Descartes' Modal Epistemology

Rough draft. In progress.

In what follows, I will explicate what I call ‘the Cartesian Package’.  The Cartesian Package provides the basis for Descartes’ unique account of modal epistemology.  It has two main components: an account of the nature of objects and an account of forming adequate conceptions of them.  

Take the former component first.  Given Descartes’ account of homogeneous objects, every object has a structure of properties that are related to one another in a way that’s crucial for his modal epistemology, as we will see.  First, they have a principal attribute, which is the most general and fundamental feature of an object.  This feature is a determinable, and it specifies the kind of object that a thing is, and all of the other sorts of properties that the thing can have.  Descartes, of course, took extension to be the principal attribute of material objects.  

Second, objects have other attributes that flow necessarily and transparently from the principal attribute.   These are also determinables.  For the principal attribute of extension, these are size, shape and position.  

Finally, objects have modes.  These are determinates of the aforementioned determinables.  For example, being six feet in diameter is a mode of the attribute of size, being circular is a mode of shape, etc.  

Now for the most important point about the Cartesian account of the property structures of objects: their constituent properties stand in relations of essential dependence.  Start with the attributes. Each attribute is mutually dependent upon the others: they all stand or fall together, so to speak.  So, for example, size can’t be instantiated without shape, nor can shape be instantiated without size.  Similarly for the other attributes.  The relation of dependence between attribute and mode, on the other hand, is one-sided: While any mode is dependent upon its corresponding attribute, no attribute is dependent upon any particular mode.   

In short, objects have a tightly integrated collection of features: a principal attribute, non-principal attributes, and modes.  The attributes are determinables and the modes are determinates. All of these features stand in relations of dependence of one sort or another: symmetric dependence between attributes, and asymmetric dependence between attributes and modes.  

The fundamental means of forming an accurate conception of objects is the Cartesian device of intellectual exclusion.  You exclude a feature of an object from your conception of it when you hold your representations of at least two of its features before your mind, and explicitly negate one of them.  So, for example, take Descartes’ famous chunk of wax.  You exclude the representation of a particular shape S from your conception of the wax by holding the representations of the other features constant, and then conceiving of it as existing without S.  This sort of intellectual act stands in stark contrast to the act of abstraction, where one chooses to focus on one or more features of an object, thereby allowing other features to fall back into the periphery of one’s consciousness.  

Take Descartes’ wax again. We abstract our representation of S from our conception of the wax when we let S fall out of our intellectual focus, so to speak.  In short, with abstraction, the mind shifts its focus to other aspects of a representation, but with exclusion, the mind “snaps off” features of a representation.

Of course, Descartes wasn’t interested in exclusion for its own sake.  Rather, he was interested in what happens to a conception of an object as a result of excluding a constituent from it.  In particular, he was interested in whether excluding a rep resulted in rendering one’s conception of an object unintelligible.  According to Descartes – and here’s where Descartes’ modal epistemology enters the picture -- if one can exclude a representation of A from a representation of B and yet retain an intelligible conception of B, then B can exist without A.  On the other hand, if the exercise renders B unintelligible, then B cannot exist without A.  Henceforth, let’s say that a representation of A is excludable from a representation of B iff exclusion of A’s representation doesn’t render B unintelligible.

Pick up Descartes’ piece of wax again.  Since my representation of the particular shape of the wax is excludable from my representation of the wax itself, the wax can exist without having that particular shape.  On the other hand, the representation of having shape is not excludable from my representation of the wax, and so the wax can’t exist without it.

So far, so good.  But how does Descartes justify the move from excludability facts to modal facts? The details will be given later, but for now, the general rationale will suffice.  First, recall our account of excludability: A’s representation is excludable from B’s representation iff excluding the former from the latter doesn’t render B unintelligible. So if A’s representation isn’t excludable from B’s, then B is unintelligible without A.  In other words, B’s representation presupposes A’s. Thus, we see that exclusion is the acid test for determining whether one representation presupposes another.  

Second, consider what sort of worldly relation is represented by the conceptual relation of presupposition. If a representation of a real property A presupposes a representation of another real property B, and the presupposition relation here corresponds to some real relation in the world between A and B, what is that relation? Answer: essential dependence: the metaphysical impossibility for A to exist without B. One may then justifiably ask: “what justifies the claim that presuppositional relations mirror essential dependence relations?”  More will be said about this below, but I think the claim is intuitively plausible.  For how could, say, a particular shape exist unless it had the property of being shaped?  This response no doubt raises thorny issues involving the nature of universals, the realism/nominalism debate, etc., but delving into these issues further is beyond the scope of this paper. I here only want to assert that such a claim is not prima facie implausible.

Unfortunately, not all essential dependence relations are captured by presuppositional relations – e.g., the property of being water essentially depends on the property of having hydrogen atoms, but the representation of neither property presupposes the other, and thus excluding either one doesn’t render the other unintelligible. On the other hand, it looks as though the converse holds: all presuppositional relations capture essential dependence relations.  Whatever your thoughts on these matters, we’ve already become acquainted with two such types of properties: determinables and determinates of principal attributes.

Now for the punchline.  Putting these points together, we get the general Cartesian rationale for the inference from excludability facts to modal facts.  If you have an accurate representation of two properties A and B, together with the fact that A essentially depends on B, then this is mirrored in your representation by a representation of A, a representation of B, and the fact that A’s representation presupposes B’s representation.  These facts legitimate exclusion as a tool for doing modal epistemology in such a case.  For application of the tool of exclusion will reveal when A’s representation presupposes B’s representation, and if these representation and the presupposition relation between them accurately represent A, B and the relation of essential dependence between them, it will thereby reveal the modal fact that A can’t exist without B. 

In short, exclusion is a powerful tool for doing modal epistemology.  For it reveals facts about presuppositional relations between representations, which in turn correspond to modal facts – in particular, essential dependence facts -- about objects and properties.  And finally, we know of at least two sorts of properties that both stand in essential dependence relations and are mirrored in presuppositional relations between representations, viz., determinables and determinates of principal attributes. In a word, Exclusion is legitimate because it exploits presupposition relations between constituents of a concept, which in turn mirror relations of dependence between the features of an object.

So far, we’ve seen that exclusion is a reliable means of gaining modal knowledge under certain conditions: when applied to representations of the determinables and determinates of a principal attribute.  But this isn’t enough to show that exclusion handles the worries raised in our discussion of the Problem Cases.  To do that, we need to do two things: give an account of objects that have all and only such determinables and determinates, and provide reasons to accept that account.  Happily, our job is already half over, for we’ve already seen an account of objects with this sort of property structure, viz. Descartes’ account.  I will therefore go straight into the second task.  My strategy for accomplishing it is as follows:  First, I will briefly state and explain the basics of Descartes’ theory of distinction.  Then, I will put it to work to form reps as if of material objects that have such features.  Finally, I will bring in two lines of reasoning from Descartes’ corpus to argue that material objects have no other properties.

As we’ve seen, the device of exclusion is the basis of Descartes’ key tool for doing modal epistemology.  From this simple device, Descartes has all he needs to generate his theory of distinction.  For just as there are three possible ways in which representations of A and B are excludable – in one direction, both directions, and neither direction – so there are three possible ways in which A and B can stand in dependence relations to one another – in one direction, both directions, or neither direction.  Corresponding to each pair of exclusion/dependence relations are relations of distinctness.  If A and B are mutually excludable/independent, then they are really distinct; if A and B are excludable/dependent in one direction only, then they are modally distinct; and if A and B aren’t identical, yet excludable/independent in neither direction, then they are rationally distinct.

(Side note: How can one discern rationally distinct features if neither is excludable in principle? There are at least two main accounts of how this might work. According to the first, humans have a basic capacity of noticing that two features differ from one another. According to the second, one can deploy the previously mentioned Cartesian device of intellectual abstraction, whereby one attends to one feature while letting the others fade into one's intellectual periphery.)

Corresponding to each relation of distinctness is a pair of modal claims: If the representation of A is excludable from the representation of B, and vice-versa, then A and B are really distinct, in which case A can exist without B and vice-versa; if the representation of A is excludable from that of B, but not vice-versa, then A and B are modally distinct, in which case A can exist without B, but not vice-versa; and if neither the representation of A nor that of B is excludable from the other, then A and B are rationally distinct, in which case neither can exist without the other.

With Descartes’ theory of distinction in hand, one can use it to form adequate conceptions of objects.  To make the process more concrete, grab the wax again.  First, form a representation of the object’s most general feature – its principal attribute – and then start excluding as many associated representations from it as possible. In the case of our trusty old piece of wax, this will turn out to be extension in length, depth and breadth. One will immediately find that representations of several other properties are neither excludable from it nor from each other – the determinables Size, Shape, Position and Motion.  Thus, the theory of distinction tells us that they are all only rationally distinct from each other.  Continuing the process, one finds that while all other properties are excludable from the ones just mentioned, the converse doesn’t hold.  So, for example, while one can exclude one’s representation of the current particular shape of the wax from that of its attribute of shape, one can’t exclude the representation of the attribute of shape from that of the particular shape.  A similar story is true for the other determinables that are the attributes of the wax, and their corresponding determinates. One thus concludes that the excludable properties are modally distinct from the object. Finally, one discovers that everything else is excludable from the conception of the wax, and conversely, in which case one concludes that there is a real distinction between the wax and everything else.

Thus, the theory of distinction gives us a conception of an object with a principal attribute, several non-principal attributes, and their modes.  The attributes are determinables, and the modes are token determinates of them.

It’s crucial to notice two key facts about conceptions of objects formed by using the theory of distinction in this way.  First, all of the representations within such conceptions are conceptually interconnected by presupposition relations: symmetric presupposition relations among the representations of the attributes, and asymmetric presupposition relations between the representations of the attributes and the modes.  

Second, these features are well within our ability to grasp them: if objects have only such properties, then they aren’t Deep Structure Cases.  Indeed, Descartes has provided the materials for arguments for thinking that they have no other features.

Descartes gave two main arguments for this “closure clause” regarding physical properties, one a priori and the other a posteriori.  

Take the a priori argument first. Descartes thought that he had demonstrated the existence of a perfect being – a being similar to Anselm’s god.  Since perfection includes moral perfection, and moral perfection precludes deceit, it follows that this being wouldn’t deceive us.  And since such a being would be deceitful if he allowed us to go wrong about the nature of objects when we use our best efforts and methods, then since our best efforts and methods only allow us to know that objects have extension, size, shape, position, motion and their corresponding modes, it follows that objects have all and only such properties.  As we’ve mentioned earlier, however, this sort of justification has limited appeal.  Thankfully, Descartes has an argument with much broader appeal.

Descartes’ a posteriori argument is an argument from empirical adequacy and theoretical parsimony.  He expresses it in a number of places, but it servers as the master argument in his The World.  In this work, Descartes undermines Aristotelian physics by explaining all natural phenomena in terms of the motion of the simple extended substances discussed above.  Thus, in response to Aristotelian explanations of terrestrial phenomena ultimately in terms of hot, cold, wet and dry, he says:

"…I shall say to you that these qualities themselves seem to me to need explanation.  Indeed, unless I am mistaken, not only these four qualities but all the others as well, including even the forms of inanimate bodies, can be explained without the need to suppose anything in their matter other than the motion, size, shape, and arrangement of their parts."

According to this argument, all physical phenomena can be explained in terms of objects that have all and only the determinables and determinates on our aforementioned list.  Therefore, considerations of empirical adequacy and theoretical parsimony lead us to postulate that material objects have these features, and no others.

In short, Descartes’ modal epistemology is captured in his theory of distinction. The latter is justified as a legitimate tool for doing modal epistemology because it utilizes the device of exclusion on representations of objects whose properties are exhausted by a small, transparently obvious, and conceptually integrated collection of determinables and determinates.  That objects have such properties (and no others) is justified by three considerations.  The first consists in the fact that applying the theory of distinction lays bare a minimal collection of such properties in objects; so, we’re justified in thinking that they have at least such features.  And the second and third consist in an a priori argument from the goodness of God and an a posteriori argument from empirical adequacy and theoretical parsimony, both of which justify the conclusion that objects have no other features. 

With the Cartesian Package before us, we are now in a position to see how Descartes can handle the difficulties raised by the Problem Cases.  For recall that the Problem Cases consist in Deep Structure Cases and Lawless Feature Cases.  

Lawless Feature Cases occur when an object’s accurate conception has representations that are cognitively accessible yet not conceptually interconnected.  But if the three considerations above are on track, then if one forms a conception of an object using Descartes’ Theory of Distinction in the way recommended, then one will form an accurate conception of an object whose constituent representations are all interconnected in the relevant way.  The further worry that they may have other properties whose representations aren’t thus connected is answered by the a priori and a posteriori arguments for the “that’s all” clause that objects have no other properties.  

Similar reasoning enables Descartes to answer the Deep Structure worries.  For recall that Deep Structure Cases occur when an essential property of an object is beyond one’s cognitive grasp.  But we’ve just seen that application of the theory of distinction tells us that objects have at least a set of essentially interdependent properties, and the closure clause arguments tell us that they have no others.  Thus, we can be confident that objects have no others, and so we can be confident that objects don’t have features beyond our grasp.  Thus, Descartes’ modal epistemology gets around the worries raised by the Problem Cases – something that cannot be said of any other account of modal epistemology on offer.

To summarize our discussion of the Cartesian Package: objects have a tightly integrated structure of determinables and determinates – Descartes’ “attributes” and “modes” - such that the former depend on each other, and the latter depend on the former.  That they have at least these features is justified by careful use of Descartes’ theory of distinction.  That they have no others is justified by an a priori argument from a non-deceiving god and a posteriori considerations of empirical adequacy and theoretical parsimony.  Careful utilization of Descartes’ theory of distinction gives us an accurate conceptual “map” of these property structures, and the presuppositional relations within these conceptions mirror the essential dependence relations among the features of these objects.  These presuppositional relations thus legitimate modal claims based on utilizing the device of exclusion, enabling one to rule out Lawless Feature cases.  The arguments for the “that’s all” clause enable one to rule out Deep Structure cases. This, then, is the classical Cartesian account of modal epistemology, and how it gets around the Problem Cases.