Introduction to the special issue “The roles of experience in a priori knowledge”

According to the influential Kantian tradition, the distinction between the a priori and the a posteriori is drawn by looking at different roles experience plays or does not play in determining the epistemic status of justified belief. Kant himself introduces his distinction between the a priori and the a posteriori as a distinction between two kinds of judgments—that is, two kinds of essentially propositional cognitions—in the opening paragraph of the Critique of Pure Reason:

Experience is without doubt the first product that our understanding brings forth as it works on the raw material of sensible sensations. (...) It tells us, to be sure, what is, but never that it must necessarily be thus and not otherwise. For that very reason it gives us no true universality, and reason, which is so desirous of this kind of cognitions, is more stimulated than satisfied by it. Now such universal cognitions, which at the same time have the character of inner necessity, must be clear and certain for themselves, independently of experience; hence one calls them a priori cognitions: whereas that which is merely borrowed from experience is, as it is put, cognized only a posteriori, or empirically.¹

Nowadays, the distinction is mostly drawn as a distinction between two types of epistemic support that beliefs enjoy, or in other words between two types of doxastic justification. Inspired by Kant’s own remarks, a version of the Kantian distinction is now the orthodox one that we find accepted as background in most epistemological contexts within analytic philosophy, or else taken as the subject of debate:

Orthodox Kantian A Priori/A Posteriori Distinction

A subject S has an a posteriori justified belief B if and only if the justification on which B is based is dependent on experience. Conversely, a subject S has an a priori justified belief B if and only if the justification on which B is based is not dependent on experience.

Philosophers’ interest in the distinction is largely due to their interest in the possibility, extent and nature of a priori justified beliefs. While it is accepted by all but the skeptic that we gain substantial amounts of knowledge about the world by exercising our perceptual capacities, it is not universally accepted that any interesting pieces of knowledge are justified purely by non-experiential sources or non-experiential reasons. The best candidates are often identified in logic and mathematics. But like Kant, many accept that the possibility of interesting distinctively philosophical knowledge depends on the possibility of interesting a priori justified beliefs. Unfortunately, the orthodox characterization of the a priori is an entirely negative one. It primarily describes what it is for a belief to be justified a posteriori, and characterizes a priori justification merely as justification that is not like this.

Given this formulation, two important questions need to be answered by a more comprehensive and illuminating account of a priori justification: First, which roles for experience in the acquisition of doxastic justification amount to epistemic dependence in the relevant sense? And second, what is meant by “experience” such that justification must be independent of experience to qualify as a priori? Let us call the first question the Dependence Question, and the second question the Experience Question. One can quickly see that the answers to both questions matter enormously for our understanding of the a priori/a posteriori distinction. Consider first the Dependence Question. A subject cannot have a justified belief that p if she cannot form the belief that p. In order to form the belief that p, the subject needs to possess the relevant concepts. One standard way to acquire many concepts is to learn a natural language in which they can be expressed, and this is typically accomplished during childhood through interaction with competent speakers of the language. Such interaction crucially involves the sensory perception of their verbal behavior. If we take reliance on sense experience of this sort as sufficient to render any belief that employs the learned concepts a posteriori, then we pretty much rule out a priori justified beliefs in a trivial way. Next, consider the Experience Question. If “experience” is understood in a very wide sense as referring to any kind of conscious mental state, then the fact that we often rely on conscious reasoning processes when we run through mathematical proofs suffices for the mathematical beliefs justified in this way to count as a posteriori justified. Again, there is little interest in an a priori/a posteriori distinction drawn in this way. If “experience” is understood in a very narrow sense as the deliverances of the five senses, then introspective beliefs and beliefs based on episodic memory come out as justified a priori (if justified at all). But intuitively—as many philosophers agree—this verdict is wrong. They are paradigmatically empirical beliefs about how things happen to be in the actual world at a particular time and place.² So, the difficulty in formulating a defensible and interesting distinction between a priori and a posteriori justification depends crucially on giving plausible answers to both the Dependence and the Experience Questions that cohere with our clear intuitions about classification of paradigmatic cases.³

We have a plausible—though not undisputed—partial answer to the Experience Question: whatever else is included in experience in the relevant sense, ordinary sense experience—that is, the deliverances of the five sensory capacities—should definitely be included. We also have plausibly—though again not undisputedly—made some progress on the Dependence Question: it is now widely recognized that we need to distinguish between experience playing a merely enabling role for a piece of knowledge from experience playing a justificatory or entitling role for a piece of knowledge.⁴

Nevertheless, skepticism about the a priori/a posteriori distinction is alive and well. For example, Timothy Williamson has recently argued that the distinction is coherent, but of little significance because it does not cut at any deep epistemic joints.⁵ And John Hawthorne has tried to show, assuming basic externalist epistemological commitments, that the distinction is not a natural one, because the classification of specific justified beliefs as justified a priori or a posteriori depends on a superficial and mostly arbitrary choice regarding which belief-forming processes to look at when assessing the belief for its epistemic status.⁶ Contemporary criticisms like these can be understood as targeting the Dependence and Experience Question. What else besides sense experience should be capable of affecting the a priori status of justification: introspection, imagination, or cognitive experience such as intuition? Or maybe any kind of experience after all? And how exactly should we understand the intuitively natural distinction between enabling and justifying conditions? Or should we even reject or significantly revise it?

In this special issue, four experts on the a priori give partial answers to these questions by looking at the different roles experience plays in a priori knowledge. Here I would like to present a guide map to the contributions in this special issue. I will not summarize the content of the four papers, as the authors have already done a better job of it in their abstracts than I could do. But I will highlight some ways in which each paper contributes to the elucidation of the roles that experience does and does not, and can and cannot, play in a priori knowledge.

Albert Casullo’s contribution Four Challenges to the A Priori—A Posteriori Distinction presents a thorough taxonomy of the most important recent and contemporary challenges to the distinction. But Casullo does more than provide us with a useful map to the problems and answers of the debate about a priori justification and a priori knowledge. He shows that the coherence of the Kantian distinction stands or falls with the coherence of the distinction between the experiential and the non-experiential, and argues that this distinction is extraordinarily hard to draw. A demarcation by means of simply enumerating cognitive capacities that count as experiential capacities is likely to leave people wondering whether any unifying reason stands behind and explains the enumeration. But according to Casullo, previous attempts to demarcate experience by referring to phenomenological features of the relevant mental states are unsatisfactory, as are attempts to demarcate experience in terms of the contents of the beliefs it justifies, its objects or the relations the subjects of experience stand in to those objects. Casullo’s conclusion is not skeptical of the experiential/non-experiential distinction, but he draws attention to the challenges that have to be met when articulating a defensible conception of experience.

In their contribution The Traditional Conception of the A Priori Carrie Ichikawa Jenkins and Masashi Kasaki focus on the Dependence Question more than the Experience Question. While Williamson aims to use cases involving the cognitive capacity of imagination to argue that there is no deep distinction between a priori and a posteriori justified beliefs, Jenkins and Kasaki show that reflection on cases involving the imagination that is inspired by Kant’s thoughts on the subject can lead us to a different diagnosis of the situation. They argue that a two-way distinction between enabling and evidential roles for experience in the process of acquiring knowledge fails to acknowledge the possibility that experience might play an epistemic role for a given piece of knowledge that is yet not evidential. It can provide the “necessary epistemic backdrop” for the justification of a belief without thereby being constitutive of this justification. As long as the role of experience is not evidential, they claim, the resulting belief is interestingly different from classic a posteriori pieces of knowledge and should count as a priori. On the account they sketch, experiences such as imaginings, rememberings and even perceivings can sometimes serve as the background whose positive epistemic status is a necessary (epistemic, but non-evidential) condition for being a priori justified in believing the conclusion of a piece of reasoning.

In How Reliabilism Saves the A Priori/A Posteriori Distinction Thomas Grundmann shares Jenkins’ and Kasaki’s resistance to Williamson’s skeptical arguments, but he does so on different grounds. For Grundmann, the key to finding a satisfactory answer to the Dependence Question and thereby saving the a priori/a posteriori distinction from serious doubts lies in the choice of epistemological theoretical frameworks. In his opinion, the evidentialist framework on which justification is understood as evidence cannot adequately capture our intuitions about the cases we should classify as justified a priori and the cases we should classify as justified a posteriori. More precisely, the evidentialist reaches intuitively incorrect verdicts about which cases are ones in which experience plays a merely enabling role, and which cases are ones in which it plays a justifying role. According to Grundmann, the process reliabilist is in a much better position: The reliabilist classifies those justified beliefs as a priori that result from reliable terminal processes that are not only non-experiential, but whose reliability does not depend on the reliability of a relevant input-process that is sense experiential. This classification matches our intuitions about the cases of justified belief in which experience plays a role that is non-justificatory and thereby compatible with a priority, and the cases in which experience plays a justificatory role. This gives us a way to defend the distinction against Williamsonian and Hawthornian criticisms.

Declan Smithies’ contribution Ideal Rationality and Logical Omniscience presents an account of a priori justification for logical beliefs on which the source of justification for these beliefs are the facts about logic themselves, rather than psychological facts about experience, reasoning or understanding. Smithies argues that this account is needed to capture the fact that rationality proper requires logical omniscience. On his account of a priori justification there is no sense in which any kind of experience, whether perceptual, memorial, intuitive, or understanding-based, ever plays anything but an enabling or disabling role for a priori knowledge, because experiences of any kind are just not the type of thing that justifies beliefs about logic and similar matters. This is what distinguishes the epistemic support these beliefs enjoy from the epistemic support of empirical claims. But the enabling role of experience on Smithies’ account is rich. It does not merely allow us to grasp the propositions believed, but also provides—in Smithies’ own terms—whatever else is needed to convert propositional justification about logical matters into doxastic justification.

There is a background story to this special issue: In September 2012, Thomas Grundmann and I co-organized a workshop-style conference on The Roles of Experience in A Priori Knowledge at the University of Cologne in Germany. The conference included eight presentations of original new research on the topic by George Bealer, Albert Casullo, Marcus Giaquinto, Thomas Grundmann, John Hawthorne, Carrie Ichikawa Jenkins and Masashi Kasaki, Nenad Miscevic, and Declan Smithies, followed by in-depth discussion sessions. The four papers that comprise this special issue were developed out of four of the presentations at the workshop. The guest co-editors would like to thank all the conference participants for creating a philosophically rich and stimulating event. The conference was mostly funded through the DFG Emmy Noether Research Group on “Understanding and the A Priori” led by Brendan Balcerak Jackson and me from 2009 to 2015. But we would also like to thank the German Association for Analytic Philosophy (GAP) for their generous support of the conference. Most importantly, thanks are due to all the referees who helped us select and greatly improve the papers you can now read in this special issue. Due to many complicated circumstances, it took us more time than we had anticipated and than we would have liked to complete work on this special issue. Thanks to the editors of Synthese, and in particular to Gila Sher, for their patience and their support.