We argue that existing learning algorithms are often poorly equipped to solve problems involving a certain type of important and widespread regularity that we call “type-2 regularity.” The solution in these cases is to trade achieved representation against computational search. We investigate several ways in which such a trade-off may be pursued including simple incremental learning, modular connectionism, and the developmental hypothesis of “representational redescription.”.

Constructivism has been a key referent for research into the learning of science for several decades. There is little doubt that the research into learners’ ideas in science stimulated by the constructivist movement has been voluminous, and a great deal is now known about the way various science topics may commonly be understood by learners of various ages. Despite this significant research effort, there have been serious criticisms of this area of work: in terms of its philosophical underpinning, the validity (...) of its most popular constructs, the limited scope of its focus, and its practical value to science teaching. This paper frames this area of work as a Lakatosian Research Programme (RP), and explores the major criticisms of constructivism from that perspective. It is argued that much of the criticism may be considered as part of the legitimate academic debate expected within any active RP, i.e. arguments about the auxiliary theory making up the ‘protective belt’ of the programme. It is suggested that a shifting focus from constructivism to ‘contingency in learning’ will allow the RP to draw upon a more diverse range of perspectives, each consistent with the existing hard core of the programme, which will provide potentially fruitful directions for future work and ensure the continuity of a progressive RP into learning science. (shrink)

Some regularities enjoy only an attenuated existence in a body of training data. These are regularities whose statistical visibility depends on some systematic recoding of the data. The space of possible recodings is, however, infinitely large – it is the space of applicable Turing machines. As a result, mappings that pivot on such attenuated regularities cannot, in general, be found by brute-force search. The class of problems that present such mappings we call the class of “type-2 problems.” Type-1 problems, by (...) contrast, present tractable problems of search insofar as the relevant regularities can be found by sampling the input data as originally coded. Type-2 problems, we suggest, present neither rare nor pathological cases. They are rife in biologically realistic settings and in domains ranging from simple animat behaviors to language acquisition. Not only are such problems rife – they are standardly solved! This presents a puzzle. How, given the statistical intractability of these type-2 cases, does nature turn the trick? One answer, which we do not pursue, is to suppose that evolution gifts us with exactly the right set of recoding biases so as to reduce specific type-2 problems to type-1 mappings. Such a heavy-duty nativism is no doubt sometimes plausible. But we believe there are other, more general mechanisms also at work. Such mechanisms provide general strategies for managing problems of type-2 complexity. Several such mechanisms are investigated. At the heart of each is a fundamental ploy – namely, the maximal exploitation of states of representation already achieved by prior, simpler learning so as to reduce the amount of subsequent computational search. Such exploitation both characterizes and helps make unitary sense of a diverse range of mechanisms. These include simple incremental learning, modular connectionism, and the developmental hypothesis of “representational redescription”. In addition, the most distinctive features of human cognition – language and culture – may themselves be viewed as adaptations enabling this representation/computation trade-off to be pursued on an even grander scale. (shrink)

Our research addresses the issue of teaching and learning concepts in science education as an empirical question. We study the process of conceptualization by closely examining the unfolding of classroom lesson sequences. We situate our work within the practice turn line of research on epistemic practices in science education. We also adopt a practice turn approach when it comes to the learning of concepts, as we consider conceptualization as being inherent within epistemic practices. In our work, pedagogical practices are modeled (...) as learning games and epistemic practices in science education are characterized as enacted epistemic games emerging through the unfolding of learning games. Science practices are modeled as source epistemic games since they are the source of the knowledge at stake in pedagogical practices. From this point, we examine closely how playing learning games can enable students to play enacted epistemic games and then in turn the source epistemic games at the core of conceptual understanding. Thus, the main contribution of this paper is to link pedagogical practices to epistemic practices in science education and to science practices in general. Our method is consistent with this epistemological framework as our case study on the concept of earthquakes in a 5th grade classroom sequence illustrates. Following an investigation of two experienced teachers and their classes during a teaching unit, our analysis shows how teaching effectiveness is determined by a dialectic. This entails on the one hand a didactic continuity between learning games and enacted epistemic games and, on the other, an epistemic continuity between enacted epistemic games and source epistemic games. (shrink)

In this article we focus on the concept of concept in conceptual change. We argue that (1) theories of higher learning must often employ two different notions of concept that should not be conflated: psychological and scientific concepts. The usages for these two notions are partly distinct and thus straightforward identification between them is unwarranted. Hence, the strong analogy between scientific theory change and individual learning should be approached with caution. In addition, we argue that (2) research in psychology and (...) cognitive science provides a promising theoretical basis for developing explanatory mechanistic models of conceptual change. Moreover, we argue that (3) arguments against deeper integration between the fields of psychology and conceptual change are not convincing, and that recent theoretical developments in the cognitive sciences might prove indispensable in filling in the details in mechanisms of conceptual change. (shrink)

This paper explores the relation between scientific knowledge and common sense intuitions as a complement to Hoyningen-Huene’s account of systematicity. On one hand, Hoyningen-Huene embraces continuity between these in his characterization of scientific knowledge as an extension of everyday knowledge, distinguished by an increase in systematicity. On the other, he argues that scientific knowledge often comes to deviate from common sense as science develops. Specifically, he argues that a departure from common sense is a price we may have to pay (...) for increased systematicity. I argue that to clarify the relation between common sense and scientific reasoning, more attention to the cognitive aspects of learning and doing science is needed. As a step in this direction, I explore the potential for cross-fertilization between the discussions about conceptual change in science education and philosophy of science. Particularly, I examine debates on whether common sense intuitions facilitate or impede scientific reasoning. While contending that these debates can balance some of the assumptions made by Hoyningen-Huene, I suggest that a more contextualized version of systematicity theory could supplement cognitive analysis by clarifying important organizational aspects of science. (shrink)

This work uses microgenetic study of classroom learning to illuminate (1) the role of pre-instructional student knowledge in the construction of normative scientific knowledge, and (2) the learning mechanisms that drive change. Three enactments of an instructional sequence designed to lead to a scientific understanding of thermal equilibration are used as data sources. Only data from a scaffolded student inquiry preceding introduction of a normative model were used. Hence, the study involves nearly autonomous student learning. In two classes, students developed (...) stable and socially shared explanations (“causal schemes”) for understanding thermal equilibration. One case resulted in a near-normative understanding, while the other resulted in a non-normative “alternative conception.” The near-normative case seems to be a particularly clear example wherein the constructed causal scheme is a composition of previously documented naïve conceptions. Detailed prior description of these naive elements allows a much better than usual view of the corresponding details of change during construction of the new scheme. A list of candidate mechanisms that can account for observed change is presented. The non-normative construction seems also to be a composition, albeit of a different structural form, using a different (although similar) set of naïve elements. This article provides one of very few high-resolution process analyses showing the productive use of naïve knowledge in learning. (shrink)

This article aims to contribute to the literature on conceptual change by engaging in direct theoretical and empirical comparison of contrasting views. We take up the question of whether naïve physical ideas are coherent or fragmented, building specifically on recent work supporting claims of coherence with respect to the concept of force by Ioannides and Vosniadou [Ioannides, C., & Vosniadou, C. (2002). The changing meanings of force. Cognitive Science Quarterly 2, 5–61]. We first engage in a theoretical inquiry on the (...) nature of coherence and fragmentation, concluding that these terms are not well‐defined, and proposing a set of issues that may be better specified. The issues have to do with contextuality, which concerns the range of contexts in which a concept (meaning, model, theory) applies, and relational structure, which is how elements of a concept (meaning, model, or theory) relate to one another. We further propose an enhanced theoretical and empirical accountability for what and how much one needs to say in order to have specified a concept. Vague specification of the meaning of a concept can lead to many kinds of difficulties.Empirically, we conducted two studies. A study patterned closely on Ioannides and Vosniadou's work (which we call a quasi‐replication) failed to confirm their operationalizations of “coherent.” An extension study, based on a more encompassing specification of the concept of force, showed three kinds of results: (1) Subjects attend to more features than mentioned by Ioannides and Vosniadou, and they changed answers systematically based on these features; (2)We found substantial differences in the way subjects thought about the new contexts we asked about, which undermined claims for homogeneity within even the category of subjects (having one particular meaning associated with “force”) that best survived our quasi‐replication; (3) We found much reasoning of subjects about forces that cannot be accounted for by the meanings specified by Ioannides and Vosniadou. All in all, we argue that, with a greater attention to contextuality and with an appropriately broad specification of the meaning of a concept like force, Ioannides and Vosniadou's claims to have demonstrated coherence seem strongly undermined. Students' ideas are not random and chaotic; but neither are they simply described and strongly systematic. (shrink)

Surveys of public opinions on climate change found that a majority of American respondents regarded global warming as a critical or an important threat . Given this consensus, one might expect that a majority of Americans are ready to take immediate action to deal with the environmental crisis. However, when they were asked whether we should begin taking steps now, only 43% of American respondents said yes; 54% of them chose either the option “until we are sure that global warming (...) is really a problem, we should not take any steps that would have economic costs,” or the option “global warming should be addressed, but its effects will be gradual, so we can deal with the problem .. (shrink)

: This paper offers a preliminary analysis of conceptual change between event concepts. It begins with a brief review of the major findings of cognitive studies on event knowledge. The script model proposed by Schank and Abelson was the first attempt to represent event knowledge. Subsequent cognitive studies indicated that event knowledge is organized in the form of dimensional organizations in which temporally successive actions are related causally. This paper proposes a frame representation to capture and outline the internal structure (...) of event concepts, in particular, their causal connections. The frame representation offers an effective method to analyze the relations between event concepts, and to expose the unique cognitive mechanisms behind conceptual change involved event concepts. Finally this paper shows that the frame representation of event concepts is instrumental to understanding an important historical episode of conceptual change in the context of nineteenth-century optics. (shrink)

I propose a new perspective on the study of scientific revolutions. This is a transformation from an object-only perspective to an ontological perspective that properly treats objects and processes as distinct kinds. I begin my analysis by identifying an object bias in the study of scientific revolutions, where it takes the form of representing scientific revolutions as changes in classification of physical objects. I further explore the origins of this object bias. Findings from developmental psychology indicate that children cannot distinguish (...) processes from objects until the age of 7, but they have already developed a core system of object knowledge as early as 4 months of age. The persistence of this core system is responsible for the object bias among mature adults, i.e., the tendency to apply knowledge of physical objects to temporal processes. In light of the distinction between physical objects and temporal processes, I redraw the picture of the Copernican revolution. Rather than seeing it as a taxonomic shift from a geocentric to a heliocentric cosmology, we should understand it as a transformation from a conceptual system that was built around an object concept to one that was built around a process concept. (shrink)

In this paper I examine a cognitive mechanism of incommensurability. Using the frame model of concept representation to capture structural relations within concepts, I reveal an ontological difference between object and event concepts: the former are spatial but the latter temporal. Experiments from cognitive sciences further demonstrate that the mind treats object and event concepts differently. Thus, incommensurability can occur in conceptual change across different ontological categories. I use a historical case to illustrate how the ontological difference between an object (...) and an event concept actually caused incommensurability in the context of nineteenth‐century optics. The cognitive and historical analyses indicate that incommensurability can be a local phenomenon and does not necessarily imply incomparability. (shrink)

Historians and philosophers of science have, in recent decades, offered evidence in support of several influential models of conceptual change in science. These models have often drawn on and in turn driven research on conceptual change in childhood and in science education. This nexus of reciprocal influences is held together by several largely unexamined analogies and by several assumptions concerning analogy itself. In this chapter, we aim to shed some light on these hidden premises and subject them to critical scrutiny. (...) Our critical survey suggests the following hypothesis: conceptual change in both childhood and science is driven by the epistemic subject’s quest for coherence. (shrink)

The focus of the current study is the individual and environmental attributes of inventiveness among children and adolescents. Research was conducted on the young inventors who were part of a nation-wide inventive ideation contest for children and adolescents in P. R. China. A total of 621 4th to 12th grade students from 112 schools all over China participated in the study. Among them, 38 reported holding one or more patents. Independent t-test showed, compared to their lower-level counterparts, higher-level young inventors (...) were more intrinsically motivated for inventive endeavours and were more open to new experiences. They also reported more encouragement and resources for invention from their schools. Logistic regression showed that school encouragement made the major contribution in discriminating these two groups. 2×3 MANONA revealed a significant main effect of gender and age group but no significant interaction between the two factors. Results of the univariate tests challenged the stereotyped view against the inventive ability of girls. Girls scored higher in Openness and lower in executive thinking style. The aesthetic appeal of their inventive products was also rated higher by experts. Albeit this superiority, girls, however, reported less encouragement from their parents to make inventions. Results of the cross-sectional study of the different age groups did not support a hypothesized growth of inventiveness from the lower to the higher grades. Instead, an uneven developmental pattern in inventiveness and the relevant domains were revealed, which was on the large part attributed to the influence of the educational environment. Taken together, the results of the current study highlight the important role the environmental factors play in fostering or hindering the development of inventiveness among children and adolescents. (shrink)