A Dyadic IRT Model

Psychometrika - We propose a dyadic Item Response Theory (dIRT) model for measuring interactions of pairs of individuals when the responses to items represent the actions (or behaviors,...     

EARSHOT: A Minimal Neural Network Model of Incremental Human Speech Recognition

Despite the lack of invariance problem (the many-to-many mapping between acoustics and percepts), human listeners experience phonetic constancy and typically perceive what a speaker intends. Most models of human speech recognition (HSR) have side-stepped this problem, working with abstract, idealized inputs and deferring the challenge of working with real speech. In contrast, carefully engineered deep learning networks allow robust, real-world automatic speech recognition (ASR). However, the complexities of deep learning architectures and training regimens make it difficult to use them to provide direct insights into mechanisms that may support HSR. In this brief article, we report preliminary results from a two-layer network that borrows one element from ASR, long short-term memory nodes, which provide dynamic memory for a range of temporal spans. This allows the model to learn to map real speech from multiple talkers to semantic targets with high accuracy, with human-like timecourse of lexical access and phonological competition. Internal representations emerge that resemble phonetically organized responses in human superior temporal gyrus, suggesting that the model develops a distributed phonological code despite no explicit training on phonetic or phonemic targets. The ability to work with real speech is a major advance for cognitive models of HSR.     

Why People Join Terrorist Groups in Kuwait: A Qualitative Examination

Qualitative interviews were conducted with nine members of the Islamic State of Iraq and Syria (ISIS) and Al-Qaeda currently incarcerated in Kuwait's Central Prison. The semistructured interviews attempted to understand psychosocial factors in Kuwait that contributed to their decision to join extremist organizations. Interviews were analyzed using Interpretive Phenomenological Analysis (IPA) and the emergent themes identified the following core themes explaining their involvement: (1) religious identity development; (2) personal connections; (3) propaganda; (4) defense of Islam; and (5) social marginalization. Participants described a process whereby their religion became a central part of their personal identity. As their religious identity developed, they became involved in Islamic organizations where they met people involved with ISIS or Al-Qaeda. These social connections exposed them to jihadi propaganda which, in addition to increased military conflict in the Middle East, crystallized their beliefs that Islam is under attack, and they were religiously obligated to defend it. The results also identified societal factors that increased the probability of engaging in terrorism including relatively low levels of education, coming from low socioeconomic groups in Kuwait, and feeling socially marginalized by broader Kuwaiti society.     

Changing counterproductive beliefs about attention,memory, and multitasking: Impacts of a brief,fully online module

Ubiquitous mobile technology is part of contemporary life, bringing with it the potential for distraction and reduction in performance associated with multitasking. The predisposition toward dysfunctional multitasking may be shaped in part by beliefs that individuals hold about memory and attention. The issue is particularly pressing for college students, given established links between distraction, multitasking, and learning. This project assessed the impact of an online learning module on beliefs about attention, memory, and learning in college students. It also contrasted these beliefs in a college and non-college community sample. Significant reductions in counterproductive beliefs were associated with completing the module; counterproductive beliefs were also no more prevalent in the college vs. the non-college sample. Our findings suggest that brief online modules are a practical way to address counterproductive beliefs related to multitasking with technology, and add to the literature on metacognition, attention, and multitasking in college and non-college populations.     

Modeling the personality of dyads and groups

The paper presents a methodological approach for assessing the personality of a dyad or a group, a concept that is not equivalent to the sum, or mean, of the individual scores. We illustrate how the logic of the multitrait multimethod approach, which is a familiar technique for establishing construct validity, can be extended to assess the construct of a relationship "personality." The model, which we call the latent group model, provides a decomposition and comparison of individual-level and group-level variance in a given trait, and the individual-level and group-level covariance or correlation between two traits. The model is also extended to the assessment of stability of the individual and group level traits. Throughout the paper, we draw connections between related methods and show how the latent group model can be estimated through hierarchical linear modeling.     

Perceptions of another person's heavy drinking as a function of one's relationship to the drinker a pilot study

Of 301 participants in a telephone survey, 201 could identify a problem drinker within their social networks. In 67 cases the drinker was a partner, in 45 another family member, in 61 a friend, and in 28 a work colleague. Another person's heavy drinking was reported as most disrupting to one's life when that person was a partner or other family member or if one was living in the same residence as the drinker. The exception to this was that reported disruption to respondents' work life was greatest when the drinker was a work colleague. While there appear to be problems for friends and work colleagues, partners and other family members appear to report the most disruption so providing services to family members seems appropriate.     

Background beliefs in Bayesian inference

We report five experiments in which the role of background beliefs in social judgments of posterior probability was investigated. From a Bayesian perspective, people should combine prior probabilities (or base rates) and diagnostic evidence with equal weighting, although previous research shows that base rates are often underweighted. These experiments were designed so that either piece of information was supplied either by personal beliefs or by presented statistics, and regression analyses were performed on individual participants to assess the relative influence of information. We found that both prior probabilities and diagnostic information significantly influenced judgments, whether supplied by beliefs or by statistical information, but that belief-based information tended to dominate the judgments made.     

Recency effects for meaning and form in word selection

Three experiments examined the contribution of phonological availability in selecting words as predicted by interactive activation models of word production. Homophonous words such as week and weak permitted a word's phonological form to be activated on priming trials without selection of its meaning or lemma. Recent production of a homophone failed to significantly increase production of its twin as a sentence completion. However, speakers were significantly more likely to complete a sentence with a recently read or generated unambiguous word. This increase in response probability was unaffected by word frequency. The results constrain the degree to which experience and phonological availability may affect word selection in spoken language production.