We were encouraged by the broad enthusiasm for building machines that learn and think in more human-like ways. Many commentators saw our set of key ingredients as helpful, but there was disagreement regarding the origin and structure of those ingredients. Our response covers three main dimensions of this disagreement: nature versus nurture, coherent theories versus theory fragments, and symbolic versus sub-symbolic representations. These dimensions align with classic debates in artificial intelligence and cognitive science, although, rather than embracing these debates, we (...) emphasize ways of moving beyond them. Several commentators saw our set of key ingredients as incomplete and offered a wide range of additions. We agree that these additional ingredients are important in the long run and discuss prospects for incorporating them. Finally, we consider some of the ethical questions raised regarding the research program as a whole. (shrink)

Recent progress in artificial intelligence has renewed interest in building systems that learn and think like people. Many advances have come from using deep neural networks trained end-to-end in tasks such as object recognition, video games, and board games, achieving performance that equals or even beats that of humans in some respects. Despite their biological inspiration and performance achievements, these systems differ from human intelligence in crucial ways. We review progress in cognitive science suggesting that truly human-like learning and thinking (...) machines will have to reach beyond current engineering trends in both what they learn and how they learn it. Specifically, we argue that these machines should build causal models of the world that support explanation and understanding, rather than merely solving pattern recognition problems; ground learning in intuitive theories of physics and psychology to support and enrich the knowledge that is learned; and harness compositionality and learning-to-learn to rapidly acquire and generalize knowledge to new tasks and situations. We suggest concrete challenges and promising routes toward these goals that can combine the strengths of recent neural network advances with more structured cognitive models. (shrink)

In this paper, I present an alternative argument for Jerry Fodor's recent conclusion that there are currently no tenable theories of concepts in the cognitive sciences and in the philosophy of mind. Briefly, my approach focuses on the 'theory-theory' of concepts. I argue that the two ways in which cognitive psychologists have formulated this theory lead to serious difficulties, and that there cannot be, in principle, a third way in which it can be reformulated. Insofar as the 'theory-theory' is supposed (...) to replace, and to rectify the problems of, the earlier 'classical' and 'probabilistic' theories, its failure confirms Fodor's original observation. Since my critique does not rest on controversial philosophical assumptions and is readily available from within the cognitive sciences, it is a stronger argument than Fodor's. (shrink)

We used a new method to assess how people can infer unobserved causal structure from patterns of observed events. Participants were taught to draw causal graphs, and then shown a pattern of associations and interventions on a novel causal system. Given minimal training and no feedback, participants in Experiment 1 used causal graph notation to spontaneously draw structures containing one observed cause, one unobserved common cause, and two unobserved independent causes, depending on the pattern of associations and interventions they saw. (...) We replicated these findings with less‐informative training (Experiments 2 and 3) and a new apparatus (Experiment 3) to show that the pattern of data leads to hidden causal inferences across a range of prior constraints on causal knowledge. (shrink)

Structural and functional descriptions of technical artefacts play an important role in engineering practice. A complete description of a technical artefact involves a description of both functional and structural features. Engineers, moreover, assume that there is an intimate relationship between the function and structure of technical artefacts and they reason from functional properties to structural ones and vice versa. This raises the question of how structural and functional descriptions are related. The kind of inference patterns that establish coherence between structural (...) and functional descriptions are explored in this paper, using the analysis of coherence creating relations of Thagard et al. Explanatory, analogical and practical inference patterns are discussed and it is argued that of these three, practical inferences may be the most important. Practical inferences, however, cannot provide a full underpinning of the coherence of structural and functional descriptions of technical artefacts. The paper ends with the suggestion that any account of the coherence of the structural and functional descriptions of technical artefacts must involve reference to their intentional features.Keywords: Technical artefact; Structural description; Functional description; Coherence relation; Practical inference. (shrink)

Model-based learning has an established position within science education. It has been found to enhance conceptual understanding and provide a way for engaging students in authentic scientific activity. Despite ample research, few studies have examined the cognitive processes regarding learning scientific concepts within MBL. On the other hand, recent research within cognitive science has examined the learning of so-called relational categories. Relational categories are categories whose membership is determined on the basis of the common relational structure. In this theoretical paper, (...) I argue that viewing models as relational categories provides a well-motivated cognitive basis for MBL. I discuss the different roles of models and modeling within MBL and discern the related cognitive aspects brought forward by the reinterpretation of models as relational categories. I will argue that relational knowledge is vital in learning novel models and in the transfer of learning. Moreover, relational knowledge underlies the coherent, hierarchical knowledge of experts. Lastly, I will examine how the format of external representations may affect the learning of models and the relevant relations. The nature of the learning mechanisms underlying students’ mental representations of models is an interesting open question to be examined. Furthermore, the ways in which the expert-like knowledge develops and how to best support it is in need of more research. The discussion and conceptualization of models as relational categories allows discerning students’ mental representations of models in terms of evolving relational structures in greater detail than previously done. (shrink)

It has often been suggested that people’s ordinary capacities for understanding the world make use of much the same methods one might find in a formal scientific investigation. A series of recent experimental results offer a challenge to this widely-held view, suggesting that people’s moral judgments can actually influence the intuitions they hold both in folk psychology and in causal cognition. The present target article distinguishes two basic approaches to explaining such effects. One approach would be to say that the (...) relevant competencies are entirely non-moral but that some additional factor (conversational pragmatics, performance error, etc.) then interferes and allows people’s moral judgments to affect their intuitions. Another approach would be to say that moral considerations truly do figure in workings of the competencies themselves. It is argued that the data available now favor the second of these approaches over the first. (shrink)

Five experiments provide evidence for a class of ‘dual character concepts.’ Dual character concepts characterize their members in terms of both (a) a set of concrete features and (b) the abstract values that these features serve to realize. As such, these concepts provide two bases for evaluating category members and two different criteria for category membership. Experiment 1 provides support for the notion that dual character concepts have two bases for evaluation. Experiments 2-4 explore the claim that dual character concepts (...) have two different criteria for category membership. The results show that when an object possesses the appropriate concrete features, but does not fulfill the appropriate abstract value, it is judged to be a category member in one sense but not in another. Finally, Experiment 5 uses the theory developed here to construct artificial dual character concepts and examines whether participants react to these artificial concepts in the same way as naturally occurring dual character concepts. The present studies serve to define the nature of dual character concepts and distinguish them from other types of concepts (e.g., natural kind concepts), which share some, but not all of the properties of dual character concepts. More broadly, these phenomena suggest a normative dimension in everyday conceptual representation. (shrink)

Two main theories of concepts have emerged in the recent psychological literature: the Prototype Theory (which considers concepts to be self-contained lists of features) and the Theory Theory (which conceives of them as being embedded within larger theoretical networks). Experiments supporting the first theory usually differ substantially from those supporting the second, which suggests that these the· ories may be operating at different levels of explanation and dealing with different entities. A convergence is proposed between the Theory Theory and the (...) intentional stance in the philosophy of language and mind. From this stance, concepts should not be thought of as concrete physical entities. (shrink)

Theories of relational concept acquisition based on structured intersection discovery predict that relational concepts with a probabilistic structure ought to be extremely difficult to learn. We report four experiments testing this prediction by investigating conditions hypothesized to facilitate the learning of such categories. Experiment 1 showed that changing the task from a category-learning task to choosing the “winning” object in each stimulus greatly facilitated participants' ability to learn probabilistic relational categories. Experiments 2 and 3 further investigated the mechanisms underlying this (...) “who's winning” effect. Experiment 4 replicated and generalized the “who's winning” effect with more natural stimuli. Together, our findings suggest that people learn relational concepts by a process of intersection discovery akin to schema induction, and that any task that encourages people to discover a higher order relation that remains invariant over members of a category will facilitate the learning of putatively probabilistic relational concepts. (shrink)

Humans are poorer at identifying smells and communicating about them, compared to other sensory domains. They also cannot easily organize odor sensations in a general conceptual space, where geometric distance could represent how similar or different all odors are. These two generalities are more or less accepted by psychologists, and they are often seen as connected: If there is no conceptual space for odors, then olfactory identification should indeed be poor. We propose here an important revision to this conclusion: We (...) believe that the claim that there is no odor space is true only if by odor space, one means a conceptual space representing all possible odor sensations, in the paradigmatic sense used for instance for color. However, in a less paradigmatic sense, local conceptual spaces representing a given subset of odors do exist. Thus the absence of a global odor space does not warrant the conclusion that there is no olfactory conceptual map at all. Here we show how a localist account provides a new interpretation of experts and cross‐cultural categorization studies: Rather than being exceptions to the poor olfactory identification and communication usually seen elsewhere, experts and cross‐cultural categorization are here taken to corroborate the existence of local conceptual spaces. (shrink)

Humans are poorer at identifying smells and communicating about them, compared to othersensory domains. They also cannot easily organise odour sensations in a general conceptual space like with colours. We challenge the conclusion that there is no olfactory conceptual map at all. Instead we propose a new framework, with local conceptual spaces.

Research in education and cognitive development suggests that explaining plays a key role in learning and generalization: When learners provide explanations—even to themselves—they learn more effectively and generalize more readily to novel situations. This paper proposes and tests a subsumptive constraints account of this effect. Motivated by philosophical theories of explanation, this account predicts that explaining guides learners to interpret what they are learning in terms of unifying patterns or regularities, which promotes the discovery of broad generalizations. Three experiments provide (...) evidence for the subsumptive constraints account: prompting participants to explain while learning artificial categories promotes the induction of a broad generalization underlying category membership, relative to describing items (Exp. 1), thinking aloud (Exp. 2), or free study (Exp. 3). Although explaining facilitates discovery, Experiment 1 finds that description is more beneficial for learning item details. Experiment 2 additionally suggests that explaining anomalous observations may play a special role in belief revision. The findings provide insight into explanation’s role in discovery and generalization. (shrink)

Conviction Narrative Theory (CNT) is a theory of choice underradical uncertainty– situations where outcomes cannot be enumerated and probabilities cannot be assigned. Whereas most theories of choice assume that people rely on (potentially biased) probabilistic judgments, such theories cannot account for adaptive decision-making when probabilities cannot be assigned. CNT proposes that people usenarratives– structured representations of causal, temporal, analogical, and valence relationships – rather than probabilities, as the currency of thought that unifies our sense-making and decision-making faculties. According to CNT, (...) narratives arise from the interplay between individual cognition and the social environment, with reasoners adopting a narrative that feels “right” to explain the available data; using that narrative to imagine plausible futures; and affectively evaluating those imagined futures to make a choice. Evidence from many areas of the cognitive, behavioral, and social sciences supports this basic model, including lab experiments, interview studies, and econometric analyses. We identify 12 propositions to explain how the mental representations (narratives) interact with four inter-related processes (explanation, simulation, affective evaluation, and communication), examining the theoretical and empirical basis for each. We conclude by discussing how CNT can provide a common vocabulary for researchers studying everyday choices across areas of the decision sciences. (shrink)

Vector-based models of word meaning have become increasingly popular in cognitive science. The appeal of these models lies in their ability to represent meaning simply by using distributional information under the assumption that words occurring within similar contexts are semantically similar. Despite their widespread use, vector-based models are typically directed at representing words in isolation, and methods for constructing representations for phrases or sentences have received little attention in the literature. This is in marked contrast to experimental evidence (e.g., in (...) sentential priming) suggesting that semantic similarity is more complex than simply a relation between isolated words. This article proposes a framework for representing the meaning of word combinations in vector space. Central to our approach is vector composition, which we operationalize in terms of additive and multiplicative functions. Under this framework, we introduce a wide range of composition models that we evaluate empirically on a phrase similarity task. (shrink)

Although mind‐wandering research is rapidly progressing, stark disagreements are emerging about what the term “mind‐wandering” means. Four prominent views define mind‐wandering as (a) task‐unrelated thought, (b) stimulus‐independent thought, (c) unintentional thought, or (d) dynamically unguided thought. Although theorists claim to capture the ordinary understanding of mind‐wandering, no systematic studies have assessed these claims. Two large factorial studies present participants (N = 545) with vignettes that describe someone's thoughts and ask whether her mind was wandering, while systematically manipulating features relevant to (...) the four major accounts of mind‐wandering. Dynamics explains between four and 40 times more variance in participants' mind‐wandering judgments than other features. Our third study (N = 153) tests and supports a unique prediction of the dynamic framework—obsessive rumination contrasts with mind‐wandering. Our final study (N = 277) used vignettes that resemble mind‐wandering experiments. Dynamics had significant and large effects, while task‐unrelatedness was nonsignificant. These results strongly suggest that the central feature of mind‐wandering is its dynamics. (shrink)

There is today much interest in research of neuronal substrata in metaphor processing. It has been suggested that the right hemisphere yields a key role in the comprehension of figurative language and, particularly, in metaphors. Figurative language is included in pragmatics, a branch of linguistics that researches the use of language, in opposition to the study of the system of language. There lingers, though, an open debate in respect to the identification of the specific aspects concerning semantics, as opposed to (...) those dominated by pragmatics. Can studies from neuronal correlates clarify questions that relate to semantics/pragmatics representation? I shall analyze neuroscientific developments about implicit language to attempt to understand, in section 2, scientific techniques available and more suitable to the phenomenology of the act of understanding an implicit, figurative or implicated message in a certain language game. To do so, I shall start by reviewing the studies in philosophy of language, and accommodate the development of the research in pragmatics underlying metaphor, particularly, in Philosophical Investigations by Wittgenstein. Finally, I discuss the possibility of interpretative capabilities being socioculturally grounded. I expect this methodological analysis to contribute to the enlightenment of the problem of phenomenology of intersubjective pragmatics, and to its future experimental paradigms. (shrink)