Chapter 12. Learning Reasoning Strategies in End-to-End Differentiable Proving

Attempts to render deep learning models interpretable, data-efficient, and robust have seen some success through hybridisation with rule-based systems, for example, in Neural Theorem Provers (NTPs). These neuro-symbolic models can induce interpretable rules and learn representations from data via back-propagation, while providing logical explanations for their predictions. However, they are restricted by their computational complexity, as they need to consider all possible proof paths for explaining a goal, thus rendering them unfit for large-scale applications. We present Conditional Theorem Provers (CTPs), an extension to NTPs that learns an optimal rule selection strategy via gradient-based optimisation. We show that CTPs are scalable and yield state-of-the-art results on the CLUTRR dataset, which tests systematic generalisation of neural models by learning to reason over smaller graphs and evaluating on larger ones. Finally, CTPs show better link prediction results on standard benchmarks in comparison with other neural-symbolic models, while being explainable. All source code and datasets are available online. (At https://github.com/uclnlp/ctp)

How statically-typed functional programmers write code

How working statically-typed functional programmers write code is largely understudied. And yet, a better understanding of developer practices could pave the way for the design of more useful and usable tooling, more ergonomic languages, and more effective on-ramps into programming communities. The goal of this work is to address this knowledge gap: to better understand the high-level authoring patterns that statically-typed functional programmers employ. We conducted a grounded theory analysis of 30 programming sessions of practicing statically-typed functional programmers, 15 of which also included a semi-structured interview. The theory we developed gives insight into how the specific affordances of statically-typed functional programming affect domain modeling, type construction, focusing techniques, exploratory and reasoning strategies, and expressions of intent. We conducted a set of quantitative lab experiments to validate our findings, including that statically-typed functional programmers often iterate between editing types and expressions, that they often run their compiler on code even when they know it will not successfully compile, and that they make textual program edits that reliably signal future edits that they intend to make. Lastly, we outline the implications of our findings for language and tool design. The success of this approach in revealing program authorship patterns suggests that the same methodology could be used to study other understudied programmer populations.

Diversity, Dissent and Representation: Lessons From The First Minority Judge in the Israeli Supreme Court

This paper seeks to enrich existing empirical research on substantive representation in the judicial system by exploring a case study of the Honorable Justice (retired) Salim Joubran, the first ethnic-minority judge appointed to the Supreme Court of Israel. By employing a dual methodology of qualitative discourse analysis and dissenting quantitative studies, the study investigates when, why and how he dissented in controversial cases, which are defined as cases that resulted in non-unanimous votes. The study shows that a quantitative study on dissenting opinions of a minority judge alone did not provide comprehensive conclusions. The complementary qualitative discourse analysis shows that in cases that challenged state actions that impacted his social group, Joubran employed distinct strategies and reasoning that are akin to feminist judgments approach. Hence, the study adds to existing research on judicial diversity indicating that women and ethnic minorities judges not only share common challenges but might also operate similar reasoning strategies. In light of these insights, the study calls for employing the combined qualitative and quantitative methodology on examining judgments focusing on dissenting opinions of women and ethnic minority judges as it offers a complex understanding of substantive representation and provides answers regarding the socio-legal effects of judgments group affiliation on.

Does causal reasoning lead to moral reasoning? Consumers' responses to scandalized athletes and endorsements

PurposeMarketers and brand managers are subject to reputational crises when their endorsers are involved in scandals. To effectively manage such crises, it is imperative to understand (1) the underlying mechanisms through which consumers process negative information surrounding morally tainted endorsers, and (2) how these mechanisms affect consumer behavior in the context of athlete scandals.Design/methodology/approachDrawing on attribution theory and the moral reasoning strategy framework, we investigate the impact of attribution on moral reasoning strategies, and the impact of such strategies on consumers' responses to scandalized athletes and endorsements.FindingsOverall, our results demonstrate that the same scandal can be evaluated differently, depending on its information, including the consensus, distinctiveness, and consistency of the scandal. The results of Study 1 show that in the context of an on-field scandal, individuals engage in a sequential cognitive process in which they go through attribution, the choice of a moral reasoning strategy, and ultimately a response. The results of Study 2 reveal that in the context of an off-field scandal, attribution directly influences consumers' responses.Originality/valueWe extend the existing literature on the moral reasoning of athlete scandals by suggesting that attribution is a determinant of moral reasoning choice in the context of on-field scandals. We also extend the sports marketing and consumer behavior literature by suggesting that consumers' diverse reactions to athlete scandals depend on their attribution patterns and moral reasoning choices.

Lesson study as a research approach: a case study

PurposeThe purpose of this paper is to explore the merits of lesson study (LS) as a research approach for research in (science) education. A lesson was developed to introduce students to model-based reasoning: a higher order thinking skill that is seen as one of the major reasoning strategies in science.Design/methodology/approachParticipants of the LS team were three secondary school teachers and two educational researchers. Additionally, one participant fulfilled both roles. Both qualitative and quantitative data were used to investigate the effect of the developed lesson on students and to formulate focal points for using the LS as a research approach.FindingsThe developed lesson successfully familiarized students with model-based reasoning. Three main focal points were formulated for using LS as a research approach: (1) make sure that the teachers support the research question that the researchers bring into the LS cycle, (2) take into account that the lesson is supposed to answer a research question that might cause extra stress for the teachers in an LS team and (3) state the role of both researchers and teachers in an LS team clearly at the beginning of the LS cycle.Originality/valueThis study aims to investigate whether LS can be used as a research approach by the educational research community.

Exploring ideation strategies as an opportunity to support and evaluate making

Purpose This paper aims to compare two types of prompts, encouraging participants to think about real-world examples or engineering principles to show how these two approaches can result in vastly different design practices. Design/methodology/approach Two studies (N = 20, N = 40) examine the impact of two different prompts. Non-expert students, from high school and university, completed a hands-on, engineering design task in pairs. Half were prompted to ideate using real-world examples, while the other half were prompted to ideate using engineering principles. The findings are based on human coding and artifact analyses. Findings In both studies, and across multiple measures, students in the principle-based condition performed better than students in the example-based condition. Research limitations/implications A seemingly small difference in how students are prompted or encouraged to approach a problem can have a significant impact on their experience. The findings also suggest that leveraging engineering principles, even when those principles are only loosely formed, can be effective even for non-experts. Finally, the findings motivate identifying student reasoning strategies over time as a potential means for assessment in Makerspaces. Practical implications Encouraging makers to think about different ways for approaching problems can be an important way to help them succeed. It may also be a useful way to chronicle their learning pathway. Originality/value To the author's knowledge, explicitly looking at ideation strategies has not been widely discussed within the Maker community as a way to support learners, or as a way to evaluate learning.

A Descriptive Analysis for a Collaborative Work Process: A Complex Real Medical Case in the Radiological Field

Collaborative acts occur daily in every human activity. In the case of medicine, and particularly in the diagnostic decision process, these acts are very frequent and occur naturally. It is very important to properly understand how these collaborative acts are developed in order to provide tools that facilitate and support them. In this article, we describe this collaborative work process in the framework of a complex real medical case in the radiological field. Usually, complex cases require several specialists. In this work, we have analyzed the intervention of several specialists and the exchange and interaction of different reasoning strategies among specialists, while considering their temporal dimension. Two types of collaboration are presented in the case analysis (1) exchange between specialists from different specialties and (2) exchange between specialists from the same specialty. The method of analysis follows five steps: (1) Case synopsis, (2) Temporal representation of the case, (3) Analysis of the general decision in the case, (4) Analysis of the reasoning in the medical case using the different strategies, and (5) Analysis of radiological collaboration. We have presented different reasoning strategies, data, hypotheses and complementary tests from different sources in the diagnostic resolution process and we have shown that collaboration is present during the entire process. The temporality and the intervention of different specialists is shown using a graphical representation. We have focused special attention on radiological collaboration, and have shown how a radiological diagnosis is achieved. We have discussed different elements present in the collaboration process. Our study has produced meta-knowledge derived from these exchanges that is of value in the context of artificial intelligence progress, in particular for the comprehension of collaborative medical work.