Steels & Belpaeme (S&B) ask how autonomous agents can derive perceptually grounded categories for successful communication, using color categorization as an example. Their comparison of nativism, empiricism, and culturalism, although interesting, does not include key biological and technological constraints for seeing color or learning color categories in realistic environments. Other neural models have successfully included these constraints.

It is becoming increasingly clear that acquiring cognitive skills is feasible only with significant developmental constraints. However, recent research provides the strongest evidence to date for constructivist development. Here, we examine how these two apparently conflicting perspectives may be reconciled. Specifically, we suggest that subcortical and cortical structures possess divergent developmental strategies, with many subcortical structures satisfying Fodor's criteria for modularity. These structures constitute an early behavioral system that guides the construction of later emerging cortical structures, for which there (...) is little evidence for modularity. Thus, we focus on how the dynamic time course of development itself implicitly constrains the emergence of cortical representations, reducing the requirement for built-in encodings of knowledge in cortical circuits, as on the traditional nativist conception. (shrink)

The paper examines constraints and preferences employed by people in learning decision rules from preclassified examples. Results from four experiments with human subjects were analyzed and compared with artificial intelligence (AI) inductive learning programs. The results showed the people's rule inductions tended to emphasize category validity (probability of some property, given a category) more than cue validity (probability that an entity is a member of a category given that it has some property) to a greater extent than (...) did the AI programs. Although the relative proportions of different rule types (e.g., conjunctive vs. disjunctive) changed across experiments, a single process model provided a good account of the data from each study. These observations are used to argue for describing constraints in terms of processes embodied in models rather than in terms of products or outputs. Thus AI induction programs become candidate psychological process models and results from inductive learning experiments can suggest new algorithms. More generally, the results show that human inductive generalizations tend toward greater specificity than would be expected if conceptual simplicity were the key constraint on inductions. This bias toward specificity may be due to the fact that this criterion both maximizes inferences that may be drawn from category membership and protects rule induction systems from developing over‐generalizations. (shrink)

Traditionally, learning organizations face certain constraints related to both exogenous and endogenous factors. This paper models three well-established constraints that employees face while being part of their organizations. One is an explicit constraint on their natural behavior, and two implicit constraints on their endeavor to acquire new knowledge and perform new actions. The implicit constraints, which are elaborated, are related to their relative performance in acquiring new knowledge and by their consecutive actions based on the (...) new knowledge gained. Therefore, this paper attempts to underline such limitations which the agents face under organizational culture and suggest possible strategic initiatives that would effectively counteract such binding limitations to stimulate positive performances from their end. (shrink)

Schyns, Goldstone & Thibaut argue that categorization experience results in the learning of new perceptual features that are not derivable from the learner's existing feature set. We explore the meaning and implications of this “nonderivability” claim and relate it to the question of whether perceptual invariants are learnable, and if so, what might be entailed in learning them.

This paper will explore the relationship between Mezirow's concept of reflexivity and Bourdieu's theory of habitus in order to develop a more robust framework within which critical reflection can take place. Nurse educators have sought to close the theory practice gap through the use of critical reflection. However, we are not convinced that this has produced the depth and quality of reflection required. Furthermore, the contexts in which critical reflection takes place is often sidelined or erased so that the whole (...) impetus in the literature is to educate nurses in reflection rather than empower nurses to understand the complex circumstances and barriers that obstruct critical reflection. This paper argues that the reason for this position is that nurse education does not always acknowledge the role that personal and cultural values systems have on reflective practices. The literature search was undertaken using CINHAL and MEDLINE. Keywords included: values systems, habitus, and critical reflection. Inclusion criteria were determined by the theoretical approach and included seminal texts, from as far back as 1956, to identify key themes. Although critical reflection is a potentially powerful way of enhancing care it has often failed to do so. It is suggested that this is because nurse educators have frequently used models of reflection without considering the impact that student's personal values systems has on their perceptions of care and subsequently care delivery. The purpose of this paper is to promote deeper levels of reflection and is part of a programme of research aimed at developing a more robust approach to reflection in educational practice. (shrink)

Levelt, reviewing the AGL field from a psycholinguistic perspective, identifies various gaps and makes a number of concrete suggestions for improving several currently used experimental designs. He raises the question whether artificial (and natural) grammar learning is about detecting ‘rules’, as is commonly assumed, or rather the detection of a set of ‘constraints’. He cautions the community to not ignore ‘semantics’, and recommends to consider less artificial tasks, that may be needed for learning more complex rules by (...) human or nonhuman animals. (shrink)

It is widely recognized that computational theories of learning must posit the existence of a priori constraints on hypothesis selection. The present article surveys the theoretical options available for modelling the dynamic process whereby the constraints have their effect. According to the 'simplicity' theory (exemplified by Fodor's treatment), hypotheses are preference-ordered in terms of their syntactic or semantic properties. It is argued that the same explanatory power can be obtained with a weaker (hence better) theory, the 'minimalist' (...) theory, which dispenses with the preference ordering. According to the 'finitistic' theory, the learner is capable of generating only finitely many hypotheses for evaluation. Chomsky maintains that the occurrence of errorless learning in language acquisition necessitates a finististic explanation. Once again, there is a weaker theory that explains the same data. Finally, Goodman's argument to the effect that there cannot be a computational theory of learning is examined and rejected. (shrink)

A theory of conceptual development must provide an account of the innate representational repertoire, must characterize how these initial representations differ from the adult state, and must provide an account of the processes that transform the initial into mature representations. In Carey, 2009 (The Origin of Concepts), I defend three theses: 1) the initial state includes rich conceptual representations, 2) nonetheless, there are radical discontinuities between early and later developing conceptual systems, 3) Quinean bootstrapping is one learning mechanism that (...) underlies the creation of new representational resources, enabling such discontinuity. I also claim that the theory of conceptual development developed in The Origin of Concepts addresses two of Fodor's challenges to cognitive science; namely, to show how learning could possibly lead to an increase in expressive power and to defeat Mad Dog Nativism, the thesis that all concepts lexicalized as mono-morphemic words are innate. A recent article by Georges Rey (Mind & Language, 29.2, 2014) argues that my responses to Fodor's challenges fail, because, he says, I fail to distinguish concept possession from manifestation and I do not confront Goodman's new riddle of induction. My response is to show that, and how, new primitives in a language of thought can be learned, that there are easy routes and hard ones to doing so, and that characterizing the learning mechanisms involved is the key to understanding both concept possession and constraints on induction. (shrink)

This dissertation concerns Plato's theory of education and the problem of how one can actually acquire knowledge of the Forms. Plato's theory of education aims to make one a good person, which requires knowledge of the Form of the Good. Yet, how exactly one would acquire such knowledge has remained a mystery. Various models of learning are presented by Plato: elenctic refutation ; hypothesis; recollection; the mathematical, dialectical, and political studies of the Republic's curriculum; and diairesis to name just (...) those that are explicit. I have argued here that as rich as these learning strategies are we need to augment our examination of Socratic pedagogy with a full exegesis of Platonic epistemology. ;In order finally to appreciate how they, Socratic education and Platonic epistemology, are meaningfully inseparable, I have explored the following questions: what kind of teacher Socrates is and the way in which he struggles against the sophistic methods of teaching; why Plato has Socrates include ten years of mathematical study in the ideal philosophical curriculum and how we can take the Republic's paradoxical curriculum seriously; how the philosopher utilizes the mathematical training in coming to know a Platonic Form; what role opposites play in Plato's metaphysical imagination; and what kind of person would one be if one actually knew the Form of the Good. I have shown that if one were to partake of a Socratic education, to pursue the requisite mathematical training, and to arrive finally at knowledge of the Platonic Forms, then one would be the sort of person that Socrates has in mind when he articulates the principles of Socratic intellectualism. ;What has been novel about this approach is its demonstration of how, despite the fact that Socratic learning and Platonic epistemology have separately been the object of perennial scholarly attention, the two taken genuinely together have been in need of fresh critical engagement. This inquiry reveals the metaphysical constraints on knowledge and in so doing reconciles Plato's theory of knowledge with Socrates' style of educating potential philosophers. (shrink)

Statistical learning is often taken to lie at the heart of many cognitive tasks, including the acquisition of language. One particular task in which probabilistic models have achieved considerable success is the segmentation of speech into words. However, these models have mostly been tested against English data, and as a result little is known about how a statistical learning mechanism copes with input regularities that arise from the structural properties of different languages. This study focuses on statistical word (...) segmentation in Arabic, a Semitic language in which words are built around consonantal roots. We hypothesize that segmentation in such languages is facilitated by tracking consonant distributions independently from intervening vowels. Previous studies have shown that human learners can track consonant probabilities across intervening vowels in artificial languages, but it is unknown to what extent this ability would be beneficial in the segmentation of natural language. We assessed the performance of a Bayesian segmentation model on English and Arabic, comparing consonant-only representations with full representations. In addition, we examined to what extent structurally different proto-lexicons reflect adult language. The results suggest that for a child learning a Semitic language, separating consonants from vowels is beneficial for segmentation. These findings indicate that probabilistic models require appropriate linguistic representations in order to effectively meet the challenges of language acquisition. (shrink)

We sought to gain more insight into the effects of attention focus and time constraints on skill learning and performance in novices and experts by means of two complementary experiments using a table tennis paradigm. Experiment 1 showed that skill-focus conditions and slowed ball frequency disrupted the accuracy of experts, but dual-task conditions and speeded ball frequency did not. For novices, only speeded ball frequency disrupted accuracy. In Experiment 2, we extended these findings by instructing novices either explicitly (...) or by analogy . Explicitly instructed novices were less accurate in skill-focused and dual-task conditions than in single-task conditions. Following analogy instruction novices were less accurate in the skill-focused condition, but maintained accuracy under dual-task conditions. Participants in both conditions retained accuracy when ball frequency was slowed, but lost accuracy when ball frequency was speeded, suggesting that not attention, but motor dexterity, was inadequate under high temporal constraints. (shrink)

The ArgumentIn this paper three canonical studies of the scientific revolution are subjected to narratological analysis. Underlying this analysis is the assumption that in any single product of historical writing it is possible to distinguish, for analytical purposes, between three levels of reference: the object of the text — the events; the representation of the events — the narrative; and the text in which a story is represented by means of narrative. Through texts one learns about historical events, but also (...) about the process of producing the texts. Techniques of representation are means of production of texts that represent, in addition to the events, also the producer and his or her authority. The distinction made for any specific text allows the reader to play the two levels against each other in order to create a space within which historical writing should be criticized. Such criticism may draw attention to the difficulty of controlling the meaning of historical stories, and to the need to analyze narrative structures in order to appreciate the ideological changes of historical texts. (shrink)

There is substantial variation in language experience, yet there is surprising similarity in the language structure acquired. Constraints on language structure may be external modulators that result in this canalization of language structure, or else they may derive from the broader, communicative environment in which language is acquired. In this paper, the latter perspective is tested for its adequacy in explaining robustness of language learning to environmental variation. A computational model of word learning from cross-situational, multimodal information (...) was constructed and tested. Key to the model's robustness was the presence of multiple, individually unreliable information sources to support learning. This “degeneracy” in the language system has a detrimental effect on learning, compared to a noise-free environment, but has a critically important effect on acquisition of a canalized system that is resistant to environmental noise in communication. (shrink)

Open peer commentary on the article “A Cybernetic Approach to Contextual Teaching and Learning” by Philip Baron. Upshot: The introduction of conversational methods into the traditional classroom is a laudable undertaking. The ability of these methods to transform education is constrained by the educational management systems that determine many aspects of educational conversations.

A variety of mechanisms contribute to word learning. Learners can track co-occurring words and referents across situations in a bottom-up manner. Equally, they can exploit sentential contexts, relying on top–down information such as verb–argument relations and world knowledge, offering immediate constraints on meaning. When combined, CSWL and SLCL potentially modulate each other's influence, revealing how word learners deal with multiple mechanisms simultaneously: Do they use all mechanisms? Prefer one? Is their strategy context dependent? Three experiments conducted with adult (...) learners reveal that learners prioritize SLCL over CSWL. CSWL is applied in addition to SLCL only if SLCL is not perfectly disambiguating, thereby complementing or competing with it. These studies demonstrate the importance of investigating word-learning mechanisms simultaneously, revealing important characteristics of their interaction in more naturalistic learning environments. (shrink)

The proposal that language has evolved to conform to general cognitive and learning constraints inherent in the human brain calls for specification of these mechanisms. We propose that just as cognition appears to be grounded in cross-modal perceptual-motor capabilities, so too must language. Evidence for perceptual-motor grounding comes from non-arbitrary sound-to-meaning correspondences and their role in word learning.

Cosmopolitan education aims to transmit cosmopolitan forms of life in order to participate morally in the world community. The primary characteristics of this cosmopolitan education are its acceptance of the shared humanity of all persons as a fact of human existence and as a motivating guide for human interaction, and the requirement of democratic inclusion in deliberations of the governance of those interactions, including morality. Such an education in cosmopolitan morality requires means that befit its core components. This paper contrasts (...) the concepts of strict and moderate cosmopolitanism, empirical and deliberative morality, and structural and dispositional cosmopolitanism to show that the moderate, inclusive and deliberative processes of deliberative dispositional cosmopolitanism are more suited to cosmopolitan education in morality than strong, empirical-focused structural cosmopolitan efforts. Though strong, structural and empirically based forms may be more likely to guarantee preferred outcomes in the learning of specific morals or the implementation of institutional norms, they are also more likely to run afoul of the core components of cosmopolitanism because they will privilege outcomes over processes, and are thus more likely to be less inclusive and more coercive. In contrast, and even though they are less certain to guarantee preferred moral outcomes or actions, moderate, deliberative dispositional forms of cosmopolitan education embody the morality they seek to inform, and are more likely to find sustained internalised support over time. (shrink)

How linguistic structures evolve so as to become easier to process is addressed by Isbilen and Christiansen for the Now‐or‐Never bottleneck. The authors suggest that this fundamental challenge in language processing is coped with by rapid compression of the transient linguistic input into chunks then to be passed on. As linguistic structures that can be chunked more easily tend to stabilize and proliferate, language evolves to fit learners’ cognitive capabilities.

In this study, a new mathematical model is presented to solve the flexible flow shop problem where transportation is reliable and there are constraints on intermediate buffers, budgets, and human resource learning effects. Firstly, the model is validated to confirm the accuracy of its performance. Then, since it is an NP-hard one, two metaheuristic algorithms, namely, MOSA and MOEA/D, are rendered to solve mid- and large-scale problems. To confirm their accuracy of performance, two small-scale problems are solved using (...) GAMS exact solution software, and the obtained results have been compared with the output of the algorithms. Since the problem in this study is multiobjective, five comparative indices are used to compare the performance of algorithms. The results show that the answers achieved using the metaheuristic algorithms are very close to the ones achieved via the GAMS exact program. Therefore, the proposed algorithms are validated, and it is proved that they are accurately designed and useable in solving the real-world problems in logical calculation time. By comparing the obtained results, it can be seen that the MOEA/D algorithm performs better in terms of computational time and Mean ideal distance. The MOSA algorithm also performs better according to the index Spread of nondominated solutions, diversity metric, and number of Pareto solutions. Considering the confirmation of precision and accuracy of performance of the proposed algorithms, it can be concluded that MOSA and MOEA/D are useful in solving the mid- and large-scale modes of the problem in the study, which is very applicable in the real world. (shrink)

This essay explores the universal cognitive bases of biological taxonomy and taxonomic inference using cross-cultural experimental work with urbanized Americans and forest-dwelling Maya Indians. A universal, essentialist appreciation of generic species appears as the causal foundation for the taxonomic arrangement of biodiversity, and for inference about the distribution of causally-related properties that underlie biodiversity. Universal folkbiological taxonomy is domain-specific: its structure does not spontaneously or invariably arise in other cognitive domains, like substances, artifacts or persons. It is plausibly an innately-determined (...) evolutionary adaptation to relevant and recurrent aspects of ancestral hominid environments, such as the need to recognize, locate, react to, and profit from many ambient species. Folkbiological concepts are special players in cultural evolution, whose native stability attaches to more variable and difficult-to-learn representational forms, thus enhancing the latter's prospects for regularity and recurrence in transmission within and across cultures. This includes knowledge that cumulatively enriches (folk expertise), overrides (religious belief) or otherwise transcends (science) the commonsense ontology prescribed by folkbiology. Finally, the studies summarized here indicate that results gathered from “standard populations” in regard to biological categorization and reasoning more often than not fail to generalize in straightforward ways to humanity at large. This suggests the need for much more serious attention to cross-cultural research on basic cognitive processes. (shrink)

This book is an excellent and accessible overview of the position that children learn the meanings of words by applying a variety of nonlinguistic cognitive tools to the problem. We take issue with Bloom's emphasis on Theory of Mind as an explanatory mechanism for language learning; and with his claim that only unitary objects are nameable.

Memory is fleeting. New material rapidly obliterates previous material. How, then, can the brain deal successfully with the continual deluge of linguistic input? We argue that, to deal with this “Now-or-Never” bottleneck, the brain must compress and recode linguistic input as rapidly as possible. This observation has strong implications for the nature of language processing: (1) the language system must “eagerly” recode and compress linguistic input; (2) as the bottleneck recurs at each new representational level, the language system must build (...) a multilevel linguistic representation; and (3) the language system must deploy all available information predictively to ensure that local linguistic ambiguities are dealt with “Right-First-Time”; once the original input is lost, there is no way for the language system to recover. This is “Chunk-and-Pass” processing. Similarly, language learning must also occur in the here and now, which implies that language acquisition is learning to process, rather than inducing, a grammar. Moreover, this perspective provides a cognitive foundation for grammaticalization and other aspects of language change. Chunk-and-Pass processing also helps explain a variety of core properties of language, including its multilevel representational structure and duality of patterning. This approach promises to create a direct relationship between psycholinguistics and linguistic theory. More generally, we outline a framework within which to integrate often disconnected inquiries into language processing, language acquisition, and language change and evolution. (shrink)

According to classical arguments, language learning is both facilitated and constrained by cognitive biases. These biases are reflected in linguistic typology—the distribution of linguistic patterns across the world's languages—and can be probed with artificial grammar experiments on child and adult learners. Beginning with a widely successful approach to typology (Optimality Theory), and adapting techniques from computational approaches to statistical learning, we develop a Bayesian model of cognitive biases and show that it accounts for the detailed pattern of results (...) of artificial grammar experiments on noun-phrase word order (Culbertson, Smolensky, & Legendre, 2012). Our proposal has several novel properties that distinguish it from prior work in the domains of linguistic theory, computational cognitive science, and machine learning. This study illustrates how ideas from these domains can be synthesized into a model of language learning in which biases range in strength from hard (absolute) to soft (statistical), and in which language-specific and domain-general biases combine to account for data from the macro-level scale of typological distribution to the micro-level scale of learning by individuals. (shrink)

This article introduces a new model of the relationship between growth and learning and tests a set of hypotheses related to the development of adult competency using time allocation, anthropometric, and experimental task performance data collected between 1992 and 1997 in a multiethnic community in the Okavango Delta, Botswana. Building on seminal work in life history theory by Hawkes, Blurton Jones and associates, and Kaplan and associates, the punctuated development model presented here incorporates the effects of both growth and (...) learning constraints on age-specific task performance. In addition, the payoff to investment in two forms of embodied capital, growth-based and learning-based, are examined in relation to features of the socioecology, including subsistence economy and family composition.The three main findings are:The development of adult competency in specific tasks entails a steplike relationship between growth- and experience-based forms of embodied capital in the ontogeny of ability acquisition.There is a trade-off between the acquisition of experience-based embodied capital in the form of skills and knowledge and immediate productivity among children. Time allocation to these alternatives is primarily determined by the short- and long-term costs and benefits to parents of investment in children’s embodied capital.The availability of laborers and the overall labor requirements of the household are major determinants of investment in alternate forms of embodied capital and resulting variation in children’s time allocation. The value of children’s labor to their parents is dependent upon the opportunity costs to engaging in other activities not only for the child in question but also for potential substitute laborers.These results have important implications for our understanding of the role of growth and learning in the evolution of the human juvenile period, as well as for our understanding of cross-cultural variation in child growth and development and patterns of work and play. (shrink)

Events can be modeled through a geometric approach, representing event structures in terms of spaces and mappings between spaces. At least two spaces are needed to describe an event, an action space and a result space. In this article, we invoke general mathematical structures in order to develop this geometric perspective. We focus on three cognitive processes that are crucially involved in events: causal thinking, control of action and learning by generalization. These cognitive processes are supported by three corresponding (...) mathematical properties: monotonicity (that we relate to qualitative causal thinking and allows extrapolation); continuity (that plays a key role in our activities of action control); and convexity (that facilitates generalization and the categorization of events, and enables interpolation). We define how such properties constrain events representations and relate them to thinking about events. We discuss the relevance of the three constraints for event segmentation and explore the implications of such constraints for semantics. We conclude by a discussion that relates our approach to other accounts of events. (shrink)

Causal structure learning algorithms have focused on learning in ”batch-mode”: i.e., when a full dataset is presented. In many domains, however, it is important to learn in an online fashion from sequential or ordered data, whether because of memory storage constraints or because of potential changes in the underlying causal structure over the course of learning. In this paper, we present TDSL, a novel causal structure learning algorithm that processes data sequentially. This algorithm can track (...) changes in the generating causal structure or parameters, and requires significantly less memory in realistic settings. We show by simulation that the algorithm performs comparably to batch-mode learning when the causal structure is stationary, and significantly better in non-stationary environments. (shrink)

This volume features work on learning by researchers in various disciplines who share an interest in the systematic study of cognition and in the study of the formal and semantic aspects of language acquisition. A recurring theme is that language learning involves the acquisition of certain competencies and the formation of a system of beliefs which are significantly underdetermined by the linguistic and nonlinguistic inputs available to the learner. Theories of language learning must confront the epistemological problem (...) of how it is possible to induce and fixate a belief-system on the basis of exposure to limited data. A typical strategy in dealing with this problem has been to specify various types of formal and empirical constraints on linguistic and conceptual development in terms of specific hypotheses about the character of what is learned and about the kinds of resources and strategies available to the learner. Most of the contributions in this volume are concerned with the specification and evaluation of such constraints. (shrink)

Recent theoretical models suggest that the difference between human and nonhuman primate life-history patterns may be due to a reliance on complex foraging strategies requiring extensive learning. These models predict that children should reach adult levels of efficiency faster when foraging is cognitively simple. We test this prediction with data on Meriam fishing, spearfishing, and shellfishing efficiency. For fishing and spearfishing, which are cognitively difficult, we can find no significant amount of variability in return rates because of experiential factors (...) correlated with age. However, for shellfish collecting, which is relatively easy to learn, there are strong age-related effects on efficiency. Children reach adult efficiency more quickly in fishing as compared to shellfish collecting, probably owing to the size and strength constraints of the latter. (shrink)

It is widely assumed that language in some form or other originated by piggybacking on pre-existing learning mechanism not dedicated to language. Using evolutionary connectionist simulations, we explore the implications of such assumptions by determining the effect of constraints derived from an earlier evolved mechanism for sequential learning on the interaction between biological and linguistic adaptation across generations of language learners. Artificial neural networks were initially allowed to evolve “biologically” to improve their sequential learning abilities, after (...) which language was introduced into the population. We compared the relative contribution of biological and linguistic adaptation by allowing both networks and language to change over time. The simulation results support two main conclusions: First, over generations, a consistent head-ordering emerged due to linguistic adaptation. This is consistent with previous studies suggesting that some apparently arbitrary aspects of linguistic structure may arise from cognitive constraints on sequential learning. Second, when networks were selected to maintain a good level of performance on the sequential learning task, language learnability is significantly improved by linguistic adaptation but not by biological adaptation. Indeed, the pressure toward maintaining a high level of sequential learning performance prevented biological assimilation of linguistic-specific knowledge from occurring. (shrink)

We consider the problem of learning argumentation frameworks from a given set of labelings such that every input is a σ-labeling of these argumentation frameworks. Our new algorithm takes labelings and computes attack constraints for each argument that represent the restrictions on argumentation frameworks that are consistent with the input labelings. Having constraints on the level of arguments allows for a very effective parallelization of all computations. An important element of this approach is maintaining a representation of (...) all argumentation frameworks that satisfy the input labelings instead of simply finding any suitable argumentation framework. This is especially important, for example, if we receive additional labelings at a later time and want to refine our result without having to start all over again. The developed algorithm is compared to previous works and an evaluation of its performance has been conducted. (shrink)

Ethical perception involves seeing what is ethically salient about the particular details of the world. This kind of seeing is like informed judgment. It can be shaped by what we know and what we come to learn about, and by the development of moral virtue. I argue here that we can learn to see food justice, and I describe some ways to do so using three narrative case studies. The mechanism for acquiring this kind of vision is a “food justice (...) narrative” that is particular and concrete. These kinds of stories are counter narratives to a popular and dominant “script” about food that disguises the identity of people who eat, and obscures how constraints on free choice are created by particular lived circumstances. Food justice narratives specify the social and political location of individual people who are trying to nourish themselves. Once this contextual surround is included we are in a position to ask why this person, in this set of circumstances, is impeded in their access to nutritious food. This is not a question we are likely to consider if we leave out the identity of food consumers. Food justice narratives are forward looking as well because they bring into clearer focus what actions and kinds of social activism are appropriate responses to constraints on free choice. (shrink)

Complex constraints like conditionals ('If A, then B') and probabilistic constraints ('The probability that A is p') pose problems for Bayesian theories of learning. Since these propositions do not express constraints on outcomes, agents cannot simply conditionalize on the new information. Furthermore, a natural extension of conditionalization, relative information minimization, leads to many counterintuitive predictions, evidenced by the sundowners problem and the Judy Benjamin problem. Building on the notion of a `paradigm shift' and empirical research in (...) psychology and economics, I argue that the model of hypothesis testing can explain how people learn complex, theory-laden propositions like conditionals and probability constraints. Theories are formalized as probability distributions over a set of possible outcomes and theory change is triggered by a constraint which is incompatible with the initial theory. This leads agents to consult a higher order probability function, or a 'prior over priors,' to choose the most likely alternative theory which satisfies the constraint. The hypothesis testing model is applied to three examples: learning a simple probabilistic constraint involving coin bias, the sundowners problem for conditional learning, and the Judy Benjamin problem for learning conditional probability constraints. (shrink)

Languages are transmitted from person to person and generation to generation via a process of iterated learning: people learn a language from other people who once learned that language themselves. We analyze the consequences of iterated learning for learning algorithms based on the principles of Bayesian inference, assuming that learners compute a posterior distribution over languages by combining a prior (representing their inductive biases) with the evidence provided by linguistic data. We show that when learners sample languages (...) from this posterior distribution, iterated learning converges to a distribution over languages that is determined entirely by the prior. Under these conditions, iterated learning is a form of Gibbs sampling, a widely-used Markov chain Monte Carlo algorithm. The consequences of iterated learning are more complicated when learners choose the language with maximum posterior probability, being affected by both the prior of the learners and the amount of information transmitted between generations. We show that in this case, iterated learning corresponds to another statistical inference algorithm, a variant of the expectation-maximization (EM) algorithm. These results clarify the role of iterated learning in explanations of linguistic universals and provide a formal connection between constraints on language acquisition and the languages that come to be spoken, suggesting that information transmitted via iterated learning will ultimately come to mirror the minds of the learners. (shrink)