Shepard has argued that a universal law should govern generalization across different domains of perception and cognition, as well as across organisms from different species or even different planets. Starting with some basic assumptions about natural kinds, he derived an exponential decay function as the form of the universal generalization gradient, which accords strikingly well with a wide range of empirical data. However, his original formulation applied only to the ideal case of generalization from a single encountered (...) stimulus to a single novel stimulus, and for stimuli that can be represented as points in a continuous metric psychological space. Here we recast Shepard's theory in a more general Bayesian framework and show how this naturally extends his approach to the more realistic situation of generalizing from multiple consequential stimuli with arbitrary representational structure. Our framework also subsumes a version of Tversky's set-theoretic model of similarity, which is conventionally thought of as the primary alternative to Shepard's continuous metric space model of similarity and generalization. This unification allows us not only to draw deep parallels between the set-theoretic and spatial approaches, but also to significantly advance the explanatory power of set-theoretic models. Key Words: additive clustering; Bayesian inference; categorization; concept learning; contrast model; features; generalization; psychological space; similarity. (shrink)

[Edited from Conclusion section:] We have looked at various challenging issues to do with getting connectionism to cope with high-level cognitive activities such a reasoning and natural language understanding. The issues are to do with various facets of generalization that are not commonly noted. We have been concerned in particular with the special forms these issues take in the arena of propositional attitude processing. The main problems we have looked at are: (1) The need to construct explicit representations of (...) generalizations, not just generalize correctly to individual cases; (2) The need to be able to match two or more complex short-term information structures, to enable rapid generalization from recent examples rather than from long-term memories; (3) The need to represent and reason with anomalous combinations of concepts; (4) The need to perform embedded reasoning. This presents special problems for systems using non-concatenative representations (as in mainstream connectionist approaches). We also touched on vague quantification in attitude report complements. Neither this topic nor that of analogies between short-term structures (point 2) has been adequately addressed in the symbolic framework, let alone in connectionism. -/- The opportunities and problems covered are put forward as things worth being optimistic about or pessimistic about, respectively. They are not put forward as decisive arguments for or against connectionism. The hope is that this chapter contributes to a greater understanding of the connectionist/symbolist gap by presenting some unusual issues and by throwing new light on some well known ones. (shrink)

This paper focuses on how experiences of trauma can lead to generalized fear of people, objects and places that are similar or contextually or conceptually related to those that produced the initial fear, causing epistemic, affective, and practical harms to those who are unduly feared and those who are intimates of the victim of trauma. We argue that cases of fear generalization that bring harm to other people constitute examples of injustice closely akin to testimonial injustice, specifically, mnemonic injustice. (...) Mnemonic injustice is a label that has been introduced to capture how injustice can occur via the operation of human memory systems when stereotypes shape what is remembered. Here we argue that injustices can also occur via memory systems when trauma leads to a generalized fear. We also argue that this calls for a reformulation of the notion of mnemonic injustice. (shrink)

Le général André Bach dans une réflexion sur l’« horizon de mort dans la violence de guerre » part d’une approche anthropologique du phénomène de violence et de la peur (quasiment biologique) qu’il engendre en soulignant les difficultés des sociétés occidentales à penser la mort. C’est l’État qui donne à la guerre un sens politique et sacré et qui crée les catégories fonctionnelles de la guerre (les concepts de paix et de guerre ne sont pas en eux-mêmes opérationnels). Dans le (...) cadre d’une réflexion clausewitzienne sur la montée aux extrêmes, l’« extrême » pour le soldat est toujours sa propre mort (plus que celle de l’ennemi). Les notions de barbarisation et humanisation ne peuvent en ce sens rendre compte de la violence et de la peur radicales présentes dans toute guerre où le rôle du commandement est toujours de mettre en œuvre un degré de violence supérieur à celui de l’adversaire, sans aucune considération de « proportionnalité » de cette violence. La mise à distance technique ne diminue pas la violence mais l’augmente à proportion de l’éloignement de la peur. Seul le politique peut et doit mettre des limites à cette violence militaire par nature illimitée. La modernisation des armes ne relève jamais d’une humanisation du combat mais d’une recherche d’une violence plus efficace. L’entraînement du soldat en temps de paix n’est jamais entraînement à la violence mais apprentissage de savoir-faire et effort de constitution d’un groupe soudé, doté d’un bon moral. L’ordre et la formation tentent de limiter la sauvagerie mais il existe une constitution endogène de comportements propres au soldat, que la hiérarchie tente de canaliser dans une symbolique guerrière fondée sur le sentiment de l’honneur (destiné à compenser la peur d’être une proie vouée à mourir et à faire préférer la lutte au coude à coude sur l’instinct de conservation). Le couple violence/honneur fait que l’armée dépourvue de contrat social n’est pas soluble dans le politique (aujourd’hui dans les écoles militaires on ne parle plus d’« honneur » mais d’« éthique »…, mais pour dire la même chose). La conscription fut une rupture de la logique endogène de l’armée et introduisit une brèche dans ce système archaïque. Mais religion (au Moyen Âge) ou logique démocratique (de nos jours) s’avèrent surtout des « béquilles idéologiques » qui, en définitive, sont souvent plus efficaces pour accroître la barbarisation que pour l’éviter (exemple de la colonisation algérienne où une armée de paix, fonctionnarisée et rigoureusement canalisée, appartenant à une génération n’ayant jamais connu de violence de guerre, développe des massacres de masse). (shrink)

The term robustness is ubiquitous in modern Machine Learning (ML). However, its meaning varies depending on context and community. Researchers either focus on narrow technical definitions, such as adversarial robustness, natural distribution shifts, and performativity, or they simply leave open what exactly they mean by robustness. In this paper, we provide a conceptual analysis of the term robustness, with the aim to develop a common language, that allows us to weave together different strands of robustness research. We define robustness as (...) the relative stability of a robustness target with respect to specific interventions on a modifier. Our account captures the various sub-types of robustness that are discussed in the research literature, including robustness to distribution shifts, prediction robustness, or the robustness of algorithmic explanations. Finally, we delineate robustness from adjacent key concepts in ML, such as extrapolation, generalization, and uncertainty, and establish it as an independent epistemic concept. (shrink)

Inductive generalization, where people go beyond the data provided, is a basic cognitive capability, and it underpins theoretical accounts of learning, categorization, and decision making. To complete the inductive leap needed for generalization, people must make a key ‘‘sampling’’ assumption about how the available data were generated. Previous models have considered two extreme possibilities, known as strong and weak sampling. In strong sampling, data are assumed to have been deliberately generated as positive examples of a concept, whereas in (...) weak sampling, data are assumed to have been generated without any restrictions. We develop a more general account of sampling that allows for an intermediate mixture of these two extremes, and we test its usefulness. In two experiments, we show that most people complete simple one-dimensional generalization tasks in a way that is consistent with their believing in some mixture of strong and weak sampling, but that there are large individual differences in the relative emphasis different people give to each type of sampling. We also show experimentally that the relative emphasis of the mixture is influenced by the structure of the available information. We discuss the psychological meaning of mixing strong and weak sampling, and possible extensions of our modeling approach to richer problems of inductive generalization. (shrink)

Écriture sans écriture, critique sans discours : il s’agit de traquer et de faire ressortir l’obsession du « projet » et de son idéologie dans la trame d’un appel à contrats doctoraux. Ctrl+B chaque fois (et elles sont très nombreuses) où le petit mot est réitéré.

Category theory has become central to certain aspects of theoretical physics. Bain has recently argued that this has significance for ontic structural realism. We argue against this claim. In so doing, we uncover two pervasive forms of category-theoretic generalization. We call these ‘generalization by duality’ and ‘generalization by categorifying physical processes’. We describe in detail how these arise, and explain their significance using detailed examples. We show that their significance is two-fold: the articulation of high-level physical concepts, (...) and the generation of new models. 1 Introduction2 Categories and Structuralism 2.1 Categories: abstract and concrete 2.2 Structuralism: simple and ontic3 Bain’s Two Strategies 3.1 A first strategy for defending Objectless 3.2 A second strategy for defending Objectless4 Two Forms of Categorical Generalization 4.1 Generalization by duality 4.2 Generalization by categorification 4.3 The role of category theory in physics5 Conclusion. (shrink)

For simplicity, most of the literature introduces the concept of definitional equivalence only for disjoint languages. In a recent paper, Barrett and Halvorson introduce a straightforward generalization to non-disjoint languages and they show that their generalization is not equivalent to intertranslatability in general. In this paper, we show that their generalization is not transitive and hence it is not an equivalence relation. Then we introduce another formalization of definitional equivalence due to Andréka and Németi which is equivalent (...) to the Barrett–Halvorson generalization in the case of disjoint languages. We show that the Andréka–Németi generalization is the smallest equivalence relation containing the Barrett–Halvorson generalization and it is equivalent to intertranslatability, which is another definition for definitional equivalence, even for non-disjoint languages. Finally, we investigate which definitions for definitional equivalences remain equivalent when we generalize them for theories in non-disjoint languages. (shrink)

The universal generalization problem is the question: What entitles one to conclude that a property established for an individual object holds for any individual object in the domain? This amounts to the question: Why is the rule of universal generalization justified? In the modern and contemporary age Descartes, Locke, Berkeley, Hume, Kant, Mill, Gentzen gave alternative solutions of the universal generalization problem. In this paper I consider Locke’s, Berkeley’s and Gentzen’s solutions and argue that they are problematic. (...) Then I consider an alternative formulation of universal generalization which depends on the view that mathematical objects are individual objects and are hypotheses introduced to solve mathematical problems, and that mathematical proofs are argument schemata. I argue that this alternative formulation allows one to overcome the problems of Locke’s, Berkeley’s and Gentzen’s solutions, and is related to the approach to generality in Greek mathematics. I also argue that there is a connection between the present formulation of universal generalization and a special form of the analogy rule which is implicit in Proclus’ approach to the universal generalization problem. (shrink)

The species problem is one of the oldest controversies in natural history. Its persistence suggests that it is something more than a problem of fact or definition. Considerable light is shed on the species problem when it is viewed as a problem in the representation of the natural system (sensu Griffiths, 1974, Acta Biotheor. 23: 85–131; de Queiroz, 1998, Philos. Sci. 55: 238–259). Just as maps are representations of the earth, and are subject to what is called cartographic generalization, (...) so diagrams of the natural system (evolutionary trees) are representations of the evolutionary chronicle, and are subject to a temporal version of cartographic generalization which may be termed systematic generalization. Cartographic generalization is based on judgements of geographical importance, and systematic generalization is based on judgements of historical importance, judgements expressed in narrative sentences (sensu Danto, 1985, Narration and knowledge, Columbia Univ. Press, New York). At higher systematic levels these narrative sentences are conventional and retrospective, but near the "species" level they become prospective, that is, dependent upon expectations of the future. The truth of prospective narrative sentences is logically indeterminable in the present, and since all the common species concepts depend upon prospective narration, it is impossible for any of them to be applied with precision. (shrink)

Drawing general inferences on the basis of single-case and small- n studies is often seen as problematic. This article suggests a logic of generalization based on thinly rationalistic social mechanisms. Ideal-type mechanisms can be derived from empirical observations in one case and, based on the assumption of thin rationality, used as a generalizing bridge to other contexts with similar actor constellations. Thus, the “portability” builds on expectations about similar mechanisms operating in similar contexts. We present the general logic behind (...) such “rationalistic generalization” and relate it to other ideas about generalization from single-case studies. (shrink)

Generalization from a case study is a perennial issue in the methodology of the social sciences. The case study is one of the most important research designs in many social scientific fields, but no shared understanding exists of the epistemic import of case studies. This article suggests that the idea of mechanism-based theorizing provides a fruitful basis for understanding how case studies contribute to a general understanding of social phenomena. This approach is illustrated with a re- construction of Espeland (...) and Sauder's case study of the effects of rankings on US legal education. On the basis of the reconstruction, it is argued that, at least with respect to sociology, the idea of mechanism-based theorizing captures many of the generalizable elements of case studies. (shrink)

We develop an axiomatic theory of “generalized Routley-Meyer logics.” These are first-order logics which are can be characterized by model theories in a certain generalization of Routley-Meyer semantics. We show that all GRM logics are subclassical, have recursively enumerable consequence relations, satisfy the compactness theorem, and satisfy the standard structural rules and conjunction and disjunction introduction/elimination rules. We also show that the GRM logics include classical logic, intuitionistic logic, LP/K3/FDE, and the relevant logics.

An image of a face depends not only on its shape, but also on the viewpoint, illumination conditions, and facial expression. A face recognition system must overcome the changes in face appearance induced by these factors. This paper investigate two related questions: the capacity of the human visual system to generalize the recognition of faces to novel images, and the level at which this generalization occurs. We approach this problems by comparing the identi cation and generalization capacity for (...) upright and inverted faces. For upright faces, we found remarkably good generalization to novel conditions. For inverted faces, the generalization to novel views was signi cantly worse for both new illumination and viewpoint, although the performance on the training images was similar to the upright condition. Our results indicate that at least some of the processes that support generalization across viewpoint and illumination are neither universal (because subjects did not generalize as easily for inverted faces as for upright ones), nor strictly objectspeci c (because in upright faces nearly perfect generalization was possible from a single view, by itself insu cient for building a complete object-speci c model). We propose that generalization in face recognition occurs at an intermediate level that is applicable to a class of objects, and that at this level upright and inverted faces initially constitute distinct object classes. (shrink)

The paper focuses on some puzzles about Carnap's intended epistemological point in the "completion" and "generalization" of the Anschauungsraum in sec. II of Der Raum (leaving aside the technical problems which also arise). Since any global structure at all requires that eidetic intuition be supplemented with freely-chosen postulates and/or intuitively unmotivated generalizations, it is unclear, as several authors have pointed out, how and in what sense "intuitive space" as a whole represents a distinctive, a priori contribution to our knowledge. (...) I suggest a way of approaching this issue based on Carnap's sources -- in particular, Husserl and Driesch, both of whom he repeatedly claims to be following. The idea of a severely finite realm of possible intuition, which both requires and allows supplementation with an infinite conceptual structure, is central to Husserl's thought, and, I argue, it would be natural for Carnap to rely on it in attempting to reconcile Husserlian eidetic intuition with the general theory of relativity. That this larger conceptual structure owes its details to free postulation is, on the other hand, decidedly un-Husserlian. But here, I claim, Carnap takes his cue from Driesch's view, in the first edition of the Ordnungslehre, that natural actuality is the result of a certain demand for order: we demand such order in natural things and reject as non-actual (hallucinatory, dreamed, etc.) whatever nature-like elements of experience fail to fit into it. In conclusion I suggest that more radical versions of this particular constellation of ideas provide the key to understanding much of Carnap's later thought. (shrink)

One of the main problems that Paul Horwich’s Minimalist theory of truth must face is the generalization problem, which shows that Minimalism is too weak to have the fundamental explanatory role Horwich claims it has. In this paper, I defend Horwich’s response to the generalization problem from an objection raised by Bradley Armour-Garb. I also argue that, given my response to Armour-Garb, Horwich’s proposal to cope with the generalization problem can be simplified. -/- L’un des principaux problèmes (...) auxquels la théorie minimaliste de la vérité de Paul Horwich doit faire face est le problème de la généralisation. Horwich soutient que le minimalisme a un rôle explicatif fondamental, mais le problème de la généralisation montre que cette théorie est trop faible pour tenir ce rôle. Dans cet article, je défends la réponse d’Horwich au problème de la généralisation à partir d’une objection soulevée par Bradley Armour-Garb. Je prétends également que ma réponse à Armour-Grab nous permet de formuler d’une manière plus simple la réponse d’Horwich au problème de la généralisation. (shrink)

For some time now, Jaegwon Kim has argued that irreducible mental properties face the threat of causal inefficacy. The primary weapon he deploys to sustain this charge is the supervenience/exclusion argument. This argument, in a nutshell, states that any mental property that irreducibly supervenes on a physical property is excluded from causal efficacy because the underlying physical property takes care of all of the causal work itself. Originally intended for mental properties alone, it did not take long for his critics (...) to suggest the argument generalizes across all of the special science properties as well. Kim responds in two different ways to the generalization problem. The first response, which I call the higher-level solution, is ably dismissed by numerous critics. The second response, which I call the identity solution, has not faced comparable scrutiny. In this paper I argue that the identity solution faces numerous problems of its own. (shrink)

In this paper, we shall present a generalization of phrase structure grammar, in which all functional categories have type restrictions, that is, their argument types are specific domains. In ordinary phrase structure grammar, there is just one universal domain of individuals. The grammar does not make a distinction between verbs and adjectives in terms of domains of applicability. Consequently, it fails to distinguish between sentences like every line intersects every line, which is well typed, and every line intersects every (...) point, which is ill typed.Our generalization relates to ordinary phrase structure grammar in the same way as the higherlevel constructive type theory of Martin-Löf relates to the simple type theory of Church . Simple type theory has been used in linguistics and related with phrase structure grammar, especially in the tradition based on the work of Montague .Our definition of the grammar will be more formal than Montague's in the sense that we shall use a formal metalanguage, that of constructive type theory, for defining both the object language and its interpretation. The grammar, both syntax and semantics, is thus readily implementable in the type-theoretical proof system ALF . Inside type theory, a distinction can be made between the object language and the model, in other words, between syntactic and semantic types. It will turn out that the object language cannot be defined independently of the model, as in ordinary Tarski semantics. This is a direct consequence of introducing the typing restrictions.The grammar presented here can be seen as a formal linguistic elaboration of the work presented in Ranta 1991 and 1994. The organization of generative grammar as categorial grammar followed by sugaring should now look more familiar to the linguist, as the grammatical formalism on which sugaring operates is no longer full type theory but a set of phrase structure trees. (shrink)

In the framework of Stewart Shapiro, computations are performed directly on strings of symbols (numerals) whose abstract numerical interpretation is determined by a notation. Shapiro showed that a total unary function (unary relation) on natural numbers is computable in every injective notation if and only if it is almost constant or almost identity function (finite or co-finite set). We obtain a syntactic generalization of this theorem, in terms of quantifier-free definability, for functions and relations relatively intrinsically computable on certain (...) types of equivalence structures. We also characterize the class of relations and partial functions of arbitrary finite arities which are computable in every notation (be it injective or not). We consider the same question for notations in which certain equivalence relations are assumed to be computable. Finally, we discuss connections with a theorem by Ash, Knight, Manasse and Slaman which allow us to deduce some (but not all) of our results, based on quantifier elimination. (shrink)

The paper contains a survey of (mainly unpublished) adaptive logics of inductive generalization. These defeasible logics are precise formulations of certain methods. Some attention is also paid to ways of handling background knowledge, introducing mere conjectures, and the research guiding capabilities of the logics.

This paper examines the generalization of the Internet address term “family” (jiārénmen), a recently emerging phenomenon in language use from the perspective of Pragmatic Identity Theory. The main thrust of this research is to reveal the pragmatic identities and pragmatic functions involved in the use of “family” (jiārénmen). The core point has been arrived at by using a combination of quantitative and qualitative analyses to explore relevant cases collected from WeChat, Weibo, and in-person conversations. Research findings demonstrated that in (...) different contexts, the address term “family” (jiārénmen) can construct various pragmatic identities, all of which belong to the deviational identity opposite to the default identity. Notably, the construction of these identities functions to patch up address terms slot, promote relationships, and satisfy transactional needs. Furthermore, the generalization of the address term “family” (jiārénmen) mirrors the effect of the psychological mechanism of communicators and the Chinese family culture on discourse practice. The findings above contribute to extending the research line of Internet address term generalization, as well as a deeper understanding of the mechanism of the Internet address term “family” (jiārénmen) towards a harmonious language-using environment. (shrink)

For simplicity, most of the literature introduces the concept of definitional equivalence only to languages with disjoint signatures. In a recent paper, Barrett and Halvorson introduce a straightforward generalization to languages with non-disjoint signatures and they show that their generalization is not equivalent to intertranslatability in general. In this paper,we show that their generalization is not transitive and hence it is not an equivalence relation. Then we introduce the Andréka and Németi generalization as one of the (...) many equivalent formulations for languages with disjoint signatures. We show that the Andréka-Németi generalization is the smallest equivalence relation containing the Barrett–Halvorson generalization and it is equivalent to intertranslatability even for languages with non-disjoint signatures. Finally,we investigate which definitions for definitional equivalences remain equivalent when we generalize them for theories with non-disjoint signatures. (shrink)

A central question in cognitive research concerns how new properties are generalized to categories. This article introduces a model of how generalizations involve a process of causal inference in which people estimate the likely presence of the new property in individual category exemplars and then the prevalence of the property among all category members. Evidence in favor of this causal‐based generalization (CBG) view included effects of an existing feature’s base rate (Experiment 1), the direction of the causal relations (Experiments (...) 2 and 4), the number of those relations (Experiment 3), and the distribution of features among category members (Experiments 4 and 5). The results provided no support for an alternative view that generalizations are promoted by the centrality of the to‐be‐generalized feature. However, there was evidence that a minority of participants based their judgments on simpler associative reasoning processes. (shrink)