Igor Douven

This paper studies the learnability of natural concepts in the context of the conceptual spaces framework. Previous work proposed that natural concepts are represented by the cells of optimally partitioned similarity spaces, where optimality was defined in terms of a number of constraints. Among these is the constraint that optimally partitioned similarity spaces result in easily learnable concepts. While there is evidence that systems of concepts generally regarded as natural satisfy a number of the proposed optimality constraints, the connection between naturalness and learnability has been less well studied. To fill this gap, we conduct a computational study employing two standard models of concept learning. Applying these models to the learning of color concepts, we examine whether natural color concepts are more readily learned than nonnatural ones. Our findings warrant a positive answer to this question for both models employed, thus lending empirical support to the notion that learnability is a distinctive characteristic of natural concepts.

In Suppose and Tell, Williamson makes a new and original attempt to defend the material conditional account of indicative conditionals. His overarching argument is that this account offers the best explanation of the data concerning how people evaluate and use such conditionals. We argue that Williamson overlooks several important alternative explanations, some of which appear to explain the relevant data at least as well as, or even better than, the material conditional account does. Along the way, we also show that Williamson errs at important junctures about what exactly the relevant data are.

Confirmation is a graded notion: evidence can confirm a hypothesis to a greater or lesser degree. There has been debate about how to measure degree of confirmation. Starting from the observation that we would like evidence to be a discriminating indicator of truth, we conduct computer simulations to determine how well the various known measures of confirmation predict the extent to which a given piece of evidence fulfills that role, given a hypothesis of interest. The outcomes show that some measures are markedly better indicators of truth than others.

Cope’s Computers and Musical Style describes a computer program that allegedly can represent and replicate musical styles solely on the basis of compositions that have been entered into it. If this claim is correct, then it must be that an oeuvre’s stylistic characteristics locally supervene on its textual features, which roughly means that its stylistic properties are entirely determined by its textual properties. In my paper I argue that stylistic properties do not locally supervene on textual properties, and thus that neither Cope’s program nor any other that essentially works like it can represent or replicate styles.

Philosophers have recently questioned the methodological status of agent-based modeling. Meanwhile, this methodology has been central to various studies of the COVID-19 pandemic. Few agent-based COVID-19 models are accessible to philosophers for inspection or experimentation. We make available a package for modeling the COVID-19 pandemic and similar pandemics and give an impression of what can be achieved with it. In particular, it is shown that by coupling an agent-based model to a standard optimizer we are able to identify strategies for implementing non-pharmacological interventions that flexibly lower or raise social activity, depending on how the outbreak develops, while balancing various desiderata that cannot be fully satisfied together. The simulation outcomes to be presented testify to the power of agent-based modeling and thereby help to push back against the recent philosophical critique of this methodology.

In _Suppose and Tell_, Williamson makes a new case for the material conditional account. He tries to explain away apparently countervailing data by arguing that these have been misinterpreted because researchers have overlooked the role of heuristics in the processing of conditionals. Cases involving the receipt of apparently conflicting conditionals play an important dialectical role in Williamson’s book: they are supposed to provide evidence for the material conditional account as well as for the defeasibility of a key procedure underlying our everyday assessments of conditionals. We argue that they can serve neither of these purposes and that Williamson overestimates the reach of heuristics. We specifically challenge Williamson’s assumption that, in the kind of cases centrally at issue in his book, the recipient of conflicting conditionals will typically accept those at face value, even granting Williamson that conditionals can be freely passed among speakers under normal conditions of testimony.

Many cognitive psychologists have come to regard graded belief as fundamental to our understanding of how humans reason and many have also come to think of probability theory as providing at least part of the norms of correct reasoning. David Over has characterized this development as the emergence of a new paradigm in the Kuhnian sense. The target article argues that the choice of this term was unwarranted and also that it has done more harm than good. This commentary argues that there is nothing in Thomas Kuhn’s work to suggest that he would object to Over’s terminological choice and that there is no evidence that the choice has caused any harm.

It is undeniable that inductive reasoning has brought us much good. At least since Hume, however, philosophers have wondered how to justify our reliance on induction. In important recent work, Schurz points out that philosophers have been wrongly assuming that justifying induction is tantamount to showing induction to be reliable. According to him, to justify our reliance on induction, it is enough to show that induction is optimal. His optimality approach consists of two steps: an analytic argument for meta-induction (that is, induction over inductive methods) and an application of meta-induction to empirical findings concerning the predictive accuracy of our actual inductive practices as compared to various non-inductive methods. The present article agrees with Schurz’s general conclusion but also argues that it leaves an important aspect of our inductive practices unaccounted for, to wit, that these practices have so far been highly successful; to satisfy the optimality requirement, induction only needs to be as successful as random guessing. To explain the success of induction, I turn to insights from social epistemology, also highlighted in Schurz’s work, and make these formally explicit in a framework that combines the Hegselmann–Krause model for opinion dynamics and Carnap’s system of λ rules. Within this framework, I use a standard evolutionary algorithm to show how communities of interacting agents may, through variation and selection, come to consist of highly successful inductive reasoners.

This position paper advocates combining formal epistemology and the new paradigm psychology of reasoning in the studies of conditionals and reasoning with uncertainty. The new paradigm psychology of reasoning is characterized by the use of probability theory as a rationality framework instead of classical logic, used by more traditional approaches to the psychology of reasoning. This paper presents a new interdisciplinary research program which involves both formal and experimental work. To illustrate the program, the paper discusses recent work on the paradoxes of the material conditional, nonmonotonic reasoning, and Adams’ Thesis. It also identifies the issue of updating on conditionals as an area which seems to call for a combined formal and empirical approach.

There is widespread agreement among philosophers about the Mens Rea Asymmetry (MRA), according to which praise requires intent, whereas blame does not. However, there is evidence showing that MRA is descriptively inadequate. We hypothesize that the violations of MRA found in the experimental literature are due to what we call “moral compositionality,” by which we mean that people evaluate the component parts of a moral problem separately and then reach an overall verdict by aggregating the verdicts on the component parts. We have subjected this hypothesis to the test and here report the results of our experiment. We explore several explanations of the experimental findings and conclude that they present a puzzle to moral theory.

There is an ongoing debate about which rule we ought to use for scoring probability estimates. Much of this debate has been premised on scoring-rule monism, according to which there is exactly one best scoring rule. In previous work, I have argued against this position. The argument given there was based on purely a priori considerations, notably the intuition that scoring rules should be sensitive to truthlikeness relations if, and only if, such relations are present among whichever hypotheses are at issue. The present paper offers a new, quasi-empirical argument against scoring-rule monism. This argument uses computational simulations to show that different scoring rules can have different economical consequences, depending on the context of use.

So far, color-naming studies have relied on a rather limited set of color stimuli. Most importantly, stimuli have been largely limited to highly saturated colors. Because of this, little is known about how people categorize less saturated colors and, more generally, about the structure of color categories as they extend across all dimensions of color space. This article presents the results from a large Internet-based color-naming study that involved color stimuli ranging across all available chroma levels in Munsell space. These results help answer such questions as how English speakers name a more complex color set, whether English speakers use so-called basic color terms (BCTs) more frequently for more saturated colors, how they use non-BCTs in comparison with BCTs, whether non-BCTs are highly consensual in less saturated parts of the solid, how deep inside color space basic color categories extend, or how they behave on the chroma dimension.

The Principle of the Benefit of Doubt dictates that, whenever reasonably possible, we interpret earlier-day scientists as referring to entities posited by current science. Putnam has presented the principle as supplementary to his Causal Theory of Reference in order to make this theory generally applicable to theoretical terms. The present paper argues that the principle is of doubtful standing. In particular, it will be argued that the principle lacks a justification and, indeed, is unjustifiable as it stands.

In a celebrated experiment, Joshua Knobe showed that people are much more prone to attribute intentionality to an agent for a side effect of a given act when that side effect is harmful than when it is beneficial. This asymmetry has become known as ‘the Knobe Effect’. According to Knobe's Moral Valence Explanation, bad effects trigger the attributions of intentionality, whereas good effects do not. Many others believe that the Knobe Effect is best explained in terms of the high amount of blame attributed in the harm condition, and the low amount of praise attributed in the help condition. This Blame Hypothesis explains the high number of intentionality attributions in the harm condition in terms of the high degree of blame people ascribe, and the low number of intentionality attributions in the help conditions in terms of the low degree of praise people attribute. We replicated Knobe's original experiment and conducted a logistic regression on the results to probe more deeply into the relationship between attributions of intentionality and responsibility. The statistical analysis revealed a hitherto unknown interaction effect: intentionality correlates with blame, but not with praise. This interaction effect is consistent with the Moral Valence Hypothesis, but inconsistent with the Blame Hypothesis, as well as with two of the three other hypotheses discussed here.

In previous published research (“Conditionals and Inferential Connections: A Hypothetical Inferential Theory,” Cognitive Psychology, 2018), we investigated experimentally what role the presence and strength of an inferential connection between a conditional’s antecedent and consequent plays in how people process that conditional. Our analysis showed the strength of that connection to be strongly predictive of whether participants evaluated the conditional as true, false, or neither true nor false. In this article, we re-analyse the data from our previous research, now focussing on the semantics of conditionals rather than on how they are processed. Specifically, we use those data to compare the main extant semantics with each other and with inferentialism, a semantics according to which the truth of a conditional requires the presence of an inferential connection between the conditional’s component parts.

Various authors have recently argued that you cannot rationally stick to your belief in the face of known disagreement with an epistemic peer, that is, a person you take to have the same evidence and judgmental skills as you do. For, they claim, because there is but one rational response to any body of evidence, a disagreement with an epistemic peer indicates that at least one of you is not responding rationally to the evidence. Given that you take your peer to have the same judgmental skills as you do, and thus regard her to be equally good at assessing the evidence as you are, you will have as much reason for thinking that it is you who is not responding rationally to the evidence as for thinking that it is her. You thus have reason for thinking that your belief on the disputed matter is not a rational response to the evidence. Hence, you cannot rationally stick to your belief.