Single-Stage Prediction Models Do Not Explain the Magnitude of Syntactic Disambiguation Difficulty

The disambiguation of a syntactically ambiguous sentence in favor of a less preferred parse can lead to slower reading at the disambiguation point. This phenomenon, referred to as a garden-path effect, has motivated models in which readers initially maintain only a subset of the possible parses of the sentence, and subsequently require time-consuming reanalysis to reconstruct a discarded parse. A more recent proposal argues that the garden-path effect can be reduced to surprisal arising in a fully parallel parser: words consistent with the initially dispreferred but ultimately correct parse are simply less predictable than those consistent with the incorrect parse. Since predictability has pervasive effects in reading far beyond garden-path sentences, this account, which dispenses with reanalysis mechanisms, is more parsimonious. Crucially, it predicts a linear effect of surprisal: the garden-path effect is expected to be proportional to the difference in word surprisal between the ultimately correct and ultimately incorrect interpretations. To test this prediction, we used recurrent neural network language models to estimate word-by-word surprisal for three temporarily ambiguous constructions. We then estimated the slowdown attributed to each bit of surprisal from human self-paced reading times, and used that quantity to predict syntactic disambiguation difficulty. Surprisal successfully predicted the existence of garden-path effects, but drastically underpredicted their magnitude, and failed to predict their relative severity across constructions. We conclude that a full explanation of syntactic disambiguation difficulty may require recovery mechanisms beyond predictability.     

Jail‐based competency treatment comes of age: Multi‐site outcomes and challenges to the implementation of an evidence‐based forensic continuum

The jail‐based competency treatment (JBCT) model has become an established forensic practice across the country. From the perspective of implementation science and the three core elements of the Promoting Action on Research Implementation in Health Service (PARiHS) framework, the JBCT model is a remarkable example of how context (an unrelenting and overwhelmingly strong demand for forensic beds) has driven multiple state governments to facilitate implementation of a methodology in the absence of empirical evidence supporting its efficacy. This 7‐year study of outcomes from four JBCT program sites provides this much‐needed evidence by showing that JBCT restored 56% of 1553 male and 336 female patients over an average of 48.7 days. At the same time, the study highlights how variations in JBCT models, methods, and preadmission stabilization time present challenges to planned and effective implementation of evidence‐based practice at the statewide system level. By identifying differential responsiveness to JBCT treatment by diagnosis and other factors, the study suggests preliminary implementation ideas for what types of patients are well served by the JBCT model as part of a continuum of restoration options that includes inpatient, outpatient and diversion. Significant findings showed that JBCT patients were restored at a higher rate and in a shorter time if they were female, < 20 years old (highest restoration rate; those < 60 years old also significantly better rates), free of co‐occurring intellectual and cognitive deficits, and malingering. Of the major diagnoses, schizoaffective disorder required a significantly longer length of JBCT treatment for restoration, and lower restoration rates than schizophrenia and bipolar disorder, although this was moderated by a significant interaction with abuse of amphetamines.     

If you want to be one of us,then become like us: The evaluation of naturalization applicants by host nationals

Immigrant naturalization is a rite of passage, making assimilationist attitudes particularly pronounced among host nationals. Three experimental studies investigate whether heritage culture maintenance violates expectations that citizenship should be deserved by proving strong attachment to the host nation (i.e., neoliberal communitarianism). Study 1 (N = 293) demonstrates that naturalization applicants’ high degree of heritage culture maintenance impairs application evaluations. Perceived attachment to the host nation and citizenship deservingness mediated this effect. Study 2 (N = 220) replicates results across two national contexts and reveals that heritage culture maintenance impairs evaluations only among naturalization applicants from devalued countries. Study 3 (N = 117) manipulates attachment to the host nation and shows that perceived citizenship deservingness mediates the negative effects of naturalization applicants' low attachment to the host nation on application evaluations. Overall, assimilationist attitudes among host nationals are best explained by the combination of neoliberal and communitarian criteria of evaluation.     

Ability evaluation by proxy: role of maximal performance and related attributes in social comparison

Four experiments tested the effects of information about a proxy's related attributes and maximal effort in social comparisons of ability. In Study 1, a proxy's prior performance at maximal effort exerted systematic effects on participants' personal performance predictions for a novel task. When information about the proxy's effort was ambiguous (Study 2). information about related attribute similarity was used in the formulation of performance predictions. In Studies 3 and 4, 2-way interaction terms revealed that participants used related attribute information in generating performance predictions when the proxy's maximum effort was unknown. However, related attribute information was disregarded when proxy's maximum effort was unambiguous. Results were consistent with the proxy comparison model of ability self-evaluation (L. Wheeler, R. Martin, & J. Suls, 1997).     

Where am I looking? The accuracy of video-mediated gaze awareness

Participants worked in pairs, with one person gazing at a flat horizontal stimulus between them. The other participant estimated where the gazer was looking. Experiment 1 used linear scales as gaze targets. The mean root mean square error of estimation equates to 3.8 degrees of head-and-eye pan and 2.6 degrees of tilt. This small error of estimation was essentially the same in a video-mediated condition and in one in which a procedure that did not allow the estimator to see the head-and-eye movement to the target position was used. Experiment 2 obtained comparable gaze estimation performance in face-to-face and video-mediated conditions, using a combined pan-and-tilt grid. It is concluded that people are very good at estimating what someone else is looking at and that such estimations should be practical during video-mediated conversation.     

Decoupling the Relation Between Risk Factors for Conduct Problems and the Receipt of Intervention Services: Participation Across Multiple Components of a Prevention Program

This study examined whether the link between risk factors for conduct problems and low rates of participation in mental health treatment could be decoupled through the provision of integrated prevention services in multiple easily-accessible contexts. It included 445 families of first-grade children (55% minority), living in four diverse communities, and selected for early signs of conduct problems. Results indicated that, under the right circumstances, these children and families could be enticed to participate at high rates in school-based services, therapeutic groups, and home visits. Because different sets of risk factors were related to different profiles of participation across the components of the prevention program, findings highlight the need to offer services in multiple contexts to reach all children and families who might benefit from them. Ellen Pinderhughes's and Karen Bierman's colleagues in the Conduct Problems Prevention Research Group are, in alphabetical order, John D. Coie, Duke University; Kenneth A. Dodge, Duke University; E. Michael Foster, University of North Carolina; Mark T. Greenberg, Pennsylvania State University; John E. Lochman, University of Alabama; and Robert J. McMahon, University of Washington.