Sebastián Martínez

A common difficulty for reasoning well concerns the vast amounts of information that are potentially at our disposal. Sometimes, it might be unclear whether or not more reasoning will bring about better results. Even if clear, sometimes other demands different from reasoning make it unclear whether we should spend more cognitive resources doing it. So, for limited reasoners like us, it's crucial to engage in simplified reasoning, reasoning that has been simplified in such a way that we don't have to deal with potentially overwhelming amounts of information. Different areas of philosophy have been interested for a number of years in simplified reasoning: epistemologists, philosophers of mind, and decision theorists have dealt with simplified reasoning in different ways. A particularly crucial point of discussion has been on whether simplified reasoning is rational. How should we theorize about the rationality of a way of reasoning that seems selective, and that seems influenced by factors that epistemologists usually associate with irrational ways of thinking? How to take seriously the idea that simplification is unavoidable for all limited reasoners? And how should we account in our theories for a potential shift in normative focus arising from taking that idea seriously? The dissertation provides a comprehensive, philosophical approach to these questions. The thesis is divided into two parts. Part I is concerned with the building blocks of the contemporary discussion of simplified reasoning. Chapter 1, presents the basic concepts invoked in contemporary approaches to simplified reasoning. I first discuss some cases that are connected with certain judgements about simplified reasoning, and propose some common features in these cases. Before appealing to formal tools, this approach will help us to have an initial outlook of the kind of debates that arise from taking these cases as theoretical targets. Chapter 2 then begins to appeal to formal tools familiar from the Bayesian tradition in order to examine two particular models of simplified reasoning. I offer a detailed argument of why decision-theoretic tools are particularly useful in the discussion, an argument that will make evident how we profit from taking a formal approach. In addition, I develop another argument showing the limits of the formal approach, with direct consequences for our normative theories. Chapter 3 takes the Bayesian approach as a common foe for different philosophical theories of simplified reasoning. Here, I offer a clear mapping of the stakes of the debate, and argue for the advantages of adopting an account that fully captures the way in which different doxastic attitudes play different roles in simplified reasoning. On such a full account, we pay attention to beliefs and credences as determining how we simplify our reasoning, but we take acceptance to be the attitude that realizes the actual simplification in reasoning. Part II deals with a positive proposal concerning how to theorize about the rationality of simplified reasoning, taken as a reasoning process. Chapter 4 lays the building blocks of a bounded-resources framework. I call this framework 'consequentialist', as it works using the basic tenet of assessing ways of reasoning in terms of what is best at a given reasoning situation. This framework is widely applied in cognitive psychology and is close to normative insights in epistemology and the philosophy of mind, two approaches that I bring to bear to motivate the use of the framework. I argue that this framework can help to build models of cases of simplified reasoning. First, I lay out the methodology, identify the normative targets, and clarify how it relates to classical approaches to rationality. Then, I construct a toy model that will get me to defend that simplified reasoning can be boundedly rational in a particular sense. Finally, Chapter 5 goes further than the previous chapter to clarify how the models of simplified reasoning are to be interpreted. According to the view defended, assessing simplified reasoning as boundedly rational means that we are doing two things at the same time: we are assessing limited agents for doing (or failing to do) the best they can with the resources they have and we are assessing their attitudes for being a good (or a bad) way for those kinds of agents to acquire epistemic and practical goods. In particular, I defend the constitutive enabling view regarding the relationship between rational simplified reasoning and epistemic rationality. On this view, the rationality of simplified reasoning serves as an enabling condition for limited agents like us to be able to achieve epistemic rationality. So, I provide a proof of concept of getting to interesting views in epistemology using tools from cognitive psychology and decision theory. More broadly, this dissertation shows how to work productively in the epistemology of reasoning by collaborating with other disciplines to increase our understanding of how we get to cope in complex and uncertain environments.:I An Opinionated Introduction 1 1 A Preliminary Characterization of Illustrative Cases 3 1.1 The Interest in Simplified Reasoning 3 1.2 A Case of Planning 5 1.3 Introducing Decision-Theoretic Frameworks 11 1.4 Cases of Theoretical Reasoning 20 1.5 Conclusion 27 2 A Bayesian Framework for Simplified Reasoning 29 2.1 The Problem of Credence Apportioning 29 2.2 Harsanyi on Strict and Practical Credences 32 2.3 Selecting Actions Based on Practical Credences 43 2.4 Rational Constraints on the Assumption Set 49 2.5 Clarke on Sensitive Credal States 56 2.6 Moving On 61 3 Philosophical Theories of Simplified Reasoning 63 3.1 The Explanatory Challenge 63 3.2 Belief Accounts 65 3.2.1 Treating-as-true and Belief 67 3.2.2 The Belief Account against the Bayesian Framework 72 3.3 Credence Accounts 75 3.3.1 Reductionist C-accounts 76 3.3.2 Non-Reductionist C-Accounts 78 3.4 Belief Accounts Keeping Up 82 3.5 Acceptance Accounts 90 3.6 Towards a Full Account of Simplified Reasoning 96 3.7 Conclusion 100 II Normative Topics 101 4 A Consequentialist Framework of Simplified Reasoning 103 4.1 Introduction 103 4.2 Methodological Remarks 104 4.3 The Philosophical Underpinnings of Resource Rationality 112 4.4 Formal Elements of Resource-Rational Frameworks 116 4.5 Approximation, Waste, and Simplification Value 123 4.6 Discussion of Optimization 131 4.7 Moving On 134 5 Simplified Reasoning as Resourcefully Rational 137 5.1 Introduction 137 5.2 Tractability Revisited 138 5.3 The Worry from Normative Robustness 140 5.4 An Account of Tractable Epistemic Rationality 146 5.5 Conclusion 150 A Intractability in Forecasting Reasoning Problems 155 Appendix 155 Glossary 163 Bibliography 165.