Examining the effect of initialization strategies on the performance of Gaussian mixture modeling

Mixture modeling is a popular technique for identifying unobserved subpopulations (e.g., components) within a data set, with Gaussian (normal) mixture modeling being the form most widely used. Generally, the parameters of these Gaussian mixtures cannot be estimated in closed form, so estimates are typically obtained via an iterative process. The most common estimation procedure is maximum likelihood via the expectation-maximization (EM) algorithm. Like many approaches for identifying subpopulations, finite mixture modeling can suffer from locally optimal solutions, and the final parameter estimates are dependent on the initial starting values of the EM algorithm. Initial values have been shown to significantly impact the quality of the solution, and researchers have proposed several approaches for selecting the set of starting values. Five techniques for obtaining starting values that are implemented in popular software packages are compared. Their performances are assessed in terms of the following four measures: (1)?the ability to find the best observed solution, (2)?settling on a solution that classifies observations correctly, (3)?the number of local solutions found by each technique, and (4)?the speed at which the start values are obtained. On the basis of these results, a set of recommendations is provided to the user.     

Early processing of emotional faces in children with autism: An event-related potential study

Social deficits are one of the most striking manifestations of autism spectrum disorders (ASDs). Among these social deficits, the recognition and understanding of emotional facial expressions has been widely reported to be affected in ASDs. We investigated emotional face processing in children with and without autism using event-related potentials (ERPs). High-functioning children with autism (n = 15, mean age = 10.5 ± 3.3 years) completed an implicit emotional task while visual ERPs were recorded. Two groups of typically developing children (chronological age-matched and verbal equivalent age-matched [both ns = 15, mean age = 7.7 ± 3.8 years]) also participated in this study. The early ERP responses to faces (P1 and N170) were delayed, and the P1 was smaller in children with autism than in typically developing children of the same chronological age, revealing that the first stages of emotional face processing are affected in autism. However, when matched by verbal equivalent age, only P1 amplitude remained affected in autism. Our results suggest that the emotional and facial processing difficulties in autism could start from atypicalities in visual perceptual processes involving rapid feedback to primary visual areas and subsequent holistic processing.     

Resistance to Socialization into Organizational Corruption: A Model of Deontic Justice

Purpose  

The purpose of this article is to develop a deontic model of corruption in organizations. This model intends to explain newcomers’ resistance to socialization into corrupt practices.     

Sexual Passion in Committed Relationships: Measurement and Conceptual Issues

The purpose of this study is to discuss the need for a new triadic model of sexual passion in relationships and to present the preliminary psychometric properties of a scale designed to measure these three approaches to passion (harmonious, obsessive, and inhibited) for use in clinical and scholarly work. Existing theory and measures of general passion are based on a dualistic model of passion that includes the harmonious and obsessive approaches to passion. We added the inhibited approach to passion from the sexuality research and develop measures for assessing sexual passion in relationships. We utilized an Amazon Mechanical Turk sample of 1,421 individuals in committed relationships to test this measure. Reliability analyses and confirmatory factory analyses evinced that these three approaches to sexual passion were unique constructs and distinct from sexual satisfaction. Sexual passion showed predictive validity above and beyond relationship length, sexual desire toward a partner, and a broader variable of sexual drive. Harmonious sexual passion robustly predicted higher sexual satisfaction and relationship satisfaction, and inhibited sexual passion moderately predicted lower sexual satisfaction and relationship satisfaction. Obsessive sexual passion had minimal associations with either outcome. These new constructs, especially harmonious and inhibited sexual passion, may help scholars and practitioners improve their understanding of sexual satisfaction and overall relationship satisfaction.     

VIDEO MODELING TO TRAIN STAFF TO IMPLEMENT DISCRETE‐TRIAL INSTRUCTION

Three new direct‐service staff participated in a program that used a video model to train target skills needed to conduct a discrete‐trial session. Percentage accuracy in completing a discrete‐trial teaching session was evaluated using a multiple baseline design across participants. During baseline, performances ranged from a mean of 12% to 63% accuracy. During video modeling, there was an immediate increase in accuracy to a mean of 98%, 85%, and 94% for each participant. Performance during maintenance and generalization probes remained at high levels. Results suggest that video modeling can be an effective technique to train staff to conduct discrete‐trial sessions.     

Hemispheric specialization for language according to grapho-phonemic transformation and gender. A divided visual field experiment

This behavioral study aimed at assessing the effect of two variables on the degree of hemispheric specialization for language. One of them was the grapho-phonemic translation (transformation) (letter-sound mapping) and the other was the participants’ gender. The experiment was conducted with healthy volunteers. A divided visual field procedure has been used to perform a phoneme detection task implying either regular (transparent) grapho-phonemic translation (letter-sound mapping consistency) or irregular (non-transparent) grapho-phonemic translation (letter-sound mapping inconsistency). Our results reveal a significant effect of grapho-phonemic translation on the degree of hemispheric dominance for language. The phoneme detection on items with transparent translation (TT) was performed faster than phoneme detection on items with non-transparent translation (NTT). This effect seems to be due to faster identification of TT than NTT when the items were presented in the left visual field (LVF)-right hemisphere (RH). There was no difference between TT and NTT for stimuli presented in the right visual field (RVF)-left hemisphere (LH). This result suggests that grapho-phonemic translation or the degree of transparency can affect the degree of hemispheric specialization, by modulating the right hemisphere activity. With respect to gender, male participants were significantly more lateralized than female participants but no interaction was observed between gender and degree of transparency.     

The effect of context on the structure of categories

Three experiments examined the effect of context on the representativeness ordering of exemplars of a category. Experiments 1 and 2 employed an online reading time paradigm to examine the effect of context on the time it takes to establish an anaphoric reference between an exemplar and a category term. Experiment 1 demonstrated that context can change the relation between a category term and an exemplar at the time of comprehension. Experiment 2 showed that category terms presented in context generate graded goodness-of-example distributions of exemplars that are different from the distributions generated in the absence of explicit context. These distributions cannot be derived by assuming that the exemplar most strongly suggested by the context serves as the category representation. Experiment 3 employed a membership verification paradigm. Response time was found to be a function of degree of relatedness to the contextdependent category representation. Typicality, as determined in the absence of explicit context, had no effect on decision time. Several models, including some extensions of current semantic memory theories, are developed to account for the results of these experiments.     

Can lemurs learn to deceive? A study in the black lemur (Eulemur macaco)

Deceptive behavior in primates has been the focus of a number of studies. Nevertheless, such abilities have never been demonstrated in prosimians. The authors' goal was to analyze possible deception in lemurs according to a paradigm used with simians. Three black lemurs were trained to communicate about the location of a hidden reward with a cooperative trainer. Afterward, when a 2nd trainer and lemurs competed to gain access to the reward, each subject differentially adapted its learned behavior to the context. Their performances with the cooperative trainer remained stable while they showed various behavioral adjustments when faced to the competitive trainer: 1 subject refused to participate, another preferentially withheld information, and the 3rd sometimes pointed deceptively to obtain the reward.