Introduction

First, some history. In December 2019, one of us—Jonathan Lewis-Jong—began inviting philosophers and psychologists to meet up for a conference on “the nature of belief,” to be held in Oxford in the summer of 2020. The COVID-19 pandemic made this impossible, and so the Oxford Belief Workshops moved online. A total of thirty-two psychologists and philosophers participated in the conversations, among whom were Eric Schwitzgebel, Tania Lombrozo, and Neil Van Leeuwen, whom Lewis-Jong invited to collaborate on this topic. Lombrozo and Van Leeuwen ended up running an online seminar series from October 2021 to July 2022 (https://cognition.princeton.edu/belief) and an in-person conference at Princeton in June 2022. Meanwhile, Schwitzgebel and Lewis-Jong worked on this volume, which started as an essay competition, advertised on Schwitzgebel’s blog, The Splintered Mind. Applicants were challenged to provide an answer to the question “What is belief?” We received sixty-four submissions, and—with some difficulty, as many were very good—invited thirteen to submit full papers. These thirteen papers are joined in this volume by one each from Eric and Neil: Each chapter represents a different answer to the question. The diversity of responses is our evidence that a book like this is sorely needed, and that our question—What is belief?—remains a good one, and one without a consensus answer. But it might be worth our saying something about why we think it is an important question. Coming from different disciplinary perspectives, we have slightly different answers to the question. Lewis-Jong is an experimental psychologist, whose work is mainly on religion and specifically on religious beliefs. As a psychologist, he spent a lot of his time and efforts trying to figure out how to measure religiosity, and religious beliefs in particular. And as measurement is simply the practical manifestation of definition, this soon led him to wonder what religious beliefs are, which in turn led him to wonder what beliefs are more generally. As soon as he posed the question, he noticed something peculiar about psychology, which is that (1) there is lot of research that goes on that seems to be research about beliefs of some kind, and (2) very little of this research uses the term “belief” as a theoretical term. There are, for example, active psychological research programs on declarative memory and cognitive schemata and stereotypical folk intuitions in various domains, including the physical, psychological, biological, and moral. And, if we were to cast our nets further, other social scientists are eager to survey things called “views” and “opinions.” These all seem, at first blush, to be about beliefs, on a common-sense understanding of the term. But cognitive psychologists are more likely to talk about these things in terms of knowledge, by which they certainly do not mean “justified true belief.” It is worth considering a counterexample to this reticence to talk about belief. The American Psychological Association Diagnostic and Statistical Manual uses the term “belief” and its cognates over two hundred times, especially but not exclusively in its discussion on delusions. Incidentally, twenty-one occurrences of “belief” in the DSM-5 are in the Technical Glossary, which conspicuously neglects to define “belief” itself. To complicate things further, psychology has had a history of theorizing about and attempting to investigate the unconscious. Freud’s obsession is with us still, transmuted into research on tacit knowledge, automaticity, and implicit attitudes. Lewis-Jong’s own work has involved indirect and so-called implicit measures of attitudes, the best known of which is the Implicit Association Test (see Jong et al. 2017 for a review). It has never been clear what these measures actually measure, and how whatever they measure (if anything) relates to whatever it is that traditional self-report questionnaires measure. All of which is to say that there are many things studied by psychologists and social scientists that could be called “beliefs,” but they rarely are. And regardless what we call them, substantive questions remain about how these different things—memories, schemata, intuitions, opinions, knowledge, delusions, and attitudes, implicit or otherwise—relate to one another. These are questions of ontology, and here Lewis-Jong will confess to having some physics envy: There is not anything like a universally accepted way of categorizing and characterizing the furniture of the mind. We have meandered far from Lewis-Jong’s initial concerns about measurement, but we should return there. Lewis-Jong’s interest in the nature of belief still revolves around measurement. He wants to know what there is to measure, for example. For any given proposition—say, “God exists”—is there only one thing to measure that we might call “belief”? Or are there two, as some imagine there to be implicit as well as explicit attitudes about the same target object. Or may there be more? Or is the entire question misguided? And he wants to know how best to measure whatever it is there is to measure. This is in part an empirical question, of course: the subject matter of psychometric research. But how we define “belief” will make a difference. Definitions that privilege, say, assent and assertion will emphasize self-report; those that privilege other behavioral dispositions will emphasize behavioral measures; and so forth. Consensus among psychologists about how to taxonomize mental states may well be beyond the realm of possibility; but conceptual resources to help with the design, implementation, and measurement of mental states need not be. Lewis-Jong favors a representationalist view of belief that is largely agnostic about the causes and consequences of belief, and even the phenomenology of belief as well as about its aims. On his view, questions about whether or not beliefs are responsive to evidence or whether they are amenable to volitional control or whether they govern behavior or whether they feel a certain way are all empirical questions, not to be settled at the point of definition. This leads him to disprefer theories of belief that define belief in terms of their dispositional profile. He therefore stands in the representationalist tradition when it comes to belief. The measurement of belief is the measurement of the content and strength of a cognitive representation. Having once argued that that beliefs are object-attribute associations of varying strength (Jong 2018), modeled after Russell Fazio’s (2007) definition of attitudes, he is now persuaded that beliefs are propositionally structured. However, the influence of associationism remains, especially in how he understands representational strength, which is in terms of association strength between a representation’s constituents and the accessibility of the representation. In this volume, Lewis-Jong’s own view is closest to Porot and Mandelbaum’s, though they are more committed to functionalism than he is. Schwitzgebel’s interest in belief grew from his interest in the philosophy of developmental psychology, in two separate ways. First, the philosopher Donald Davidson notoriously argued that belief is not possible among creatures who don’t understand that their own beliefs might be mistaken (Davidson 1984). Given the widely accepted view in developmental psychology that children don’t fully grasp the concept of false belief until about age four (Wellman et al. 2001), Davidson’s position seems to imply, bizarrely, that three-year-olds don’t have beliefs (Schwitzgebel 1997). The apparent sharp conflict between Davidson’s view and ordinary understandings of “belief” prompted Schwitzgebel to wonder what conceptual criteria were necessary for possession of belief. Second, knowledge seems sometimes to emerge gradually in development. Plausibly, infants and children do not learn in a flash, for example, that objects can continue to exist unperceived or that people can have mistaken beliefs. Understandings come together slowly and unstably, in a way that seems poorly captured by simple representational models of belief according to which a representation like “objects sometimes exist unperceived” or “beliefs can be false” is either stored or not stored in a “belief box” or in long-term memory (Schwitzgebel 1999; McGeer and Schwitzgebel 2006). These issues led Schwitzgebel to propose a “dispositional” approach to belief (Schwitzgebel 2002). On this view, inspired by the twentieth-century philosopher Gilbert Ryle (Ryle 1949), to believe is just to be disposed to act and react in a manner that ordinary belief ascribers would regard as characteristic of someone with that attitude, somewhat like having a personality trait such as extraversion is just to be disposed to act and react in a manner characteristic of an extravert. (For a more careful statement, see Schwitzgebel’s contribution to this volume.) Such patterns of action and reaction can develop slowly, in degrees. They need not require a stored representation with the same content as the belief. And they need not require sophisticated conceptual tools such as self-ascription or knowledge that beliefs can be mistaken. It seemed to Schwitzgebel that this view, by being minimalist about cognitive architecture, avoided a variety of puzzles that about what people “really believe” in cases not only of gradual learning but also of delusion, conceptual confusion, self-contradiction, implicit bias, partial forgetting, procedural knowledge, and so on. On Schwitzgebel’s view, once you have described the person’s dispositional structure, you’re done with the substance of the question. There need be no further fact about whether, “underneath it all,” they really believe or fail to believe. Schwitzgebel’s approach to belief stands in conflict with the dominant representationalist paradigm in philosophy of mind and cognitive science; with approaches that connect belief tightly with self-ascription of belief, or with feelings or judgments that some proposition is true, or with other demanding cognitive achievements; and with functionalist approaches that treat beliefs as states or entities that must have a certain causal history. And it’s unclear how comfortably it fits with “normative” approaches that treat beliefs as necessarily subject to norms of evidence, truth, or rationality. Philosophical disputes of this sort weave throughout these contributions. This book does not offer one single definitive theory or definition of belief. Rather, it displays the shape of philosophical disagreement about the nature of belief. It will surprise at most 0.3 readers that there are multiple ways of characterizing these philosophical differences. One approach is to first ask the question, “How many things are there that count as belief?” The most common answers are: zero, one, two, or rejection of the presupposition of the question. Other answers are possible—including, for example, twelve—but have not been developed. Zero represents a deflationist view that eliminates belief altogether, replacing it with something else: behavioral dispositions perhaps, or brain states. Some interpretivist, dispositionalist, or neuroscientific theories of belief might count as eliminativist, if they hold that interpretations, dispositions, or brain states exist while beliefs do not. There are no straightforwardly eliminativist views represented in this book, but Angela Mendelovici contends that beliefs are fictions, and Tim Crane and Katalin Farkas are also happy to call their view a fictionalist one. Their approaches differ, however. Mendelovici’s view is a close cousin of interpretivism—according to which to believe some proposition P is to behave in a manner that would make sense to an interpreter who ascribed to you the belief that P. In Mendelovici’s view, unlike most interpretativist views, the relevant interpreter is the believing agent herself rather than some other actual or idealized agent, and the “behavior” needn’t be outwardly observable. Although particular internal states might correspond to beliefs, such states are neither tantamount to nor necessary for belief. Crane and Farkas go somewhat further: Not only do they argue against there being a kind of internal state tantamount to belief but also they are skeptical about there being a distinct type of trait that corresponds to belief. They argue that “belief” is just a convenient fiction, a useful theoretical construct that stands in for a complex collection of psychological dispositions. It is more common to say that there is one thing—most often a kind of mental representation or a kind of psychological disposition—that deserves the name “belief.” This can be motivated by a methodological preference for parsimony, which seeks to unify diverse phenomena under as few categories as possible, or by the supposition that all the phenomena we are inclined to treat as beliefs will have some essential common core. As with all unification projects, the challenge is finding the unifying thread behind an apparent diversity of phenomena. Is it, for example, a “feeling of conviction” (Smithies)? A “motivational role” (Sullivan-Bissett)? A “commitment to truth” (Singh)? Most of the contributions to this volume implicitly or explicitly treat belief as a unified phenomenon. The view that there are two things we should call “belief” is most closely associated with social psychologists who posit two distinct processes and/or systems that involve two different kinds of representations, for example the conscious versus unconscious, explicit versus implicit, and propositional versus associative. But there is another, more recent, view that posits the existence of two types of mental state that both get called “belief.” In this volume, this view is represented by Neil Van Leeuwen’s distinction between “Mundane” and “Groupish” beliefs. On Van Leeuwen’s view, Mundane beliefs (e.g., this switch is connected to this light) and Groupish beliefs (e.g., God is Father, Son, and Holy Spirit) have sufficiently different functional roles that they should be considered different types of cognitive entity. Mundane beliefs, for example, are involuntarily responsive to evidence and generate truth-dependent action, while Groupish beliefs are voluntary and generate signals of group membership. Addressing Van Leeuwen’s earlier work on this topic, Sullivan-Bissett argues that the difference between mundane and religious belief is not so extreme as to prevent them from both being called types of belief. Accounts of belief also differ in their appeal to functional conditions, structural conditions, and normative conditions, as we will now explain. Beliefs might tend to be caused in certain types of ways, or they might tend to have certain sorts of effects. These are functional conditions on belief. Functional conditions can either be essential (so that mental states lacking those conditions necessarily do not qualify as beliefs) or typical (characteristic of belief, but not strictly necessary). Nanay, for example, describes three widely accepted functional roles of belief: that they are sensitive to evidence, that they feature in inferences, and that they guide action. Van Leeuwen’s Mundane versus Groupish distinction complicates the matter by positing social roles for beliefs alongside epistemic ones. Most of the chapters in this volume discuss sensitivity to evidence, and vary on how this condition is interpreted and applied to belief. Porot and Mandelbaum argue that belief is “dumb” (formed quickly and reflexively), “cold” (readily updated in accord with evidence, even evidence we dislike), and “cynical” (involves rejecting information that is implausible or dubious). The overall trend in this volume is in the direction of relaxing the criterion of strict rational responsiveness to evidence. According to Ganapini’s “minimialist threshold,” belief is rational in that when confronted with the is motivated to a different that may be but in some cases of belief, such as in delusions. Van Leeuwen made possible by his distinction between Mundane and Groupish The tend to be caused by evidence while the are caused by social the that beliefs are sensitive to evidence is associated with the that they are not amenable to volitional This is beliefs are to be and to be updated on evidence. Porot and who argue that our beliefs are very sensitive to evidence we as a characteristic of belief. challenge the argues that belief a volitional act of argues that the of control can as a Mendelovici’s theory that belief is does not explicitly the but it also does not to require these conditions, some require that beliefs have particular on or There is of this criterion in this volume, though functionalism does its in and that there can be that of the relevant functional that believing no difference to or dispositions but to what one has to and dispositions. on in that belief is that to as in a variety of possible Van Leeuwen argues that Mundane Belief Sullivan-Bissett out by on of belief as an and to the subject to behave in ways which would if the of the belief were Mundane Van Leeuwen, truth-dependent action, Groupish Beliefs and social Crane and and and Schwitzgebel argue that believing P necessarily being disposed to act and react in particular ways. For example, and belief being disposed to the information that P to guide and There are also other or of belief its or behavioral a view, that the dispositional belief that P in that P. Crane and fictionalist view also that conscious of or “the of belief.” Mendelovici argues that beliefs that P involve being disposed to P argues that beliefs be fully their content be fully out by the Some hold that believing stored for example, the view that which content is to to the types of functional described Porot and and treat belief as the and of and hold that beliefs are states or that but their view with views that hold that the is necessarily Mendelovici a view on which all attitudes, including belief, involve propositional Schwitzgebel argues in for a that posits no structural on belief, as as the is disposed to act and react as if P is the and in particular he to the that belief that P stored with the content P. Crane and Farkas the need for an internal structural criterion for belief, as as the target of can be modeled as having the belief, on the of the and the of a between and belief as over while that what people believe is what they explicitly Some normative conditions on belief. they the that believing necessarily being subject to certain norms or of or for example, both and about structure, and that belief should be in terms of a causal Rather, on view, to believe is to be in a position to and feel things in a of a whose would make them the concept of belief in patterns of and criteria of argues that someone as having beliefs that as being in the of that as to normative of and from this an according to which if someone that they must have the to their belief in to evidence. that believing that P a to the of to the of P representation of P as true to a normative of as representation of P as if P is not in fact Sullivan-Bissett argues that our of belief can have different such that some beliefs from to true beliefs while from that to useful beliefs cases when the is not on normative in belief: view in to the challenge of sense of in belief. argues that his of belief as a of why are subject to norms of even when they are not at all disposed to with such the of and in belief a type of normative according to which a state be a belief the with that belief to that There are many ways to and even taxonomize theories of belief. The is a one. and offer a more complex and one in their But we would also like to call to the of and in the chapters of this book, some explicit and some to the and more theory of belief is and on about Crane and Farkas two different one with models and the other with The is also a feature of Schwitzgebel’s which personality as and also two is like or like being in a particular of on in belief to in This on and has been a characteristic of the cognitive though this has not been It is of the of and theorizing about the a For the of the mind leads to and for it leads to for it should to