Laura Soter
Affiliation: York University
Category: Philosophy
Keywords: belief, doxastic voluntarism, emotions, process models, cognitive science
Date: Thursday 4th of September
Time: 15:00
Location: Room 161 (161)
View the full session: Concepts & Propositional Attitudes
In this project, I aim to propose a new way of thinking about the nature of belief from a cognitive scientific perspective. The core idea is that instead of thinking of belief as a “state,” we should think of it as a dynamic, componential cognitive process.
The motivation for this approach comes from looking to cognitive scientific approaches to other kinds of mental states. Emotions are a paradigmatic example. Though we often talk loosely of emotions as mental “states,” psychologists really model them as dynamic psychological processes that unfold over time and involve the coordination of distinct cognitive components, which iteratively influence each other throughout the emotion-process (Scherer, 2009). On this view, there is not single privileged component of the process which “really is” “the emotion itself”—rather, this approach reconceptualizes the psychological reality of emotions as the whole process. This componential model of emotions is a dominant approach among contemporary emotion scientists (even if there is disagreement among theorists as to which components are most important especially for differentiation of emotion kinds).
My methodological proposal is that perhaps we could aim for the same kind of thing to understand belief: perhaps we can model belief as a dynamic, componential psychological process. The aim in developing this process model of belief would be to identify the different components of the belief-process and the mechanistic relations among those components, and the many different ways the process can be instantiated. On this view, there would be no ground truth answer regarding which component of the process “really is” “the belief itself:” it invites us to move past thinking of belief as some kind of static state, to a temporally extended, psychologically rich process.
Why might we want to do this? I’ll argue that a process model of belief, if successful, could have (at least) four significant benefits: (1) Psychological plausibility: Quilty-Dunn and Mandelbaum (2018) write, “Beliefs are stored in the mind the way episodic and semantic memories are” (emphasis added). But memory science has moved far beyond thinking of memories as statically stored; instead, memory is now widely accepted to be a reconstructive process. From emotion to memory, many developments in cognitive science have involved rethinking seemingly static phenomena as dynamic processes; it thus may be fruitful to consider whether belief is amenable to the same treatment. (2) Enabling empirical investigation: understanding exactly what components constitute the belief-process, their different possible instantiations, and their relations to each other will ultimately be a heavily empirical process. An obstacle for developing a systemic science of belief has been widespread unclarity about exactly what a belief is, or what counts as measuring belief. The process model offers a conceptual framework for pursuing empirical cognitive-scientific investigation of belief, while dodging classic worries about whether psychological investigation is “really measuring” “belief itself.” It encourages us to develop mechanistically specific, detailed predictions about the components of the belief process and their various relations and possible instantiations, and seek to test them—which can enable iterative theory revision of the kind we have seen in emotion science (Moors 2024). (3) Theory integration: it may turn out that some existing debates about the nature of belief can be resolved by observing that they map onto different parts of the belief-process. Scherer (2022) has argued that theory integration is one of the primary benefits of the componential model of emotions; I will show several ways in which this may offer similar benefits for debates about belief, including debates about representationalism vs. dispositionalism (Quilty-Dunn & Mandelbaum 2018; Schwitzgebel 2002), and about how many kinds of “belief” attitude there are (Van Leeuwen forthcoming). (4) A doxastic regulation/control framework: the science of emotion regulation has been a hugely successful scientific endeavor, and has capitalized on the componential model of emotions by observing that we have (and lack) the capacity to intervene in different ways at different parts of the emotion-process (Gross, 2015). The process model may afford a similar opportunity for belief, allowing us to develop a systematic framework to understand the capacities and limits of doxastic control—and take debates about doxastic voluntarism out of the philosophical armchair and into the realm of empirical(ly informed) cognitive science. This will also reveal that standard voluntarism questions about whether belief is under direct voluntary control are woefully underspecified: we should instead be striving to understand the ways in which various components of the belief-process are and are not amenable to agential intervention, and in what ways and to what extent.
Finally, having motivated the potential theoretical benefits of the PMB approach, I’ll close with a sketch of a first-pass attempt of drawing out a candidate process model, shown in Figure 1. The goal of putting forward my first-pass model will not be to defend it as the one we should expect to settle on, but instead as an initial attempt that will inevitably call for much editing. I’ll point to some good candidates for initial components the process model of belief should aim to capture. I’ll then offer some initial thoughts about the productivity of this starting point. I’ll show how different theories of belief might suggest different modifications; e.g., Spinozans vs. Cartesians (Mandelbaum, 2014; Sperber et al., 2010) about belief-fixation may disagree about where the appraisal component should sit. I’ll highlight how isolating components can allow for more precise discussion about specific functions, such as whether the appraisal function is most plausibly only evidence/truth-directed, or whether it could also capture social-affiliative goals. And I’ll highlight how the process model can allow us to scaffold a framework for doxastic control, arguing that classic involuntarism can be understood as identifying a particular link in the process where we likely lack direct agential control, but that we can specify the capacity for intervention at many other places in the model.