RESPONSIBLE BELIEVING

S. Joel Garver

CHAPTER FOUR

Seeing Isn't Believing:
(And Other Thoughts on Doxastic Practices)

Section III

Presuppositions and Groundless Beliefs

Before leaving behind Millar's views, one more kind of case he discusses ought to be mentioned--those that involve groundless beliefs. These groundless beliefs often function as what he calls the "normative propositions" of forms of (quasi-)inference and or conceptual frameworks. I tend to call these entities "presuppositions." Millar gives the example that one can justifiably believe that someone wants to see you based upon the fact that someone is knocking on your door. This belief involves a move from a (knocking sound coming from the region of one's office door)-type experience to the conclusion that someone wants to see one. According to Millar this move presupposes a generalization that is expressed in the "belief that the occurrence of knocking sounds coming from the region of one's office-door is strong fallible evidence that someone wants to see one" (1991:177).

This belief, he suggests, can be rightly held even if one is "unable to recall any specific occasions in which a knocking sound at the door herald[ed] someone wishing to see you" (1991:176). It might not be possible to test the generalization in any way since you lack the appropriate memories. In fact, you are probably more confident of the truth of the generalization than you are of any memories you might have to support it. Millar urges a picture in which we come to take the generalization to be

The point is not that every practice of this sort is acquired on one's own in isolation from others engaging that practice. Such practices can be, and surely are, inculcated by others--parents, teachers, peers, and so on. Rather, the point is that there are not always clear chains of justification running from certain experiences to the belief in the generalization. No one instance could justify the belief and no memory of multiple instances exists in most cases.

The generalization, then, is groundless, on Millar's account, and functions as what he calls a "normative proposition," an entity much like those presuppositions or language game-rules that Wittgenstein invokes in On Certainty (1969: 95, 96, 167, etc.; the numbers in reference to On Certainty are paragraph numbers). While this groundlessness does not amount to irrationality or illegitimacy, Millar sees this groundlessness as akin to bare assumption or taking a proposition for granted. Furthermore, if such groundless beliefs play the role that Millar suggests, we have a category of belief that is neither immediately nor mediately justified and so a foundationalist account of justification is threatened.

I want to stop here for a moment and question Millar's account of the knock on the door. I think he is quite right regarding the etiology of our practice of telling that someone wants to see us from knocks on doors (let's call this practice a K-inference). Nevertheless, a couple of points can be questioned. First, I think it is far from clear that the practice of making K-inferences necessarily involves the employment of the belief that K-inferences are valid (that is, the belief or generalization that knockings of the right sort are strong fallible evidence for someone's wanting to see you; we are not thinking here of strictly logical validity). When a K-inference is made, it is made as a matter of practice or intellectual habit or, one might say, "automatically." It is not necessarily made in response to or as a conscious application of any occurent belief to the effect that such inferences are valid.

Second, this is not to say that most people who make K-inferences do not believe them to be valid. If asked, surely most would agree that K-inferences can be rightly made and that, yes, they have for quite some time believed as much. Nevertheless, that does not mean that the belief was employed in making the inference. Introspectively, at least, such a claim does not often ring true.

Third, this picture might suggest something about the grounds of the generalization that Millar took to be groundless. Perhaps it is the case that the practice--the actual habit of making K-inferences--serves as the ground for the generalization concerning the validity of K-inferences. After all, the practice itself was rightly formed by actual cases, proves to be truth-conducive, and has great utility in maneuvering through social space. The practice also presupposes the truth of the generalization--that is to say, by utilizing K-inferences as justification-conferring (or knowledge-producing) the doxastic practice logically requires the truth of the generalization that K-inferences are valid. If that is so, then the belief expressed by the generalization may not, in fact, be groundless, but rather be grounded in and emergent from the practice as it was formed.

As a game-rule or presupposition, then, the generalization governs the practice and its truth is logically prior to the practice and accounts for its reliability. As a belief, however, held by the subject engaging in the practice, its formation and status as justified can be seen as causally and logically posterior to the practice and to be accounted for by the practice's reliability. The generalization, then, is reliably inferred or quasi-inferred from the experienced shape and usefulness of the truth-conducive practice itself.

Of course, not every generalization of the sort that validates K-inferences need be formed in this way. One could, for instance, come to justifiably believe such a generalization on the basis of the testimony of someone else--especially as one is learning from another how to process information in a specialized field or discipline. Nevertheless, I would suggest, generalizations of that sort do regularly and fundamentally emerge in the way I have sketched.

If my account has any weakness, it is not, I think, so much a matter of wrongly identifying the origin of some of Millar's "groundless beliefs." Their origins are in the practices they validate. It is instead a matter of infelicitously filling in how the beliefs emerge from their origins. One might urge that the process of belief-formation for such generalizations is so unconscious, so inevitable, and so automatic, that we must avoid any talk of grounds as if they were rational bases for belief or the inputs of inferential patterns. It is admissible, of course, that the generalizations are reliably caused, have justification or warrant (or other positive assessment), but, one might suppose, they cannot be said to have grounds. In this respect, the objection goes, they are like "fact memory."

"Fact memory" is to be contrasted with "event memory." In event memory we remember something that has happened to us and facts emerging from that event by calling up certain mental experiences--a replay of the past--and reporting their contents. In this way I remember visiting England's Lake District and that Wordsworth had lived there in the house we toured. By way of contrast, fact memory involves recalling facts, straight up, so to speak, without any accompanying mental replay of past experience. In this way I remember that Lusaka is a city in Zambia, a country in southern Africa.

One might suggest that beliefs emerging from event memory are based on the mental replays of past experience. Thus they have grounds. Fact memory, on the other hand, lacks any necessary mental replay and, while it produces justified or warranted or reliable beliefs, it does not do so on the basis of grounds (i.e., beliefs or experiences that serve as rational bases for belief). No form of inference or quasi-inference is employed in fact memory. Fact memory is, then, rightly caused or reliably formed, but this does not amount to having grounds. Thus, fact memory is groundless.

This line of thought might provide another account for the groundless generalizations that Millar invokes. Such generalizations are not, on this alternative account, grounded (by means of inference or quasi-inference) in the belief-forming practices of which they serve as presuppositions. Rather, even while the practice of, say, K-inference, is itself being formed on the basis of experiences, the mind may be busy reliably committing to memory facts from these experiences. For instance, the mind may commit to memory the fact that a K-inference (though not yet a true habit or practice) was successfully employed or would have yielded true belief had it been employed. Once this fact has been adduced, actual event memory is dispensable. These facts in turn together cause a new element to be added to one's fact memory--that K-inferences always or for the most part have worked or would have worked. This element of fact memory amounts to a reliably formed possession of the generalization or presupposition in question.

Thus, on this alternative account, Millar's point that there are no specific event memories behind the generalizations is accurate and consistent with the etiology of such generalizations. My earlier suggestion that such generalizations originate in the practices themselves is also correct. What is different, however, is that on this alternative account the positive epistemic status of the generalizations is not attributed to some sort of inferential ability that moves from practices to generalizations. Rather, such generalizations are rightly and reliably caused and so count as justified or warranted (or otherwise positively assessed) belief, though groundless.

This suggestion has some force and instances of it may actually obtain in some cases of these generalized beliefs. If that is so, however, it is important to note that it succeeds by appealing to a form of fairly pure reliablism concerning certain classes of beliefs. As such it is consistent with a modified foundationalism that allows reliably formed groundless beliefs of certain classes to serve among the foundations of knowledge along with immediately justified beliefs. It does not even tend towards any sort of coherentism or real alternative to foundationalism. But does this alternative account serve to explain all cases of generalized and presuppositional belief? I think not.

Keep in mind that the central issue is whether our beliefs in these generalizations are simply reliably caused or whether they are based on practices in some more inferential way. Recall the case of fact memory. The appeal of a purely reliablist account in respect to fact memory stems, I think, largely from our inability to ever point to any event memory that would identify the origin or serve as grounds for the belief produced by fact memory. That is to say, fact memory is taken to be groundless since grounded beliefs always or for the most part admit of a certain accessibility to their grounds. Thus, my earlier account of how we come to believe the presuppositions or generalizations can be made more plausible if we do, in fact, take such generalizations to be grounded and can point to their grounds.

Consider the following by way of evidence in favor of the groundedness of presuppositional generalizations where the grounds consist in the experience of the practice itself. The earlier account seems to be consistent with instances in which the generalized beliefs (though often dormant), do seem to emerge only after the practice has been formed and successfully engaged for some time. It also accounts for the instances in which we feel that we come to hold such generalizations on the basis of our "seeing" what we have been doing.

My earlier account takes stock of the pragmatics of these generalizations. We believe them, I think we would say, because they are presupposed by what we do in belief-formation and what we do works well.

Less pragmatically, my earlier account helps explain the fondness for transcendental arguments in philosophical epistemology. We account for our belief in the generalizations by an appeal to their truth as the necessary precondition for the knowledge we believe we possess. That is to say, the practices presuppose them.

All in all, these factors point to the idea that we take practices to be the grounds of the generalizations and, thus, we do not take the generalizations to be ungrounded. This does not amount to a demonstration, but it carries weight.

Thus Millar may be incorrect to take generalizations (normative beliefs) to be regularly groundless and, even if correct, he is wrong to take that point to be inimical to foundationalism (since if groundless, they may be acquired reliably). It remains to be seen whether Millar (and Wittgenstein) are similarly incorrect about the groundlessness of more general and global presuppositions (e.g., that there is an external world). It seems to me, however, that they are incorrect. Consider the proposition that there is an external world. Now many of our belief-forming habits (forms of inference and quasi-inference and their attendant conceptual framework), presuppose that there is an external world. In fact, it is probably necessary for these habits to presuppose this in order for us to get through life. The success of these habits in getting us through life, in producing what we take to be justified belief and knowledge, and the generally well-established (coherent, self-supporting) nature of perceptual practice (as well as its mutual support with wider spheres of doxastic and non-doxastic practices), all taken together presuppose the existence of the external world. In various ways, then, they ground our belief in an external world, and, I would suggest, justify that belief. I am confident that such a picture may be constructed.

The stage is almost set to turn to Foucault's account of doxastic practices. Before turning to that matter, however, let us take a detour through Hume and some other practices.