archives@tulane.edu / Heuristics allow us to navigate a complex and uncertain world, react to new situations, and make decisions without needing all the information. While this is often helpful, using heuristics can sometimes lead to biased actions and potentially costly errors. Many algorithms are developed with the goal of improving human performance: either by offering recommendations or by seeking to predict and thereby mitigate human error. Such tools benefit when they can account for human bias and behavior, either through analysis of behavioral data in the design process or by incorporating computational cognitive models that simulate and predict behavior. This dissertation explores several novel computational approaches for simulating bias and heuristics in human behavior at different cognitive levels of abstraction. First, we develop a model of attentional bias in spatial auditory attention for the cognitive architecture, ACT-R. We then move to a more complex task, modeling confirmation bias that occurs when making decisions with noisy feedback. Finally, we consider behavior in a multi-agent setting, collecting and analyzing experimental data about heuristics in voting. In all three instances, we implement different computational models of human behavior to evaluate them on behavioral data. This work provides tools that support a better understanding of human behavior and enables the integration of bias and heuristics in human-aware computing, laying the groundwork for designing new computational approaches inspired by human heuristic behavior. / 1 / Jaelle Scheuerman

A three parameter model of Item Response Theory reported by Lord (1968, 1980) was used to determine whether cultural item bias existed in the Reading Comprehension subtest of the California Achievement Tests. Item responses from 1500 second graders from Southern Arizona schools with 500 responses from each of three groups (Anglo, Black, and Hispanic) were analyzed via MULTILOG program (Thissen, 1984) using the likelihood ratio chi-square (IRT-LR) method. Results indicated that there were no significant differences in item difficulty across items and groups. However, the typical group mean differences often reported wherein Anglos usually score between.5 to 1 standard deviation above the group means of other minorities on standardized tests did appear. The results of this study were also consistent with and supportive of the notion that most standardized tests today are not biased against specific minority groups. Selected items from the Reading Comprehension subtest on the California Achievement Tests are being used to date as anchor items for newer achievement tests. As such, this study lends support and credence to those items. Other implications for future research are discussed.

Lors de leur mise en forme par thermo- estampage, les pré-imprégnés à matrice polymère thermoplastique et renfort fibreux tissé se déforment selon différents modes encore mal maîtrisés, dont les cisaillements plan et transverse, ce dernier traduisant la friction entre les plis tissés. Une fois formés, des contraintes résiduelles de fabrication sont également générées au cours du refroidissement des pré- imprégnés et altèrent leur stabilité dimensionnelle, tout comme les variations d’humidité relative. Dans ce travail, nous avons étudié ces deux aspects. Dans un premier temps, des essais de traction de biais ont été réalisés sur des pré-imprégnés de polyamide renforcés par des tissus de fibres de verre à hautes températures. L’analyse multi-échelle de la structure des échantillons déformés par tomographie à rayons X met en exergue des mécanismes de déformation des mèches de fibres dont certains ne sont pas pris en compte dans les modèles rhéologiques actuels. En outre, nous avons mis au point un dispositif pour l’étude du cisaillement transverse et la friction inter-plis. Les résultats obtenus montrent et quantifient les influences de la contrainte normale, de la vitesse de déformation et de la température sur le coefficient de frottement inter-plis. Un modèle de frottement basé sur le nombre d’Hersey est ensuite identifié. Dans un deuxième temps, l’exploitation de la mesure de la courbure globale des échantillons après refroidissement a permis d’estimer, via la théorie modifiée des stratifiées, les contraintes résiduelles et leur évolution avec les conditions hygrothermiques ambiantes. / During their thermo-stamping, prepregs made of thermoplastic polymer matrices and woven fabrics are subjected to large deformation including in-plane and transverse shear, the last deformation mechanism being related to inter-ply friction. Once shaped, prepregs can still deform due to residual stresses induced by their cooling and/or hygrometric variations. In this work, we focused on these two aspects. Firstly, prepregs made of polyamide reinforced by layers of woven fabrics (glass fibre bundles) were subjected to bias extension at high temperatures. The fibrous structure deformed specimens was analysed with X-ray microtomography, allowing fiber bundles deformation mechanism to be emphasized.Some of them are not taken into account in current rheological model developed for prepregs. In addition, using a dedicated apparatus, we performed transverse shear tests to analyse inter-ply friction within prepregs. Experimental results underline the influence of the normal stress, the shear strain rate and the temperature on the interplay friction coefficient. A friction model based on the Hersey number is then fitted. Secondly, we measure the curvatures of deformed prepregs was to estimate, via the modified laminated theory, residual stresses and their variation with temperature and relative humidity.

The purpose of this study was to examine instructor-student interactions in the College Algebra classroom for gender bias. Three measuring instruments were constructed to answer five research questions. These instruments included a Researcher Observation Code Sheet, an Instructor Questionnaire, and a Student Questionnaire. One of the research questions required triangulation of all three perspectives for the interactions.Participants included four mathematics instructors, 54 female students and 45 male students. Eighty-one students filled out the Student Questionnaire. The researcher coded 764 interactions.Findings showed most interactions involved instructors posing open questions to the class. Students mostly called out answers. Lower-level questions were asked the most by instructors.Male and female students responded almost equally to male and female instructor's questions. Male students received more positive responses from female instructors and females received more negative responses from male instructors. More students with male instructors were silent in class. Female students interacted more in male taught classes and female students communicated more via questions/comments in female-taught classes.Students perceived instructors called on them by name, pointing, or eye contact. Although male students thought they had more interactions than females, male students in male-taught classes thought females had more interactions with instructors. Female instructors tended to ignore students more than male instructors. More male students thought they were ignored than females. Female students thought instructors responded to them more positively than male students did. Students also thought that male and female students interacted with questions/comments equally regardless of instructor sex. Furthermore, students reported that instructor sex did not matter.Instructors reported their interactions with students as equitable, they knew all or most of the students' names, and treated both sexes equally when responding to them. Male instructors thought female students participated more, whereas female instructors thought males did. Most instructors classified their classroom climate as warm, friendly, or laid back.Triangulation of the three different perceptions showed that in most instances, instructor-student interactions were not perceived the same. However, all three were in agreement on classroom climate.

No description available.

Thesis (Ph. D.)--University of Wisconsin--Madison, 1984. / Typescript. Vita. Description based on print version record. Includes bibliographical references (leaves 99-104).

Thesis (M.S.)--Boston University

Indiana University-Purdue University Indianapolis (IUPUI) / Previous research has provided promising evidence for the effectiveness of video interventions to raise awareness of the gender bias that exists in science, technology, engineering, and math (STEM) fields. However, creating these beneficial video interventions can be costly and its possible that the success of these interventions may be hindered if a certain amount of resources are not invested into their creation (e.g., into high quality actors). Therefore, the current study expands on this research by investigating the use of personal testimonials of women’s experiences with gender bias in STEM as gender bias interventions. More specifically, I examined whether certain characteristics (e.g., genuineness) of the testimonial’s communicator would influence the relevant gender bias outcome (e.g., awareness of bias and sexism). I predicted that watching a woman genuinely talking about her experiences with bias, relative to a scripted re-telling of this story, would lead to better perceptions of the woman (e.g., more positive perceptions and perceptions of genuineness), greater feelings of empathy and connection with the woman, as well as less greater awareness of gender bias and less sexism from participants. To test this possibility, participants in the study were randomly assigned to interact with one of four story formats, a genuine telling of the woman’s experiences with bias, a scripted re-telling of these experiences, a written version of the story, or a unrelated control video. After interacting with the story participants then completed measures related to perceptions of the woman and gender bias. The findings of the study were limited, possibly due to a lack of power, but were encouraging as they were in the expected direction. The implications of the findings as well as future research ideas to expand and improve on these findings are discussed.

Note:

Social phobia is maintained in part by judgmental biases concerning the probability and cost of negative social events. One hypothesized mechanism of action of cognitive behavioral therapy (CBT) for anxiety disorders is its reduction in the exaggerated probabilities and costs associated with feared outcomes, termed the “cognitive mediation hypothesis” (Foa & Kozak, 1986). A number of studies have examined the cognitive mediation hypothesis; some investigations find cost bias to be more important to treatment outcome, whereas others find probability bias to be more important. However, methodological limitations of several of these studies leave open the possibility that changes in judgmental biases are simply correlates or consequences of social anxiety reduction. Attentional processes, which mark the first discrimination of incoming information, may serve as precursors to cognitive processes like probability and cost estimates. Though intuitively linked, whether social phobics’ pattern of attending to external threat cues is correlated with their appraisals of the cost and probability of negative events has yet to be examined empirically. The current project examines cost and probability biases and their relation to attention bias and treatment outcome in a randomized controlled trial of CBT for social phobia. It was found that, contrary to hypotheses, greater attentional bias for threat in either direction (vigilance or avoidance) did not predict higher cost and probability estimates. However, a significant relation was observed between attentional vigilance and outcome probability estimates, such that greater vigilance for threat predicted greater estimates of the likelihood that negative social events will occur. As hypothesized, early changes in cost and probability biases predicted later changes in social anxiety symptoms (and not vice versa). Changes in probability estimates were a stronger predictor of treatment outcome than changes in cost estimates. Broadly, findings provide support for the cognitive mediation hypothesis of social phobia and point to both outcome cost and outcome probability as potential treatment mechanisms. Findings are discussed in the context of extant theories of social phobia, and directions for future research are proposed.