Numerous studies have found that selective attention affects category learning. However, previous research did not distinguish between the contribution of focusing and filtering components of selective attention. This study addresses this issue by examining how components of selective attention affect category representation. Participants first learned a rule‐plus‐similarity category structure, and then were presented with category priming followed by categorization and recognition tests. Additionally, to evaluate the involvement of focusing and filtering, we fit models with different attentional mechanisms to the data. (...) In Experiment 1, participants received rule‐based category training, with specific emphasis on a single deterministic feature (D feature). Experiment 2 added a recognition test to examine participants’ memory for features. Both experiments indicated that participants categorized items based solely on the D feature, showed greater memory for the D feature, were primed exclusively by the D feature without interference from probabilistic features (P features), and were better fit by models with focusing and at least one type of filtering mechanism. The results indicated that selective attention distorted category representation by highlighting the D feature and attenuating P features. To examine whether the distorted representation was specific to rule‐based training, Experiment 3 introduced training, emphasizing all features. Under such training, participants were no longer primed by the D feature, they remembered all features well, and they were better fit by the model assuming only focusing but no filtering process. The results coupled with modeling provide novel evidence that while both focusing and filtering contribute to category representation, filtering can also result in representational distortion. (shrink)

General principles of human cognition can help to explain why languages are more likely to have certain characteristics than others: structures that are difficult to process or produce will tend to be lost over time. One aspect of cognition that is implicated in language use is working memory—the component of short‐term memory used for temporary storage and manipulation of information. In this study, we consider the relationship between working memory and regularization of linguistic variation. Regularization is a well‐documented process whereby (...) languages become less variable (on some dimension) over time. This process has been argued to be driven by the behavior of individual language users, but the specific mechanism is not agreed upon. Here, we use an artificial language learning experiment to investigate whether limitations in working memory during either language learning or language production drive regularization behavior. We find that taxing working memory during production results in the loss of all types of variation, but the process by which random variation becomes more predictable is better explained by learning biases. A computational model offers a potential explanation for the production effect using a simple self‐priming mechanism. (shrink)

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

In this chapter, we outline the range of argument forms involving causation that can be found in everyday discourse. We also survey empirical work concerned with the generation and evaluation of such arguments. This survey makes clear that there is presently no unified body of research concerned with causal argument. We highlight the benefits of a unified treatment both for those interested in causal cognition and those interested in argumentation, and identify the key challenges that must be met for a (...) full understanding of causal argumentation. (shrink)

This chapter subsumes David Marr’s levels of analysis account of explanation in cognitive science under the framework of mechanistic explanation: Answering the questions that define each one of Marr’s three levels is tantamount to describing the component parts and operations of mechanisms, as well as their organization, behavior, and environmental context. By explicating these questions and showing how they are answered in several different cognitive science research programs, this chapter resolves some of the ambiguities that remain in Marr’s account, and (...) shows that many different areas and traditions of cognitive scientific research can be unified under the mechanistic framework. (shrink)

Though individual categorization or decision processes have been studied separately in many previous investigations, few studies have investigated how they interact by using a two-stage task of first categorizing and then deciding. To address this issue, we investigated a categorization-decision task in two experiments. In both, participants were shown six faces varying in width, first asked to categorize the faces, and then decide a course of action for each face. Each experiment was designed to include three groups, and for each (...) group, we manipulated the probabilistic contingencies between stimulus, category assignments, and decision consequences. For each group, each participant received three different sequences of category response, category feedback, decision response, and decision feedback. We found that participants were only partially responsive in the appropriate directions to the contingencies assigned to each group. Comparisons of results from different sequences provided evidence for empirical interference effects of categorization on decisions. The empirical interference effect is defined as the difference between the probability of taking a hostile action in decision-alone conditions and the total probability of taking a hostile action in categorization-decision conditions. To test competing accounts for multiple empirical results, including two-stage choice probabilities and empirical interference effects, we compared a quantum cognition model versus a two-stage exemplar categorization model at both aggregate and individual levels. Using a Bayesian information criterion, we found that the quantum model provided an overall better model fit than the exemplar model. Although both models predicted empirical interference effects, the exemplar model was able to generate probabilistic deviation by incorporating category information of the first stage into the feature representation of the subsequent decision stage, while the quantum model produced interference effect by superposition, measurement, and quantum entanglement. (shrink)

Bayesian reverse-engineering is a research strategy for developing three-level explanations of behavior and cognition. Starting from a computational-level analysis of behavior and cognition as optimal probabilistic inference, Bayesian reverse-engineers apply numerous tweaks and heuristics to formulate testable hypotheses at the algorithmic and implementational levels. In so doing, they exploit recent technological advances in Bayesian artificial intelligence, machine learning, and statistics, but also consider established principles from cognitive psychology and neuroscience. Although these tweaks and heuristics are highly pragmatic in character and (...) are often deployed unsystematically, Bayesian reverse-engineering avoids several important worries that have been raised about the explanatory credentials of Bayesian cognitive science: the worry that the lower levels of analysis are being ignored altogether; the challenge that the mathematical models being developed are unfalsifiable; and the charge that the terms ‘optimal’ and ‘rational’ have lost their customary normative force. But while Bayesian reverse-engineering is therefore a viable and productive research strategy, it is also no fool-proof recipe for explanatory success. (shrink)

The extent to which we generalize a novel property from a sample of familiar instances to novel instances depends on the sample composition. Previous property induction experiments have only used samples consisting of novel types (unique entities). Because real‐world evidence samples often contain redundant tokens (repetitions of the same entity), we studied the effects on property induction of adding types and tokens to an observed sample. In Experiments 1–3, we presented participants with a sample of birds or flowers known to (...) have a novel property and probed whether this property generalized to novel items varying in similarity to the initial sample. Increasing the number of novel types (e.g., new birds with the target property) in a sample produced tightening, promoting property generalization to highly similar stimuli but decreasing generalization to less similar stimuli. On the other hand, increasing the number of tokens (e.g., repeated presentations of the same bird with the target property) had little effect on generalization. Experiment 4 showed that repeated tokens are encoded and can benefit recognition, but appear to be given little weight when inferring property generalization. We modified an existing Bayesian model of induction (Navarro, Dry, & Lee, 2012) to account for both the information added by new types and the discounting of information conveyed by tokens. (shrink)

The sound system of a language must be able to support a perceptual contrast between different words in order to signal communicatively relevant meaning distinctions. In this paper, we use a simple agent-based exemplar model in which the evolution of sound-category systems is understood as a co-evolutionary process, where the range of variation within sound categories is constrained by functional pressure to keep different words perceptually distinct. We show that this model can reproduce several observed effects on the range of (...) sound variation. We argue that phonological systems can be seen as finding a relative optimum of variation: Efficient communication is sustained while at the same time, hidden category variation provides pathways for future evolution. (shrink)

The ability to learn cause–effect relations from experience is critical for humans to behave adaptively — to choose causes that bring about desired effects. However, traditional experiments on experience-based learning involve events that are artificially compressed in time so that all learning occurs over the course of minutes. These paradigms therefore exclusively rely upon working memory. In contrast, in real-world situations we need to be able to learn cause–effect relations over days and weeks, which necessitates long-term memory. 413 participants completed (...) a smartphone study, which compared learning a cause–effect relation one trial per day for 24 days versus the traditional paradigm of 24 trials back- to- back. Surprisingly, we found few differences between the short versus long timeframes. Subjects were able to accurately detect generative and preventive causal relations, and they exhibited illusory correlations in both the short and long timeframe tasks. These results provide initial evidence that experience-based learning over long timeframes exhibits similar strengths and weaknesses as in short timeframes. However, learning over long timeframes may become more impaired with more complex tasks. (shrink)

There is an active debate within the field of phonology concerning the cognitive status of substantive phonetic factors such as ease of articulation and perceptual distinctiveness. A new framework is proposed in which substance acts as a bias, or prior, on phonological learning. Two experiments tested this framework with a method in which participants are first provided highly impoverished evidence of a new phonological pattern, and then tested on how they extend this pattern to novel contexts and novel sounds. Participants (...) were found to generalize velar palatalization (e.g., the change from [k] as in keep to [t⌢∫S] as in cheap) in a way that accords with linguistic typology, and that is predicted by a cognitive bias in favor of changes that relate perceptually similar sounds. Velar palatalization was extended from the mid front vowel context (i.e., before [e] as in cape) to the high front vowel context (i.e., before [i] as in keep), but not vice versa. The key explanatory notion of perceptual similarity is quantified with a psychological model of categorization, and the substantively biased framework is formalized as a conditional random field. Implications of these results for the debate on substance, theories of phonological generalization, and the formalization of similarity are discussed. (shrink)

Cognitive scientists have long used distributional semantic representations of categories. The predominant approach uses distributional representations of category‐denoting nouns, such as “city” for the category city. We propose a novel scheme that represents categories as prototypes over representations of names of its members, such as “Barcelona,” “Mumbai,” and “Wuhan” for the category city. This name‐based representation empirically outperforms the noun‐based representation on two experiments (modeling human judgments of category relatedness and predicting category membership) with particular improvements for ambiguous nouns. We (...) discuss the model complexity of both classes of models and argue that the name‐based model has superior explanatory potential with regard to concept acquisition. (shrink)

Traditionally, theories of concepts in psychology assume that concepts are a single, uniform kind of mental representation. But no single kind of representation can explain all of the empirical data for which concepts are responsible. I argue that the assumption that concepts are uniformly the same kind of mental structure is responsible for these theories’ shortcomings, and outline a pluralist theory of concepts that rejects this assumption. On pluralism, concepts should be thought of as being constituted by multiple representational kinds, (...) with the particular kind of concept used on an occasion being determined by the context. I argue that endorsing pluralism does not lead to eliminativism about concepts as an object of scientific interest. (shrink)

Certain of our concepts are innate, but many others are learned. Despite the plausibility of this claim, some have argued that the very idea of concept learning is incoherent. I present a conception of learning that sidesteps the arguments against the possibility of concept learning, and sketch several mechanisms that result in the generation of new primitive concepts. Given the rational considerations that motivate their deployment, I argue that these deserve to be called learning mechanisms. I conclude by replying to (...) the objections that these mechanisms cannot produce genuinely new content and cannot be part of genuinely cognitive explanations. (shrink)

Mechanistic explanation has an impressive track record of advancing our understanding of complex, hierarchically organized physical systems, particularly biological and neural systems. But not every complex system can be understood mechanistically. Psychological capacities are often understood by providing cognitive models of the systems that underlie them. I argue that these models, while superficially similar to mechanistic models, in fact have a substantially more complex relation to the real underlying system. They are typically constructed using a range of techniques for abstracting (...) the functional properties of the system, which may not coincide with its mechanistic organization. I describe these techniques and show that despite being non-mechanistic, these cognitive models can satisfy the normative constraints on good explanations. (shrink)

The renewed interest in concepts and their role in psychological theorizing is partially motivated by Machery’s claim that concepts are so heterogeneous that they have no explanatory role. Against this, pluralism argues that there is multiplicity of different concepts for any given category, while hybridism argues that a concept is constituted by a rich common representation. This article aims to advance the understanding of the hybrid view of concepts. First, we examine the main arguments against hybrid concepts and conclude that, (...) even if not successful, they challenge hybridism to find a robust criterion for concept individuation and to show an explanatory advantage for hybrid concepts. Then we propose such a criterion of individuation, which we will call ‘functional stable coactivation’. Finally, we examine the prospects of hybridism to understand what is involved in recent approaches to categorization and meaning extraction. 1 The Heterogeneity of Conceptual Representations2 Two Challenges for Hybrid Concepts: Individuation and Explanation2.1 The coordination criterion2.2 Concepts as constituents of thoughts3 Individuating Hybrids: Functional Stable Coactivation4 The Explanatory Power of Hybrid Concepts4.1 Categorization4.2 Meaning extraction4.2.1 Linguistic comprehension and rich lexical entries4.2.2 Polysemy and hybrid concepts5 Conclusion. (shrink)

The last two decades have produced unprecedented successes in the fields of artificial intelligence and machine learning (ML), due almost entirely to advances in deep neural networks (DNNs). Deep hierarchical memory networks are not a novel concept in cognitive science and can be traced back more than a half century to Simon's early work on discrimination nets for simulating human expertise. The major difference between DNNs and the deep memory nets meant for explaining human cognition is that the latter are (...) symbolic networks meant to model the dynamics of human memory and learning. Cognition-inspired symbolic deep networks (SDNs) address several known issues with DNNs, including (1) learning efficiency, where a much larger number of training examples are required for DNNs than would be expected for a human; (2) catastrophic interference, where what is learned by a DNN gets unlearned when a new problem is presented; and (3) explainability, where there is no way to explain what is learned by a DNN. This paper explores whether SDNs can achieve similar classification accuracy performance to DNNs across several popular ML datasets and discusses the strengths and weaknesses of each approach. Simulations reveal that (1) SDNs provide similar accuracy to DNNs in most cases, (2) SDNs are far more efficient than DNNs, (3) SDNs are as robust as DNNs to irrelevant/noisy attributes in the data, and (4) SDNs are far more robust to catastrophic interference than DNNs. We conclude that SDNs offer a promising path toward human-level accuracy and efficiency in category learning. More generally, ML frameworks could stand to benefit from cognitively inspired approaches, borrowing more features and functionality from models meant to simulate and explain human learning. (shrink)

Deep Neural Networks (DNNs) are popular for classifying large noisy analogue data. However, DNNs suffer from several known issues, including explainability, efficiency, catastrophic interference, and a need for high‐end computational resources. Our simulations reveal that psychologically‐inspired symbolic deep networks (SDNs) achieve similar accuracy and robustness to noise as DNNs on common ML problem sets, while addressing these issues.

According to the Perceptual Symbols Theory of cognition (Barsalou, 1999), modality‐specific simulations underlie the representation of concepts. A strong prediction of this view is that perceptual processing affects conceptual processing. In this study, participants performed a perceptual detection task and a conceptual property‐verification task in alternation. Responses on the property‐verification task were slower for those trials that were preceded by a perceptual trial in a different modality than for those that were preceded by a perceptual trial in the same modality. (...) This finding of a modality‐switch effect across perceptual processing and conceptual processing supports the hypothesis that perceptual and conceptual representations are partially based on the same systems. (shrink)

The evaluative conditioning effect refers to the change in the liking of a neutral stimulus due to its pairing with another stimulus. We examined whether the extinction rate of the EC effect is moderated by feature-specific attention allocation. In two experiments, CSs were abstract Gabor patches varying along two orthogonal, perceptual dimensions. During the acquisition phase, one of these dimensions was predictive of the valence of the USs. During the extinction phase, CSs were presented alone and participants were asked to (...) categorise the CSs either according to their valence, the perceptual dimension that was task-relevant during the acquisition phase, or a perceptual dimension that was task-irrelevant during the acquisition phase. As predicted, explicit valence measures revealed a linear increase in the extinction rate of the EC effect as participants were encouraged to assign attention to non-evaluative stimulus information during the extinction phase. In Experiment 1, Affect Misattribution Paradigm data mimicked this pattern of results, although the effect just missed conventional levels of significance. In Experiment 2, the AMP data revealed an increase of the EC effect if attention was focused on evaluative stimulus information. Potential mechanisms to explain these findings are discussed. (shrink)

Bayes’ inference cannot reliably account for uncertainty in mental processes. The reason is that Bayes’ inference is based on the assumption that the order in which the relevant features are evaluated is indifferent, which is not the case in most of mental processes. Instead of Bayes’ rule, a more general, probabilistic rule of inference capable of accounting for these order effects is established. This new rule of inference can be used to improve the current Bayesian models of cognition. Moreover, it (...) should play an essential role in the search for artificial emotional intelligence. (shrink)

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)

Some philosophers argue that all concepts cannot have the same representational structure, because no single kind of representation has been successful in accounting for the phenomena related to the formation and application of concepts. Here, I argue against this “appeal to cognitive science” by demonstrating that different theories of the kind concept cohere with different interpretations of the argument. To circumvent the threat of relativism, I argue that theories of concept should be understood as working hypotheses, which are provisionally accepted (...) to facilitate the investigation and explanation of the mind/brain. From this perspective, theories of concept are to be evaluated as part of a comparative analysis of the different kinds of cognitive science they inspire. (shrink)

Human participants and recurrent (“connectionist”) neural networks were both trained on a categorization system abstractly similar to natural language systems involving irregular (“strong”) classes and a default class. Both the humans and the networks exhibited staged learning and a generalization pattern reminiscent of the Elsewhere Condition (Kiparsky, 1973). Previous connectionist accounts of related phenomena have often been vague about the nature of the networks’ encoding systems. We analyzed our network using dynamical systems theory, revealing topological and geometric properties that can (...) be directly compared with the mechanisms of non-connectionist, rule-based accounts. The results reveal that the networks “contain” structures related to mechanisms posited by rule-based models, partly vindicating the insights of these models. On the other hand, they support the one mechanism (OM), as opposed to the more than one mechanism (MOM), view of symbolic abstraction by showing how the appearance of MOM behavior can arise emergently from one underlying set of principles. The key new contribution of this study is to show that dynamical systems theory can allow us to explicitly characterize the relationship between the two perspectives in implemented models. (shrink)

Neuroethological investigations of mammalian and avian auditory systems have documented species-specific specializations for processing complex acoustic signals that could, if viewed in abstract terms, have an intriguing and striking relevance for human speech sound categorization and representation. Each species forms biologically relevant categories based on combinatorial analysis of information-bearing parameters within the complex input signal. This target article uses known neural models from the mustached bat and barn owl to develop, by analogy, a conceptualization of human processing of consonant plus (...) vowel sequences that offers a partial solution to the noninvariance dilemma the locus equations orderly output constraint”) is hypothesized based on the notion of an evolutionarily conserved auditory-processing strategy. High correlation and linearity between critical parameters in the speech signal that help to cue place of articulation categories might have evolved to satisfy a preadaptation by mammalian auditory systems for representing tightly correlated, linearly related components of acoustic signals. (shrink)

Phonemes have variant pronunciations depending on context. For instance, in American English, the [t] in pat [pæt] and the [d] in pad [pæd] are both realized with a tap [ɾ] when the –ing suffix is attached, [pæɾɪŋ]. We show that despite greater distributional and acoustic support for the [t]-tap alternation, 12-month-olds successfully relate taps to stems with a perceptually-similar final [d], not the dissimilar final-[t]. Thus, distributional learning of phonological alternations is constrained by infants' preference for the alternation of perceptually-similar (...) segments. Further, the ability to relate variant surface forms emerges between 8- and 12-months. Our findings of biased learning provide further empirical support for a role for perceptual similarity in the acquisition of linguistically-relevant categories. We discuss the implications of our findings for phonological theory, language acquisition and models of the mental lexicon. (shrink)

Conceptual spaces are a frequently applied framework for representing concepts. One of its central aims is to find criteria for what makes a concept natural. A prominent demand is that natural concepts cover convex regions in conceptual spaces. The first aim of this paper is to analyse the convexity thesis and the arguments that have been advanced in its favour or against it. Based on this, I argue that most supporting arguments focus on single-domain concepts (e.g., colours, smells, shapes). Unfortunately, (...) these concepts are not the primary examples of natural concepts. Building on this observation, the second aim of the paper is to develop criteria for natural multi-domain concepts. The representation of such concepts has two main aspects: features that are associated with the concept and the probabilistic correlation pattern which the concept captures. Conceptual spaces, together with probabilistic considerations, provide a helpful framework to approach these aspects. With respect to feature representation, the existence of characteristic features (i.e., that apples have a specific taste) is essential. Moreover, natural concepts capture peaks of a probabilistic distribution over complex spaces. They carve up nature at its joints, that is, at areas with no or low probabilistic density. This last aspect is shown to be closely related to the convexity demand. (shrink)