Abstract: One of the principal bottlenecks in applying learning techniques to classification problems is the large amount of labeled training data required. Especially for images and video, providing training data is very expensive in terms of human time and effort. In this paper we propose an active learning approach to tackle the problem. Instead of passively accepting random training examples, the active learning algorithm iteratively selects unlabeled examples for the user to label, so that human effort is focused on labeling the most “useful” examples. Our method relies on the idea of uncertainty sampling, in which the algorithm selects unlabeled examples that it finds hardest to classify. Specifically, we propose an uncertainty measure that generalizes margin-based uncertainty to the multi-class case and is easy to compute, so that active learning can handle a large number of classes and large data sizes efficiently. We demonstrate results for letter and digit recognition on datasets from the UCI repository, object recognition results on the Caltech-101 dataset, and scene categorization results on a dataset of 13 natural scene categories. The proposed method gives large reductions in the number of training examples required over random selection to achieve similar classification accuracy, with little computational overhead.

Abstract: 1. Perception and Illusion 2. Attention and Search 3. Memory Processes 4. Spatial Cognition 5. Timing and Counting 6. Conceptualization and Categorization 7. Pattern Learning 8. Tool Fabrication and Use 9. Problem Solving and Behavioral Flexibility 10. Social Cognition Processes

Abstract: One of the most provocative and exciting issues in cognitive science is how neural specificity for semantic categories of common objects arises in the functional architecture of the brain. More than two decades of research on the neuropsychological phenomenon of category-specific semantic deficits has generated detailed claims about the organization and representation of conceptual knowledge. More recently, researchers have sought to test hypotheses developed on the basis of neuropsychological evidence with functional imaging. From those two fields, the empirical generalization emerges that object domain and sensory modality jointly constrain the organization of knowledge in the brain. At the same time, research within the embodied cognition framework has highlighted the need to articulate how information is communicated between the sensory and motor systems, and processes that represent and generalize abstract information. Those developments point toward a new approach for understanding category specificity in terms of the coordinated influences of diverse regions and cognitive systems.

Abstract: What information is available from a brief glance at a novel scene? Although previous efforts to answer this question have focused on scene categorization or object detection, real-world scenes contain a wealth of information whose perceptual availability has yet to be explored. We compared image exposure thresholds in several tasks involving basic-level categorization or global-property classification. All thresholds were remarkably short: Observers achieved 75%-correct performance with presentations ranging from 19 to 67 ms, reaching maximum performance at about 100 ms. Global-property categorization was performed with significantly less presentation time than basic-level categorization, which suggests that there exists a time during early visual processing when a scene may be classified as, for example, a large space or navigable, but not yet as a mountain or lake. Comparing the relative availability of visual information reveals bottlenecks in the accumulation of meaning. Understanding these bottlenec...

Abstract: Human subjects are extremely efficient at categorizing natural scenes, despite the fact that different classes of natural scenes often share similar image statistics. Thus far, however, it is unknown where and how complex natural scene categories are encoded and discriminated in the brain. We used functional magnetic resonance imaging (fMRI) and distributed pattern analysis to ask what regions of the brain can differentiate natural scene categories (such as forests vs mountains vs beaches). Using completely different exemplars of six natural scene categories for training and testing ensured that the classification algorithm was learning patterns associated with the category in general and not specific exemplars. We found that area V1, the parahippocampal place area (PPA), retrosplenial cortex (RSC), and lateral occipital complex (LOC) all contain information that distinguishes among natural scene categories. More importantly, correlations with human behavioral experiments suggest that the information present in the PPA, RSC, and LOC is likely to contribute to natural scene categorization by humans. Specifically, error patterns of predictions based on fMRI signals in these areas were significantly correlated with the behavioral errors of the subjects. Furthermore, both behavioral categorization performance and predictions from PPA exhibited a significant decrease in accuracy when scenes were presented up-down inverted. Together these results suggest that a network of regions, including the PPA, RSC, and LOC, contribute to the human ability to categorize natural scenes.

Abstract: Human observers are able to rapidly and accurately categorize natural scenes, but the representation mediating this feat is still unknown. Here we propose a framework of rapid scene categorization that does not segment a scene into objects and instead uses a vocabulary of global, ecological properties that describe spatial and functional aspects of scene space (such as navigability or mean depth). In Experiment 1, we obtained ground truth rankings on global properties for use in Experiments 2–4. To what extent do human observers use global property information when rapidly categorizing natural scenes? In Experiment 2, we found that global property resemblance was a strong predictor of both false alarm rates and reaction times in a rapid scene categorization experiment. To what extent is global property information alone a sufficient predictor of rapid natural scene categorization? In Experiment 3, we found that the performance of a classifier representing only these properties is indistinguishable from human performance in a rapid scene categorization task in terms of both accuracy and false alarms. To what extent is this high predictability unique to a global property representation? In Experiment 4, we compared two models that represent scene object information to human categorization performance and found that these models had lower fidelity at representing the patterns of performance than the global property model. These results provide support for the hypothesis that rapid categorization of natural scenes may not be mediated primarily though objects and parts, but also through global properties of structure and affordance.

Abstract: A variety of studies have demonstrated that organizing stimuli into categories can affect the way the stimuli are perceived. We explore the influence of categories on perception through one such phenomenon, the perceptual magnet effect, in which discriminability between vowels is reduced near prototypical vowel sounds. We present a Bayesian model to explain why this reduced discriminability might occur: It arises as a consequence of optimally solving the statistical problem of perception in noise. In the optimal solution to this problem, listeners’ perception is biased toward phonetic category means because they use knowledge of these categories to guide their inferences about speakers’ target productions. Simulations show that model predictions closely correspond to previously published human data, and novel experimental results provide evidence for the predicted link between perceptual warping and noise. The model unifies several previous accounts of the perceptual magnet effect and provides a framework for exploring categorical effects in other domains.

Abstract: What is the effect of option categorization on choosers' satisfaction? A combination of field and laboratory experiments reveals that the mere presence of categories, irrespective of their content, positively influences the satisfaction of choosers who are unfamiliar with the choice domain. This "mere categorization effect" is driven by a greater number of categories signaling greater variety among the available options, which allows for a sense of self-determination from choosing. This effect, however, is attenuated for choosers who are familiar with the choice domain, who do not rely on the presence of categories to perceive the variety available.

Abstract: In this paper, I review evidence from two theories in psychology relevant to diagnosis and diagnostic errors. “Dual Process” theories of thinking, frequently mentioned with respect to diagnostic error, propose that categorization decisions can be made with either a fast, unconscious, contextual process called System 1 or a slow, analytical, conscious, and conceptual process, called System 2. Exemplar theories of categorization propose that many category decisions in everyday life are made by unconscious matching to a particular example in memory, and these remain available and retrievable individually. I then review studies of clinical reasoning based on these theories, and show that the two processes are equally effective; System 1, despite its reliance in idiosyncratic, individual experience, is no more prone to cognitive bias or diagnostic error than System 2. Further, I review evidence that instructions directed at encouraging the clinician to explicitly use both strategies can lead to consistent reduction in error rates.

Abstract: Compared with other objects, faces are processed more holistically and with a larger reliance on configural information. Such hallmarks of face processing can also be found for nonface objects as people develop expertise with them. Is this specifically a result of expertise individuating objects, or would any type of prolonged intensive experience with objects be sufficient? Two groups of participants were trained with artificial objects (Ziggerins). One group learned to rapidly individuate Ziggerins (i.e., subordinate-level training). The other group learned rapid, sequential categorizations at the basic level. Individuation experts showed a selective improvement at the subordinate level and an increase in holistic processing. Categorization experts improved only at the basic level, showing no changes in holistic processing. Attentive exposure to objects in a difficult training regimen is not sufficient to produce facelike expertise. Rather, qualitatively different types of expertise with objects of a given geometry can arise depending on the type of training.

Abstract: Constituent structure is considered to be the very foundation of linguistic competence and often considered to be innate, yet we show here that it is derivable from the domain-general processes of chunking and categorization. Using modern and diachronic corpus data, we show that the facts support a view of constituent structure as gradient (as would follow from its source in chunking and categorization) and subject to gradual changes over time. Usage factors (i.e., repetition) and semantic factors both influence chunking and categorization and, therefore, influence constituent structure. We take as our example the complex prepositions of English, for instance, on top of, in back of, and in spite of, whose internal constituent structure has been much debated. From observing strong (but not absolute) usage trends in the corpus data, we find that these complex preposition sequences display varying degrees of emerging constituency. We conclude that constituent reanalysis, like language change generally, proceeds gradually.

Abstract: A series of 8 experiments investigated the association between pictorial and verbal representations and the psychological distance of the referent objects from the observer. The results showed that people better process pictures that represent proximal objects and words that represent distal objects than pictures that represent distal objects and words that represent proximal objects. These results were obtained with various psychological distance dimensions (spatial, temporal, and social), different tasks (classification and categorization), and different measures (speed of processing and selective attention). The authors argue that differences in the processing of pictures and words emanate from the physical similarity of pictures, but not words, to the referents. Consequently, perceptual analysis is commonly applied to pictures but not to words. Pictures thus impart a sense of closeness to the referent objects and are preferably used to represent such objects, whereas words do not convey proximity and are preferably used to represent distal objects in space, time, and social perspective.

Abstract: What information is available from a brief glance at a novel scene? Although previous efforts to answer this question have focused on scene categorization or object detection, real-world scenes contain a wealth of information whose perceptual availability has yet to be explored. We compared image exposure thresholds in several tasks involving basic-level categorization or global-property classification. All thresholds were remarkably short: Observers achieved 75%-correct performance with presentations ranging from 19 to 67 ms, reaching maximum performance at about 100 ms. Global-property categorization was performed with significantly less presentation time than basic-level categorization, which suggests that there exists a time during early visual processing when a scene may be classified as, for example, a large space or navigable, but not yet as a mountain or lake. Comparing the relative availability of visual information reveals bottlenecks in the accumulation of meaning. Understanding these bottlenec...

Abstract: Three experiments (N=370) investigated the effects of social categorization on the experience of empathy. In Experiment 1, university students reported their empathy for, and intentions to help, a student who described a distressful experience. As predicted, participants reported stronger empathy and helping intentions when the student belonged to an ingroup compared to an outgroup university. Experiments 2 and 3 demonstrated that stronger empathy for outgroup members was experienced following the activation of an ingroup norm that prescribed the experience of this emotion. Activating this norm also led to the expression of more positive attitudes towards the outgroup (Experiment 3), and empathy fully mediated this effect. These findings indicate that like other emotions, empathy is influenced by social categorization processes.

Abstract: In this article, we examine the extent to which membership categorization analysis (MCA) can inform an understanding of reasoning within the public domain where morality, policy and cultural politics are visible (Smith and Tatalovich, 2003). Through the examination of three examples, we demonstrate how specific types of category device(s) are a ubiquitous feature of accountable practice in the public domain where morality matters and public policy intersect. Furthermore, we argue that MCA provides a method for analysing the mundane mechanics associated with everyday cultural politics and democratic accountability assembled and presented within news media and broadcast settings.

Abstract: Humans have an extremely flexible ability to categorize regularities in their environment, in part because of attentional systems that allow them to focus on important perceptual information. In formal theories of categorization, attention is typically modeled with weights that selectively bias the processing of stimulus features. These theories make differing predictions about the degree of flexibility with which attention can be deployed in response to stimulus properties. Results from 2 eye-tracking studies show that humans can rapidly learn to differently allocate attention to members of different categories. These results provide the first unequivocal demonstration of stimulus-responsive attention in a categorization task. Furthermore, the authors found clear temporal patterns in the shifting of attention within trials that follow from the informativeness of particular stimulus features. These data provide new insights into the attention processes involved in categorization.

Abstract: A substantial literature has examined the nature of social categorization, a fundamental process having important implications for a wide variety of social phenomena. The great majority of this research has focused on the role of particular, clearly identified social categories (e.g. race, nationality, etc.) while ignoring or holding constant other identity dimensions. This approach has afforded considerable leverage for understanding how salient social identities influence perception, judgment, and behavior. However, it leaves unaddressed many questions about how particular social identities become salient and how (and whether) identities might be inferred when category membership is ambiguous or unknown. Everyday social perception often occurs under conditions of volatility (dynamic contexts), uncertainty (missing information), complexity (multiple bases for categorization), and ambiguity (unclear meaning of available cues). As a consequence, research must address how these factors might qualify basic processes of social categorization. Available evidence is reviewed, and directions for future research are discussed. The central importance of social categorization in shaping perception, judgment, and behavior has long been recognized. The writings of seminal theorists such as Allport (1954), Sherif (1948), and Tajfel (1974) emphasized the significance of the psychological borderlines that define membership in particular social groups, and countless studies have examined the consequences of salient category memberships on social functioning (for reviews, see Brewer & Brown, 1985; Macrae & Bodenhausen, 2000). In this article, we focus on research examining the role played by social categories in basic processes of social perception and cognition. The dominant research strategy in this domain has been straightforward and powerful: manipulate the identity of a target in a manner that makes membership in a given category salient and clear, while holding constant all else that is known about the target. Goldberg’s (1968) classic study of sex discrimination provides a prototypic example. In this research, participants evaluated the quality of written essays; essay content was held constant, but the name of the ostensible author was manipulated in a manner clearly

Abstract: Text categorization is the task of assigning predefined categories to natural language text. With the widely used 'bag of words' representation, previous researches usually assign a word with values such that whether this word appears in the document concerned or how frequently this word appears. Although these values are useful for text categorization, they have not fully expressed the abundant information contained in the document. This paper explores the effect of other types of values, which express the distribution of a word in the document. These novel values assigned to a word are called distributional features, which include the compactness of the appearances of the word and the position of the first appearance of the word. The proposed distributional features are exploited by a tf idf style equation and different features are combined using ensemble learning techniques. Experiments show that the distributional features are useful for text categorization. In contrast to using the traditional term frequency values solely, including the distributional features requires only a little additional cost, while the categorization performance can be significantly improved. Further analysis shows that the distributional features are especially useful when documents are long and the writing style is casual.

Abstract: Learning to recognize the contrasts of a language-specific phonemic repertoire can be viewed as forming categories in a multidimensional psychophysical space. Research on the learning of distributionally defined visual categories has shown that categories defined over 1 dimension are easy to learn and that learning multidimensional categories is more difficult but tractable under specific task conditions. In 2 experiments, adult participants learned either a unidimensional or a multidimensional category distinction with or without supervision (feedback) during learning. The unidimensional distinctions were readily learned and supervision proved beneficial, especially in maintaining category learning beyond the learning phase. Learning the multidimensional category distinction proved to be much more difficult and supervision was not nearly as beneficial as with unidimensionally defined categories. Maintaining a learned multidimensional category distinction was only possible when the distributional information that identified the categories remained present throughout the testing phase. We conclude that listeners are sensitive to both trial-by-trial feedback and the distributional information in the stimuli. Even given limited exposure, listeners learned to use 2 relevant dimensions, albeit with considerable difficulty.

Abstract: In addition to its communicative functions, language has been argued to have a variety of extracommunicative functions, as assessed by its causal involvement in putatively nonlinguistic tasks. In the present work, I argue that language may be critically involved in the ability of human adults to categorize objects on a specific dimension (e.g., color) while abstracting over other dimensions (e.g., size). This ability is frequently impaired in aphasic patients. The present work demonstrates that normal participants placed under conditions of verbal interference show a pattern of deficits strikingly similar to that of aphasic patients: impaired taxonomic categorization along perceptual dimensions, and preserved thematic categorization. A control experiment using a visuospatial-interference task failed to find this selective pattern of deficits. The present work has implications for understanding the online role of language in normal cognition and supports the claim that language is causally involved in nonverbal cognition.

Abstract: To elucidate the neural mechanisms underlying the development of perceptual expertise, we recorded ERPs while participants performed a categorization task. We found that as participants learned to discriminate computer generated "blob" stimuli, feedback modulated the amplitude of the error-related negativity (ERN)-an ERP component thought to reflect error evaluation within medial-frontal cortex. As participants improved at the categorization task, we also observed an increase in amplitude of an ERP component associated with object recognition (the N250). The increase in N250 amplitude preceded an increase in amplitude of an ERN component associated with internal error evaluation (the response ERN). Importantly, these electroencephalographic changes were not observed for participants who failed to improve on the categorization task. Our results suggest that the acquisition of perceptual expertise relies on interactions between the posterior perceptual system and the reinforcement learning system involving medial-frontal cortex.

Abstract: Two Studies examined the role of phonological cues in the lexical categorization of new words when children could also rely on learning by exclusion and whether the role of phonology depends oil extensive experience with a language. Phonological Cues were assessed via phonological typicality - an aggregate measure of the relationship between the phonology of a word and the phonology of words in the same lexical class. Experiment I showed that when monolingual English-speaking seven-year-olds could rely oil learning by exclusion, phonological typicality only affected their initial Inferences about the words. Consistent with recent computational analyses, phonological Cues had stronger impact on the processing of verb-like than noun-like items. Experiment 2 revealed an impact of French on the performance of seven-year-olds in French immersion when tested in a French language environment. Thus, phonological knowledge may affect lexical categorization even in the absence of extensive experience.