The Bundle Theory of Belief
We have a tendency to suppose that beliefs are discrete states of mind, cleanly separated from each other. They exist like so many peas in a pod or sentences printed on a page. But this is not a realistic picture: beliefs come in groups. The unit of belief is the bundle. Take my belief that the goldfish in my pond are thriving: that belief comes with a set of other beliefs about how I acquired the goldfish, how they are cohabiting with the frogs, how they look, the number of them, the fact that I have a pond in my garden, that I don’t live in an apartment any more, etc., etc. This collection of beliefs may not be endless, but it is certainly large; and one belief gains comfort and support from the others. Many of my beliefs don’t belong to this bundle, having their own bundles, but it is really not possible to have a single belief in isolation. Beliefs proceed from premises or assumptions and they have consequences as well as corollaries and associations; they nestle in among these ramifying beliefs. Psychologically speaking, a belief is a component of a belief package, not merely one belief set beside others. Just as properties of objects always coexist with other properties, so beliefs coexist with other beliefs. I don’t just believe, I co-believe—I harbor bristling bundles of organically connected beliefs. The idea of a single isolated belief is an artificial abstraction; the psychological reality is more holistic (if I may appropriate an overused term). Words make up sentences (word bundles) and have no existence independently of that; similarly, beliefs make up belief bundles and have no existence otherwise.
We fail to see this, or to grasp its implications, because we like to individuate beliefs by their content: distinct content, distinct belief. Thus an array of distinct propositions is supposed to generate an array of distinct beliefs—one for each proposition. But that is really a non sequitur, since an array of propositions might be the content of a single complex psychological state. Call such a state of belief a T-belief (“T” for “total”): then we can say that the subject T-believes the whole set of propositions (those that make up its bundle). The idea, then, is that from a cognitive point of view the basic unit of assent is T-belief—a pattern of assent. Talk of isolated beliefs is just so much abstraction. Assent is always to multiple propositions, so we may as well recognize that psychologically beliefs are members of a family—they travel as a team. They are parts of something larger—though parts that cannot be broken off and expected to survive intact.
Let me illustrate the point with a well-worn example. Suppose someone knows well enough that Hesperus is Phosphorous, so that they will assent equally to “Hesperus is F” and “Phosphorous is F”. We can agree that the two propositions are not identical, but surely there is no difference in the belief the subject has in respect of those propositions. It is a matter of indifference whether they express their belief state using “Hesperus” or “Phosphorous”. They have a single state of belief that can take either proposition as its content—just as many sentences could be used to express this belief. The distinct propositions correspond to no psychological distinction for them. If you ask them whether a certain planet (Venus) is pretty, they will respond with either name, since they believe the names refer to the same thing. Maybe the beliefs were once different, before the subject discovered the identity, but now they have merged, despite the distinction in the propositions. Or consider the beliefs of animals: we needn’t fuss over the precise proposition that captures the belief state of a dog or orangutan because any of a range of propositions will do to capture their tendencies to assent—their psychology is just not that fine-grained. In the case of my belief bundles all that matters is the overall pattern of assent—I assent to all of them simultaneously. I T-believe in a certain set of propositions: that is the basic fact, not the many discrete beliefs beloved of philosophers. I certainly don’t introspect discrete belief states that make up one of my bundles, as if each belief is an isolable inscription.
Compare desire. Suppose I desire to see a certain film: isn’t this really a whole bundle of desires? I want to see a particular star, I want to see a certain genre of film, I want to sit in the dark and be entertained, I want to sit next to my significant other, I want to forget my troubles, I want to pass the time, etc. I have a complex desire state in respect of seeing this film, a whole set of pro-attitudes. If you ask me why I want to see the film, I could cite any of a large number of desires that motivate me. There is never just one; I contain multitudes. Desires come in bundles not as isolated elements. When I desire one thing I desire many things. The unit of desire is the collection—that is what motivates me. It makes little sense to try to abstract a single desire unconnected to other desires—what would it mean to say that I desire to see a film but have no other desires that go with that? The proposition is distinct enough, but the psychological state is embedded in a larger totality. All desire is really T-desire. We mustn’t reify propositions into psychological realities.
The ontology of belief-desire psychology should therefore follow the bundle conception: bundles are what dispose to action and have functional properties. It is never the case that an action is prompted by a single belief and a single desire; rather, the action results from an interconnected web of beliefs and desires. The model of an ordered pair of belief and desire causing an action is at best an oversimplification. This is not merely the point that individual beliefs and desires necessarily belong with other beliefs and desires; it is the further point that psychological reality is not well represented by the model of discrete beliefs and desires. Instead we are to think of the mind as operating in larger more inclusive units—what I have been calling bundles.
Think of the way you form beliefs. You don’t form them one at a time, now this one, now another. There isn’t a sequence of isolated beliefs laid out in time, like a series of inner explosions. Rather, a package inserts itself in toto—as when you take in a scene perceptually. Hundreds of beliefs may be formed in a matter of seconds. Nor is this the case only in perception: even in testimony cases we form multiple beliefs—about the speaker’s appearance, his history and character, the consequences of what he is asserting, and so on. At every waking moment you are updating your set of beliefs in big chunks not seriatim. It is done by the bundle, and without effort. It is entirely natural to operate with belief ensembles, because beliefs like to work in groups; lonely isolation is not their preferred mode of existence. So let’s recognize the gregarious character of belief: it’s never merely the belief that pbut always the beliefs that p, q, r, etc. We don’t so much have individual beliefs as we have belief patterns. If we think of beliefs as dispositional, then it is the bundle that has the disposition not beliefs singly. 
 Wittgenstein would say that we are captivated by a picture of belief (and desire—knowledge too) according to which beliefs are laid own like exhibits in a glass case, each self-sufficient and separate. This picture has many sources, some deriving from the physical world, some from language, some from aspects of the mind itself; but we should resist the picture and accept that our talk of belief implies something like a holistic pattern of assent. I have tried to avoid putting the point in Wittgenstein’s terms, but it may be helpful from an expository point of view.