Precis of “From Human Reasoning to Belief”
Joshua Mugg
Many thanks to the Brains Blog and Dan Burnston for the opportunity to discuss my book with Devin Curry, Aliya Rumana, and Lesley Walker. In this initial post, I will provide an overview of the book. Those interested in a longer overview can check out my contribution to The Nature of Belief, edited by Jong and Schwitzgebel.
I open the book with three vignettes. A brother tells his sister that cheese made from almonds is just as real cheese as the cheese made from dairy. His sister is puzzled how he could think this, given his love for mammals’ milk brie, until she recalls that her brother recently decided to become dairy-free when he discovered that his daughter is highly allergic to dairy. Does he believe almond cheese is real cheese?
In the Gospel of Mark, a father begs Jesus to heal his sick child, but the father adds a caveat, “if you can,” to which Jesus replies, “all things are possible for one who believes.” In response, the father exclaims, “I believe, help my unbelief!” (Mark 9:22-24). Did the father believe (and not believe) Jesus could heal?
A self-professed egalitarian walks out of a fancy restaurant in a southern US state and sees two men in suits standing by the valet stand. One is white, the other black. The diner hands his keys to the well-dressed black man, thinking him to be a valet. Did the restaurant-goer believe the well-dressed black man was a valet?
These cases point to a more basic question: what is a belief? In this book, I have three goals. First, I answer the “what is a belief?” question. Second, I offer an empirically viable alternative to Dual-Process Theory: the Soundboard Account of Human Reasoning. Third, I offer a method that connects the nature of belief to empirical work by way of cognitive architecture. In this short introductory post, I outline these goals in reverse order.
How should we answer the “what is a belief?” question? While many philosophers have answered using an a priori method, my method is to treat belief as a putative kind within the taxonomy of cognitive science, which I take to be clusters of causal properties. If kinds are clusters of causal properties, then a good way to distinguish belief from acceptance, credence, propositional faith, and imagination—our taxonomy of cognitive states—is to look at models of cognition, often referred to as cognitive architectures. Noticing that the contestable cases of belief, such as those in the vignettes above, all involve reasoning, I explore the cognitive architecture of human reasoning.
Which cognitive architecture should we adopt? One might start with Dual-Process Theory (DPT) because it is influential and has been applied to the nature of belief. I take DPT to be committed to the existence of distinct cognitive kinds: Type 1 and Type 2 processing. If Type 1 and Type 2 processing are real kinds, we should see a commonality across the domains in which they are supposed to occur. Examining DPT in social psychology, cognitive psychology, and cognitive science of religion, I argue that what we find is a grab-bag of properties that are employed or ignored depending on the task that is to be explained. I argue that updated versions of DPT face a dilemma: either the Type 1/Type 2 distinction loses its explanatory power or is not empirically supported. I suspect one reason DPT has stuck around so long is the absence of a viable alternative.
In response, I offer the Soundboard Account of Human Reasoning, according to which the reasoning system can operate quickly or slowly, concretely or abstractly, propositionally or pictorially. These property contrasts crosscut, and many admit of gradation. Think of each contrast as a slide or switch on a sound-mixing board. Just as a DJ listens to the output of a track and can adjust the reverb or bass, so too subjects can re-run reasoning processes one after another, adjusting the way in which they reason. Just as a DJ can add a high-hat track, so too a reasoner can recruit additional beliefs. I argue that this affords us a degree of control over our beliefs.
Here is a concrete example. Suppose I am on a train to Brussels that is running late, and I have a tight connection on the other end for the last train of the night to Antwerp. Perhaps the initial response from my reasoning system is that I will not make my connection. An initial fast mode of operation moves from the belief that the train is late and my connection is tight to the output that I will miss my connection. Of course, I don’t want this to be the case, and so I engage in more mathematical reasoning, determining the amount of time left on my current train, adding that amount to the current time along with the 5 minutes I will need to get to the next platform. Perhaps the output is that I will just make the train. Of course, these are not the only modes of operation. I could also reason abductively here: if my train is delayed, it is plausible that other trains are too, including my next train. In addition to mode alteration, I can recruit additional beliefs and heuristics. I might recall that I have often felt anxious about making tight connections but that I have generally caught my train. I might reason heuristically: train companies make things work out for the customers. We can exercise some control over our beliefs by continuing to rerun the process.
I argue for the Soundboard Account by way of an inference to the best explanation, explaining some of the heuristics and biases that led to DPT. In addition, I contrast my account with Spinozan Accounts of belief fixation and show how the Soundboard Account can explain why different implicit measures do not correlate well with one another and can be easily manipulated. In short, once we abandon DPT, it becomes clear that implicit is not a unified kind, and the Soundboard Account can makes sense of different types of manipulations and part of processes that measure falling under the heading ‘implicit’ proport to measure. I then turn back to the question of belief.
Examining the output of reasoning processes on the Soundboard Account, I argue that beliefs are settling states, semantically evaluable, controlled through the rerunning of reasoning processes, inferentially promiscuous, and fragmented based on domain of content. My view casts belief widely, and each opening vignette involves belief in one way or another. The brother has come to believe that almond cheese is just as much real cheese as that which is made from dairy. He has done so by rerunning his reasoning processing until it produces his desired output. The father in the Gospel of Mark simultaneously believes and does not believe that Jesus can heal his son. In some fragments are one belief, but in others are a contradictory belief. The self-professed egalitarian diner in the US south does believe that the well-dressed black man is a valet, even though he may not be aware of this belief at the time. I conclude the book by looking at future directions in the ethics of belief and how one might test aspects of the Soundboard Account.
I hope that the book appeals to philosophers and psychologists looking for an alternative to DPT, and those working in the metaphysics of belief. I am delighted to have commentators who focused on both of these aspects of the work. Rumana’s comments raise concerns about the framing of my account against DPT. Walker draws out possible implications for the ethics of belief. Curry connects my argument for voluntarism about belief to the rerunning of the reasoning process. Thanks again to Dan for organizing and Aliya, Lesley, and Devin for their comments.