Norman Malcolm famously argued that Anselm developed two versions of the ontological argument for God’s existence in his Proslogion, the first of which (in P. 2) turns on the premise that existence (in re) is a perfection, so that the greatest possible being must exist both in re as well as in cogitatione (p. 44), and the second of which (in P. 3) turns on the premise that necessary existence is a perfection: “His first ontological proof uses the principle that a thing is greater if it exists than if it does not exist. His second proof employs the different principle that a thing is greater if it necessarily exists than if it does not necessarily exist” (p. 46). Malcolm interprets necessary existence as existence that doesn’t depend on others (in the way that fragile dishes depend on gentle handling — otherwise identical dishes that were unbreakable would be “superior” dishes) or that’s unlimited by outside constraints (in the way that an engine that could drive 100 mph with no fuel would be greater than a similarly fast engine that required fuel) (cf. p. 47). Malcolm takes it that the first proof falls prey to Kant’s criticism that existence isn’t a predicate (since otherwise it would be impossible to form a concept of anything that happened not to exist), but that the second doesn’t — necessary existence really is a great-making property, and so belongs necessarily to the greatest possible being, whose existence must thus either be impossible or necessary (p. 49-50).
Now Malcolm suggests that Anselm didn’t distinguish these proofs, and I don’t know of any evidence that he did. However, a passage from Bonaventure’s Quaestiones Disputatae de Mysterio Trinitatis makes me wonder if he distinguished them in something like Malcolm’s way. Bonaventure sketches three different proofs from Anselm, the first and third of which seem to me to correspond to the proof from necessary existence as a perfection, and the second of which corresponds to the proof from existence as a perfection. We’ll take the second first:
Ens, quo nihil maius potest cogitari est talis naturae, quod non potest cogitari, nisi sit in re; quia, si est in cogitatione sola, iam ergo non est ens, quo nihil maius cogitari possit: ergo si tale ens cogitatur esse, necesse est, quod non posset cogitari non esse” (q. 1, a. 1, sec. 23; Quaracchi, v. 5, p. 47a).
Here, the proof turns on existence as a perfection which the greatest possible being must possess. It’s obviously drawn from Anselm’s first formulation of the proof, in Prologion 2:
Certe id quo maius cogitari nequit, non potest esse in solo intellectu. Si enim vel in solo intellectu est, potest cogitari esse et in re; quod maius est. Si ergo id quo maius cogitari non potest, est in solo intellectu: id ipsum quo maius cogitari non potest, est quo maius cogitari potest. Sed certe hoc esse non potest. Existit ergo procul dubio aliquid quo maius cogitari non valet, et in intellectu et in re.
It’s curious, and perhaps revealng, that Bonaventure presents the proofs out of order — I wonder if he brings this one out second because he regards it as the weaker of Anselm’s two formulations…
In any case, the first and third versions of the argument sketched by Bonaventure clearly turn on the premise that necessary existence is a genuine perfection:
Deus est quo nihil maius cogitari potest; sed quod sic est, quod non potest cogitari non esse, verius est, quam quod cogitari potest non esse: ergo si Deus est quo nihil maius cogitari potest, Deus non poterit cogitari non esse (Ibid., sec. 22).
This draws on Anselm’s second formulation of the proof in Proslogion 3:
Quod utique sic vere est, ut nec cogitari possit non esse. Nam potest cogitari esse aliquid, quod non possit cogitari non esse; quod maius est quam quod non esse cogitari potest. Quare si id quo maius nequit cogitari, potest cogitari non esse: id ipsum quo maius cogitari nequit, non est id quo maius cogitari nequit; quod convenire non potest. Sic ergo vere est aliquid quo maius cogitari non potest, ut nec cogitari possit non esse.
After sketching Anselm’s two overt constructions of the ontological argument, Bonaventure offers a third derivation of the proof, from a later comment in the Proslogion about the criterion for ascribing properties to God:
“Tu solus es quidquid esse melius est quam not esse” (cf. P. 9); sed omne verum indubitabile melius est quam omne verum dubitabile; ergo Deo magis est attribuendum esse indubitabiliter quam dubitabiliter (Ibid., sec. 24).
In context, Anselm’s comment was about God’s being “iustus, verax, beatus, vel quidquid melius est esse quam non esse”: Bonaventure suggests that this criterion requires that God’s existence be regarded as “indubitable” rather than “doubtful.” Here again, it’s not that God has to possess existence outside the mind rather than only in the mind; it’s rather that God’s existence must be of the sort that the mind can’t coherently doubt it at all, because it is possible and (by definition) unconditioned by anything else.
That Bonaventure regards these as two different arguments is clear from the way he formulates and addresses objections to the indubitability of God’s existence. For instance, he considers Gaunilo’s argument that if Anselm proves anything, he proves too much, since the ontological argument can be reformulated to prove the existence of “an island, than which nothing greater can be thought,” which is only such if it exists in re as well as in cogitatione (Qu. Disp. de Myst. Trin. a. 1, q. 1, obj. 6; Quar., v. 5, p. 48b). Now this objection is clearly aimed at the argument of Proslogion 2, and Bonaventure defuses it by proposing that the idea of a island greater than which nothing can be thought is incoherent, since “insula dicit ens defectivum” (Ibid., ad 6, p. 50b). Maximal greatness isn’t repugnant to the concept of God in the same way as it is to the concept “island.” (Bonaventure clearly thinks that this argument works just as well as the next one, and I think he and Anselm have a point — but the point of this post is to show that Bonaventure already recognized two distinct arguments in Anselm, rather than just one.)
He immediately goes on to consider a different objection, this time to the second formulation of the ontological proof: how, the objector wonders, can you say that God’s non-existence cannot be thought? It’s obvious that I can assert his non-existence. And if you say that those assertions are false, then you’re not saying anything more than can be said for all necessary truths, many of which are plainly dubitable (Qu. Disp. de Myst. Trin. a. 1, q. 1, obj. 7; Quar., v. 5, p. 48b). (An extreme case is Fermat’s Last Theorem, which is apparently true, but known (rather than believed) to be so only by a few people in the world.) Bonaventure’s response to this objection is strikingly similar to Malcolm’s interpretation of Anselm’s second ontological argument: it’s not that we can’t form the thought, “Deus non est,” but rather that “ipsum adeo est evidens in se et certum cognoscenti, quod si recte considerare velit, non est aliquid, per quod possit ab hac veritate avelli. Est enim verum evidentissimum et praesentissimum, quod nulli deest loco, nulli tempori, nulli rei, nulli cogitationi; nec sic est de aliis veris creatis” (Ibid., ad 7; Quar. v. 5, p. 50b). The reason that God cannot coherently be thought not to exist, is that if he exists at all, he exists in such a way that his existence is totally unconditioned by anything else (so that “non est aliquid, per quod possit ab hac veritate avelli”), because that existence totally conditions everything else (“nulli deest loco, nulli tempori, nulli rei, nulli cogitationi; nec sic est de aliis veris creatis”).