Knowledge and Belief

The traditional analysis of knowledge as true justified belief encourages the following picture (though it doesn’t strictly entail it): there is a common psychological element to knowledge and non-knowledge, which we can call “belief”, and knowledge is the result of supplementing this element with further conditions that together add up to knowledge properly so called. Thus when we combine this common element with truth and justification the upshot is an item of knowledge; the element itself can exist quite independently of knowledge, and often does. For example, beliefs can exist as a consequence of desires and be neither true nor rationally justified, as in “wishful thinking”. This psychological state has no intrinsic connection to knowledge, being capable of existing in the absence of the conditions that lead to knowledge. It is not in the nature of belief to constitute knowledge; there could be a whole system of beliefs none of which count as knowing anything. Beliefs have no intrinsic connection to truth either: they are the same whether true or false—a false belief is just as much a belief as a true belief is. Beliefs that result from faith, or any other irrational source, are no different from beliefs that meet the highest standards of epistemic warrant. Religious belief is just as much real belief as scientific belief is: rational justification is externalto the nature of belief as such. This is why belief can never be sufficient for knowledge, no matter what kind of belief it may be—external conditions need to be superadded. If you were to pluck a belief from a context of wishful thinking and provide it with a proper justification, you would end up with knowledge, since no (true) belief is inherently prohibited form counting as knowledge; it all depends on further extrinsic conditions. Belief itself is knowledge-neutral, rationality-indifferent. That is why knowledge must be analyzed as belief plus truth and justification. To put it differently, belief is a natural psychological kind that crops up uniformly in contexts both rational and irrational. The psychological context never fixes the very nature of the state in question.

            But why should we adopt this view? Why can’t the history of a belief operate to fix its identity? Why can’t rational belief be a different kind of thing from irrational belief? Plato contrasted knowledge and opinion, assuming that these were antithetical types of mental state; knowledge isn’t opinion plus some. It isn’t opinion at all—it’s knowledge. When reason operates to produce knowledge it is operating in a way that generates a distinctive kind of psychological state, not merely the kind of state that can be produced by forces that are the opposite of reason—the kind that generally lead to error, ignorance, insanity, and sheer stupidity. Stupid belief can never be the core of genuine knowledge. The history of a belief shapes the belief; it isn’t a free-floating mental atom capable of any old kind of history. The belief is embedded in a psychological formation, and its nature reflects this embedding. The embedding can be rational or irrational, so that we get two kinds of beliefs not a single kind that can be supplemented in various ways. It is like sex in animals: the animal can be male or female, but it can never be neutral between these two options. Intuitively, a rational belief that counts as knowledge provides insight into the world—a certain transparent connection to reality—while irrational belief is caught up in the inner workings of the psyche (wishes, fears, neuroses). So knowledge is not rightly conceived as a composite of a rationality-neutral element called “belief” plus the extra conditions of truth and justification. The belief is already imbued with rationality and in paradigm cases with truth: it is a case of rational-belief (note the hyphen) not belief that is made rational by its contingent history or context. It would be good to have a word for this kind of belief (analogous to “mare” or “vixen”) but in fact no such word seems to exist; we just speak indiscriminately of “belief” (or “opinion”, “conviction”, and “commitment”). Of course, it is true that both sorts of psychological state share certain important characteristics, such as a connection to action and an inner feeling of being persuaded; but that shouldn’t stop us from registering the important respect in which they differ—namely, their connection to the rational faculties. There is really all the difference in the world between a belief that results from rash and foolish fantasy and a belief that results from carefully considered rational judgment; indeed, we should wonder why we choose to neglect this distinction in our ordinary talk of “belief”. The term is far too broad, far too inclusive. It distorts our conception of knowledge to suppose that knowledge can incorporate any old type of belief, as if even the wildest belief could find a home inside an instance of knowledge. In fact knowledge can only contain beliefs of the right sort (that have the right stuff)—the pure and noble kind, as we might say. Knowledge comprises beliefs only of the “knowledgey” kind, if I may be excused the adverbial neologism.

            This is why we have mixed reactions to certain sorts of possible case, e.g. a person who forms a belief by sheer wishful thinking but who subsequently comes across some supporting evidence for her belief. Does such a person really know the proposition in question? What if the evidence plays no causal role in sustaining the belief, the wish carrying all the weight? In such a case an attribution of knowledge seems suspect simply because the subject is not rationally motivated: she has the wrong kind of belief to form the core of the state of knowing; she would believe the same thing even if no evidence had ever come into her possession. We want the would-be knower to form her beliefs by a rational procedure, i.e. to have rational-beliefs. It is a sad fact that beliefs can depart from this admirable norm, but in attributions of knowledge we require that the belief in question should measure up to certain rational standards—and not just have a rational etiology but also have a rational nature. We want beliefs that have rationality built into them. This is why we accord a special kind of respect to beliefs that cannot fail to count as knowledge, such as beliefs about one’s own mind or elementary logic or anything else that admits of certainty. And the more likely it is for a belief to amount to knowledge the more value it has for us—the better it is qua belief. Ideally, we would like to have all our beliefs to be so rational that knowledge is guaranteed, so that no beliefs could be of the defective kind exemplified by wishful thinking. Then we could say simply that knowledge is to be analyzed as belief in that sense—once you are in that psychological state you are automatically in a state of knowledge. Things get complicated only because we are also capable of false and unjustified beliefs; but that is our fault, so to speak, not a consequence of the nature of knowledge as such. Knowledge itself is really just belief—of a special sort (the “knowledgey” sort). There is a type of belief that is such that anything of that type will be a case of knowledge—in this sense knowing is a psychological fact. We could analyze “x knows that p” as “x has a belief that p of type T”, where “T” describes the special kind of belief I am referring to. Knowledge is thus not a composite of some perfectly general state of belief and certain extra non-psychological conditions; psychology should be expanded to allow for a special class of belief states.[1]

            The spirit of this view might remind us of so-called disjunctive views of perceptual experience. The thought here is that there is no common psychological element between a veridical and a hallucinatory experience: there are two types of state that we call by the same name, viz. “experience”—there is nothing like a sense- datum that is shared by both types of so-called experience. Similarly, there is no unitary state called “belief” as between rational belief and irrational belief—or, more cautiously, it is wrong to assume a single natural kind denoted by the word “belief”. Rather, very different kinds of state fall under the umbrella term “belief” (for intelligible reasons), and we should firmly distinguish these states. The state of belief of the wishful thinking religionist is really quite different from that of the evidence-guided scientist: the former is in a psychological state of a different type from that of the latter (one might be tempted to call it “make-believe”). The religionist is oblivious to evidence and rational argument; the scientist is exquisitely sensitive to such considerations. The will is involved in the former but not the latter (the “leap of faith”). So we might favor a disjunctive view of the concept of belief: so-called beliefs can be either of the fantasy-driven type or of the reason-driven type–and these types are different in their history, their hold on the mind, and their functional characteristics (though similar in certain ways). We do well to mark the difference plainly and not assume a deep similarity or identity. There are good beliefs and bad ones, impostor beliefs and the genuine article (cf. fool’s gold and real gold). We shouldn’t be misled by the superficial form of our talk about belief (as we shouldn’t be about our talk of desire[2]) into assuming more similarity than there actually is. In particular, we shouldn’t run away with the idea that knowledge is a compound state with a belief constituent that is exactly the same as that which can occur completely outside of cases of knowledge or even elementary rationality. The reasons for belief shape the nature of belief; whether a belief is justified affects the inner character of the belief. Thus our picture of knowledge changes once we recognize that knowledge springs from a particular type of belief: now we see it as more unified than it would be if composed of disparate elements combined together—as if the belief element had nothing intrinsically to do with knowledge until brought into proximity with the other elements. The belief element (if that is the word) is already steeped in justification and hence truth-oriented; it doesn’t need a dollop of these ingredients drawn from elsewhere. Knowledge is really part of the psychological reality of belief (of the type that can lead to knowledge anyway). We could even say that knowledge just is belief—once we appreciate the true nature of belief. There is no distinction between knowledge and belief when it comes to certain subject matters (e.g. one’s own conscious thoughts), and the case is not that different in cases where fallibility is a possibility. At any rate, beliefs that count as knowledge are intrinsically cut out for the job. Knowledge is therefore a more closely knit phenomenon than we have tended to suppose, less conjunctive in its essence. It is not a tripartite thing.[3]

[1] We could compare this position to externalism about mental content: psychology needs to be expanded to include externally individuated content and not be restricted to content narrowly individuated. Similarly, beliefs need to be seen as incorporating their context within the mind, with rationality entering into their nature (or not as the case may be). 

[2] We can speak of moral desires but it would obviously be quite wrong to assimilate these to other sorts of desire.

[3] Even if the concept of knowledge were a tripartite thing, consisting of three separate and independent concepts, it would not follow that knowledge itself (the phenomenon) is a tripartite thing; ontologically, knowledge can be regarded as essentially unitary. The gap between knowledge and belief is not as large as traditional thinking suggests.