The Role of Negative Information in Distributional Semantic Learning

Distributional models of semantics learn word meanings from contextual co‐occurrence patterns across a large sample of natural language. Early models, such as LSA and HAL (Landauer & Dumais, 1997; Lund & Burgess, 1996), counted co‐occurrence events; later models, such as BEAGLE (Jones & Mewhort, 2007), replaced counting co‐occurrences with vector accumulation. All of these models learned from positive information only: Words that occur together within a context become related to each other. A recent class of distributional models, referred to as neural embedding models, are based on a prediction process embedded in the functioning of a neural network: Such models predict words that should surround a target word in a given context (e.g., word2vec; Mikolov, Sutskever, Chen, Corrado, & Dean, 2013). An error signal derived from the prediction is used to update each word's representation via backpropagation. However, another key difference in predictive models is their use of negative information in addition to positive information to develop a semantic representation. The models use negative examples to predict words that should not surround a word in a given context. As before, an error signal derived from the prediction prompts an update of the word's representation, a procedure referred to as negative sampling. Standard uses of word2vec recommend a greater or equal ratio of negative to positive sampling. The use of negative information in developing a representation of semantic information is often thought to be intimately associated with word2vec's prediction process. We assess the role of negative information in developing a semantic representation and show that its power does not reflect the use of a prediction mechanism. Finally, we show how negative information can be efficiently integrated into classic count‐based semantic models using parameter‐free analytical transformations.     

Attention Problems and Restlessness as Transdiagnostic Markers of Severity and Treatment Response in Youth with Internalizing Problems

Research on Child and Adolescent Psychopathology - Transdiagnostic models of psychopathology suggest that disorders may share common features that could influence their severity. Attention problems...     

Direction of threat attention bias predicts treatment outcome in anxious children receiving cognitive-behavioural therapy

BackgroundA bias to selectively direct attention to threat stimuli is a cognitive characteristic of anxiety disorders. Recent studies indicate that individual differences in pre-treatment threat attention bias predict treatment outcomes from cognitive-behavioural therapy (CBT) in anxious individuals. However, there have been inconsistent findings regarding whether attention bias towards threat predicts better or poorer treatment outcome.MethodThis longitudinal study examined treatment outcomes in 35 clinically-anxious children following a 10-week, group-based CBT program, as a function of whether children showed a pre-treatment attention bias towards or away from threat stimuli. The effect of CBT on attention bias was also assessed.ResultsBoth groups showed significant improvement after receiving CBT. However, anxious children with a pre-treatment attention bias towards threat showed greater reductions not only in anxiety symptom severity, but also in the likelihood of meeting diagnostic criteria for anxiety disorders at post-treatment assessment, in comparison with anxious children who showed a pre-treatment attention bias away from threat. Children who had a pre-treatment bias away from threat showed a reduction in this bias over the course of CBT.ConclusionsFindings suggest that pre-existing differences in the direction of attention towards versus away from threat could have important implications for the treatment of anxious children.     

Facial affect recognition deficits: A potential contributor to aggression in psychotic illness

The current literature suggests a relationship between psychosis and the likelihood of engaging in aggressive behavior, though the factors involved are unclear. In this paper we use a social information processing framework to consider mediators of aggressive behavior in psychosis, especially facial affect recognition (FAR) deficits. The theoretical underpinnings of aggressive behavior are explored using the General Aggression Model (GAM) and FAR deficits are posited as a possible contributor to increased aggression in psychosis. Current research investigating the relationship between FAR, psychosis, and aggression is critically examined, and the role of potential confounding variables including positive symptoms, psychopathic personality traits, childhood trauma, and substance use briefly explored. In conclusion, we argue that socioemotional processing deficits, such as impaired FAR, are a fruitful area for research aimed at understanding, and hence reducing the risk of violence in psychosis.     

Determining the number of factors to retain in an exploratory factor analysis using comparison data of known factorial structure

Exploratory factor analysis (EFA) is used routinely in the development and validation of assessment instruments. One of the most significant challenges when one is performing EFA is determining how many factors to retain. Parallel analysis (PA) is an effective stopping rule that compares the eigenvalues of randomly generated data with those for the actual data. PA takes into account sampling error, and at present it is widely considered the best available method. We introduce a variant of PA that goes even further by reproducing the observed correlation matrix rather than generating random data. Comparison data (CD) with known factorial structure are first generated using 1 factor, and then the number of factors is increased until the reproduction of the observed eigenvalues fails to improve significantly. We evaluated the performance of PA, CD with known factorial structure, and 7 other techniques in a simulation study spanning a wide range of challenging data conditions. In terms of accuracy and robustness across data conditions, the CD technique outperformed all other methods, including a nontrivial superiority to PA. We provide program code to implement the CD technique, which requires no more specialized knowledge or skills than performing PA.     

Spatial working memory for locations specified by vision and audition: Testing the amodality hypothesis

Spatial working memory can maintain representations from vision, hearing, and touch, representations referred to here as spatial images. The present experiment addressed whether spatial images from vision and hearing that are simultaneously present within working memory retain modality-specific tags or are amodal. Observers were presented with short sequences of targets varying in angular direction, with the targets in a given sequence being all auditory, all visual, or a sequential mixture of the two. On two thirds of the trials, one of the locations was repeated, and observers had to respond as quickly as possible when detecting this repetition. Ancillary detection and localization tasks confirmed that the visual and auditory targets were perceptually comparable. Response latencies in the working memory task showed small but reliable costs in performance on trials involving a sequential mixture of auditory and visual targets, as compared with trials of pure vision or pure audition. These deficits were statistically reliable only for trials on which the modalities of the matching location switched from the penultimate to the final target in the sequence, indicating a switching cost. The switching cost for the pair in immediate succession means that the spatial images representing the target locations retain features of the visual or auditory representations from which they were derived. However, there was no reliable evidence of a performance cost for mixed modalities in the matching pair when the second of the two did not immediately follow the first, suggesting that more enduring spatial images in working memory may be amodal.