating an illusion, I think you don't have any trouble trusting your ally and believing that the illusion isn't real, but I assume that you don't disbelieve the illusion strongly enough to see through it. This is something like the roller-coaster effect. Your ally just told you that the illusory wall she just created isn't real, but it sure looks real to you – unless you make your disbelief Will save at +4. By the way, once you have been told that an illusion isn't real, I allow you to make disbelief Will saves repeatedly; that is, I let you make another disbelief Will save every time you take another move action to study the illusion. This is something that I don't normally allow, because if you don't have reason to believe that an illusion isn't real, I assume that you usually stick with your first assumption. Being told by an ally that an illusion isn't real should give you a powerful advantage. And of course, if your +4 disbelief Will save fails, you can just walk up to the wall and confidently thrust your fist through it, which enables you to disbelieve the illusion automatically.

If you actually stand inside an illusion, I always consider you to be penetrating this illusion and always allow you to disbelieve it instantly. I assume, as a house rule, that no matter what an illusion looks like on the outside, it is always perfectly hollow on the inside. So when you're inside an illusion, you automatically disbelieve it and see it only as a translucent outline – indeed, an outline that you look through from the inside.

The special bubble-like frailty of illusions created by the Major Image spell is a problem that needs an interpretive solution. Here is my solution. The Major Image spell, in contrast to the Silent Image spell and the Minor Image spell, enables you to create an illusion that you can move about throughout the spell's range. So I assume that, by default, the illusion that you create with the Silent Image spell or the Minor Image spell cannot be moved outside the space in which you first created it. When you use the Major Image spell, which for the most part is based upon the Silent Image spell and the Minor Image spell, I allow you to create images that are hologram-like and oblivious of all physical attacks, but only if you confine these images to the space in which you first cast the Major Image spell. If you create a big creature, for example an illusory dragon, and cause it to move outside the space in which you created it, it becomes as frail as a bubble and bursts when it is struck by an attack. I make this rule because I believe it makes no sense that the Major Image spell should be weaker than the Silent Image spell and the Minor Image spell.

[Seward]

At most tables you got the basics of....

If you interact and save, yeah, it goes transparent or whatever.

Interaction means working through a second sense generally (touch, noticing a silent image makes no sound if it is moving, scanning with detect magic and seeing it is radiating illusion, noticing a lack of smell if it isn't a major image. Note that all figments fail to deal with touch sense, so a major image tends to need to be poked physically or scanned with magic to "interact"). However some GMs will say that perceiving the illusion is interacting and always give a save for ghost sound, say, but I think that is clearly wrong, barring something like using a disguise spell and getting details wrong (mechanically a spot check vs disguise check) and even then you won't make the illusion go transparent, you'll just know that the person is in disguise. (hitting them with sword gives the will save against disguise self, but penetrating the illusion might not reveal the identity if they are also using a mundane disguise under the illusion)

If you know it is an illusion but don't save, it is still there, and can block vision etc, but it is no more effective at blocking you from, say, moving than a fog cloud. Claire knows the wall is an illusion but blew her save so she can't see through it without poking for more saves, which might give away the party. So she sucks up being blind until her party members tell her to move. This sort of thing can be important (a shadow conjuration fog cloud will block vision past 5' for anybody who fails a save, but not for those who make a save until 25' is reached, as it is 20% real).

Being told it is an illusion or doing a spellcraft check of high enough to identify it as an illusion(figment) spell seems to give a save at +4, but only once. More interaction is needed if you blow the save.

If you don't interact with it, and have no outside evidence it is an illusion you don't get a save, period, and you perceive the illusion. This is useful if it is being used to signal or communicate, and very useful for the case of making an illusory wall to screen against goblins, unless one is bright with the Kn Architecture/Engineering or they know the terrain well enough to be suspicious that an alcove isn't there anymore or the tunnel is narrower.

That said, illusion(figments) are about tied with grapple/combat maneuvers for causing table variation and cautions of "this is a great spell, unless your GM hates it and makes sure it never works" are common. It is something that if you want to indulge beyond disguise self, illusory walls, making minor distractions or shadow magic, you absolutely need to talk to your GM before playing that character.

[NichG]

I think it'd be useful to have a metaphysics of 'how illusions work' and derive disbelief/etc effects from that. To that end, and not trying to agree 100% with RAW, I could imagine something like this:

Illusions (figments and glamers) are disembodied information which is passively taken in by minds that can perceive their anchor - the true form of an illusion is extremely weak vibrations, color variations, murmurs of sound, etc - possibly related to the intended content but not identical to it. When processed by a mind, those things cause the target's mind to fill in details as if whatever the illusion is trying to be were really there. This is a sort of multistable perceptual illusion like a Necker cube or magic eye picture, where the target can either perceive the fluctuations as what they represent, or as those fluctuations themselves, but not both at once. Furthermore, while conscious thought can prime such things, ultimately it's a subconscious process that determines which way things fall.

The thing that matters to whether an illusion is broken to a perceiver is the balance of evidence between those perceptual modes. So if someone interacts with an illusion such that they themselves find it hard to subconsciously rationalize the result as consistent with what the illusion is trying to be, that's what matters for disbelief.

Or in other words, it's like dream logic. You can notice something is wrong without fully disrupting the dream (lucid dreaming), which for an illusion would be a disagreement between conscious and subconscious perception of that illusion. The character has failed a save against the illusion, but also knows that it is an illusion (e.g. they were told so). But if their subconscious is pushed so far that it can't fill in the details, it can flip to the other perceptual mode and actually not even perceive the illusion fully.

This would also mean that different viewers could disagree on the details of what they see when perceiving an illusion, because the illusion relies on the subject's brain to fill in missing details. More skillful illusion work (higher DCs) would fill in more of those gaps.

Patterns extend this brain hacking to produce physiological reactions, not just perceptual effects. Phantasms lack the anchor point in the physical world and just transmit the hostile information directly.

Shadows are what happens when the line between perception, belief, and reality is crossed, on the basis of the illusion being embodied in a material whose reactions and interactions are sensitive to such things, the same way that e.g. Limbo is a morphic plane and reactive to focused will.

[Doctor Awkward]

Conclusion
Illusion spells are hard to adjudicate, and anybody pitching an easy solution is selling something. The more these spells come up in a game, the more the DM is just going to have to go with their gut and try not to let any players feel either omnipotent or cheated.

In games that I run if at any point a player suspects that their character may have encountered an illusion, they may spend their standard action making a Spot check, which is rolled as normal.

If there is actually an illusion, this counts as an interaction, and I roll their will save in secret to see if they disbelieve successfully. If it's not an illusion, they get whatever the result of their Spot check entitles them to.

They may do this regardless of the distance between their character and the effect.

[bekeleven]

Good lord, this topic is so deep I didn't even touch on Skip Williams mentioning that you can willingly not roll for disbelief. I even included that quote in the post. I can't think of many applications, though... maybe blocking your LOS to a nymph? (RAW, Seeming wouldn't help you there, but an image of a barrier would if the nymph can't cross it)

[Seward]

If there is actually an illusion, this counts as an interaction, and I roll their will save in secret to see if they disbelieve successfully. If it's not an illusion, they get whatever the result of their Spot check entitles them to.

.

Yeah, I like that. Passive senses don't get a roll, but an active check for discrepancies could get you a save (eg, a flaw in the ghost sound of bird song, or the fact that the wall isn't a perfect match for the other stonework). For scent, survival might be the skill check since it's the one that makes scent tracking better. It also means the will save doesn't have to be done before the game starts on a notecard without raising suspicions, since they know you're rolling just for form sake if there is no illusion.

[remetagross]

Good lord, this topic is so deep I didn't even touch on Skip Williams mentioning that you can willingly not roll for disbelief. I even included that quote in the post. I can't think of many applications, though... maybe blocking your LOS to a nymph? (RAW, Seeming wouldn't help you there, but an image of a barrier would if the nymph can't cross it)

That Shadowcraft Mage build that gets 120% real illusions/shadow effects might want to intentionally not roll for disbelief against some illusion effect having a positive outcome for them?

[YellowJohn]

That Shadowcraft Mage build that gets 120% real illusions/shadow effects might want to intentionally not roll for disbelief against some illusion effect having a positive outcome for them?

I thought shadow conjuring a Magic Tattoo was the classic example for intentionally not rolling disbelief :smallsmile:

[icefractal]

The "turning transparent once you disbelieve them" thing is weird, and IMO causes more problems than it solves. I think if I were rewriting them, I'd split it into two categories:

Figments - These are real holograms; they don't become translucent at all, and you don't get a save. When looking at it closely or carefully, you can make a Perception (vs Spellcraft, probably) check to see flaws in the image. Even if you don't spot flaws, it's still not physical and you can stick your hand through, although you wouldn't necessarily know it's an illusion (as opposed to an incorporeal creature, for example).

Phantasms - These only exist in the minds of observers. So you always get a Will save, the first time you'd sense it, but if you fail the save then not only do you not notice any flaws, but your mind actively works to excuse any discontinuities. So if you tried to walk through a phantasm wall, you'd perceive that you bumped into it. But an unaffected observer would see you walk up to the translucent wall and voluntarily stop just before touching it. Since it isn't physical though, someone could shove you through the wall, at which point it'd be either auto-disbelief or make another save.

[Seward]

The split icefractal describes is exactly how the Hero Games system does it.

They have images and they have mental illusions. The former create stuff that is absolutely visible to whatever sense they are tuned to. Light images create photons, sound images sound waves, scent actual scents etc or they at least fool your nervous system into reporting those things in a way identical to the real thing, depending on special effect. You might figure out the image isn't real but that doesn't affect how it affects your senses.

Mental illusions range from fuzzing details to putting you in an alternate reality while your body interacts with it and stumbles around or in most extreme cases doesn't move at all, and can knock you out or kill you if they are powerful enough. (light so bright it hurts is either flash if it blinds you temporarily, darkness if it is so bright you can't see while in it or an energy or killing attack if it does damage. It might ALSO provide an image effect. something like Blinding color Surge would be very much in the Hero Games style). If you break out of a mental image you come back to reality, it doesn't persist, and if they don't stick it in the first place it is similar to effects like Phantasmal Killer where you know you were under mental attack but might not know what it was if you make that will save and can't easily find the source of the attack or have some unusual sense to see it trying to affect you.

D&D has legacy issues involved. Originally there was only Phantasmal Force, which was both figments and phantasm. It could convince you "rocks fall everybody dies" or be used to change the color of everybody's hair. It also lasted as long as you could concentrate and barring taking damage, you never had to stop concentrating. In one campaign I ran PC's put their illusionist on the mage/thief's tenser disk, built a fort around him and made it look like a chest of loot. The illusionist (level 1 due to dual-classing) spent a couple adventures pretending he didn't exist and aiding the party with various illusions (eg, wizard points a wand and does a command word and shoots a real lightning bolt out of it. Phantasmal Force lightning Bolt also goes off....pretty convincing no? Smell, scorch marks etc. Just does more damage if you fail save vs spells). It was a new party member so the bad guys had no idea there was an extra opponent in the treasure chest (which even had a false bottom that let them toss loot on top to add to the effect).

It worked great till the party encountered a bunch of fire priests and the fort was flamestriked into oblivion while targeting other party members and the illusionist's cover was blown, literally, to bits.

The illusionist class was created and added multisensory versions of it but also spells that split out some of those functions, like color spray, that had much more clearly delineated boundaries. But there are weird remnants of the original phantasmal force spell in the magic system. The figment going transparent is one relic. The entirety of the shadow magic line of magic with special effect reskinned as "shadow plane energy" for the partial damage on a save is another relic.

[Duke of Urrel]

Good lord, this topic is so deep I didn't even touch on Skip Williams mentioning that you can willingly not roll for disbelief. I even included that quote in the post. I can't think of many applications, though... maybe blocking your LOS to a nymph? (RAW, Seeming wouldn't help you there, but an image of a barrier would if the nymph can't cross it)

I think the most important use of the option to forego a disbelief Will save against an illusion is to make certain applications of the Shadow Conjuration spell work for you. For example, suppose you use the Greater Shadow Conjuration spell to mimic the Wall of Stone spell, so that you can create a bridge across a chasm. You're definitely going to want to forego your disbelief Will save so that the bridge will remain solid under your feet.