Chemistry, Society and Uncertainty

In recent years, besides the increased interest in philosophy of chemistry, we have witnessed a "material turn" in philosophy and the history of sciences with an interest in putting instruments, objects, materials and practices at the core of historical reports. Since its alchemic past, chemistry has worked with and on materials, so that its history is also a "material history". Thus, in the wake of this "material turn", it is up to philosophy and the history of chemistry to perceive the chemical substances, the chemists that create them and the industries that produce them as part of culture, society and politics. This overlap between chemical reasoning and materiality as well as the artificial character of its products makes chemistry an eminently technoscientific science. In this context, we will analyze the most general aspect that led us to identify it as "technoscientific", the hybrid that exists between chemistry and society. With that, we intend to argue in favor of considering the modern societal necessities (material, environmental, and human) with chemistry, in an effort to build a more harmonious relationship, being that it will be long and, maybe, indissoluble. Following that, our aim is to develop a concept that cannot be separated from the capillarity of chemistry in societies and the environment, the imprevisibility and essential uncertainty of the behavior of chemical entities in multiple contexts. Finally, we will highlight some reflections concerning chemical ethics associated with the production and creation of new substances that may become a part of the lifeworld.

Integrating Deep Neural Networks and Symbolic Inference for Organic Reactivity Prediction

<div>Accurate <i>in silico</i> models for the prediction of novel chemical reaction outcomes can be used to guide the rapid discovery of new reactivity and enable novel synthesis strategies for newly discovered lead compounds. Recent advances in machine learning, driven by deep learning models and data availability, have shown utility throughout synthetic organic chemistry as a data-driven method for reaction prediction. Here we present a machine-intelligence approach to predict the products of an organic reaction by integrating deep neural networks with a probabilistic and symbolic inference that flexibly enforces chemical constraints and accounts for prior chemical knowledge. We first train a graph convolutional neural network to estimate the likelihood of changes in covalent bonds, hydrogen counts, and formal charges. These estimated likelihoods govern a probability distribution over potential products. Integer Linear Programming is then used to infer the most probable products from the probability distribution subject to heuristic rules such as the octet rule and chemical constraints that reflect a user's prior knowledge. Our approach outperforms previous graph-based neural networks by predicting products with more than 90% accuracy, demonstrates intuitive chemical reasoning through a learned attention mechanism, and provides generalizability across various reaction types. Furthermore, we demonstrate the potential for even higher model accuracy when complemented by expert chemists contributing to the system, boosting both machine and expert performance. The results show the advantages of empowering deep learning models with chemical intuition and knowledge to expedite the drug discovery process.</div>

Integrating Deep Neural Networks and Symbolic Inference for Organic Reactivity Prediction

<div>Accurate <i>in silico</i> models for the prediction of novel chemical reaction outcomes can be used to guide the rapid discovery of new reactivity and enable novel synthesis strategies for newly discovered lead compounds. Recent advances in machine learning, driven by deep learning models and data availability, have shown utility throughout synthetic organic chemistry as a data-driven method for reaction prediction. Here we present a machine-intelligence approach to predict the products of an organic reaction by integrating deep neural networks with a probabilistic and symbolic inference that flexibly enforces chemical constraints and accounts for prior chemical knowledge. We first train a graph convolutional neural network to estimate the likelihood of changes in covalent bonds, hydrogen counts, and formal charges. These estimated likelihoods govern a probability distribution over potential products. Integer Linear Programming is then used to infer the most probable products from the probability distribution subject to heuristic rules such as the octet rule and chemical constraints that reflect a user's prior knowledge. Our approach outperforms previous graph-based neural networks by predicting products with more than 90% accuracy, demonstrates intuitive chemical reasoning through a learned attention mechanism, and provides generalizability across various reaction types. Furthermore, we demonstrate the potential for even higher model accuracy when complemented by expert chemists contributing to the system, boosting both machine and expert performance. The results show the advantages of empowering deep learning models with chemical intuition and knowledge to expedite the drug discovery process.</div>

Targeting complex plutonium oxides by combining crystal chemical reasoning with density-functional theory calculations: the quaternary plutonium oxide Cs2PuSi6O15

The stability of the novel Pu(iv) silicate, Cs2PuSi6O15, was predicted from a combination of crystal chemical reasoning and DFT calculations and confirmed by its synthesis via flux crystal growth.

HEURISTIC REASONING OF PRE-SERVICE SCIENCE TEACHERS IN CHEMISTRY TOPICS

The aim of the present research is to explain how the heuristics utilized by the students in a multiple choice examination on the general chemistry subject of “chemical bonding theories and molecular structures” caused biases on intuitive judgment and decision making processes, using the three characteristics of associative memory (attribute substitution, fluency process and associative coherence). A mixed-methods approach, both qualitative and quantitative research methods, were used in this research. Therefore, both questionnaire and individual interview were utilized to collect data. The results of the current research demonstrated that the participants used 4 different decision making strategies. Detailed evaluation of these strategies demonstrated that most of the participants did not prefer the processes related to the use of chemical knowledge and thus, were not able to assess the target attribute. Furthermore, it was identified that most of the students’ decision making processes were dependent on one or more of these three associative memory processes. It was also determined by this research that the most dominant of these three associative memory processes is the fluency effect, since participants often prefer to use superficial features. The dependence of participants on associative memory processes caused various biases, so participants often responded incorrectly to questions. Keywords: chemistry education, chemical reasoning, intuitive judgments, science education.

An examination of preservice elementary teachers’ representations about chemistry in an intertextuality- and modeling-based course

The chemistry education aspect of elementary teacher education faces a unique set of challenges. On one hand, preservice and in-service elementary teachers tend to not like chemistry and have negative feelings toward chemistry. On the other hand, learning chemistry requires reasoning about natural phenomena from the submicroscopic perspective that deals with the properties and behaviors of unobservable particles. The present study addresses these challenges in chemistry education for preservice elementary teachers (PSETs) by designing a chemistry curriculum that improves the relevance of chemistry learning to studentsviaintertextuality and modeling practices. An analysis of chemistry representations that PSETs generated before and after taking the designed chemistry course demonstrates that they initially perceived chemistry as vivid chemical changes occurring in lab spaces or a discipline related to atoms while failing to provide connections between the chemical reactions and atoms. After taking the course, many students came to see doing chemistry as epistemic practices that construct submicroscopic explanations for observable phenomena and its relevance to everyday lives such as food, car emissions, and their local surroundings. They also came to recognize various epistemic roles that people play in doing chemistry. We provide important implications for engaging PSETs in chemical reasoning and designing chemistry curricula that are more approachable and build on learners’ knowledge resources.

The nature of students' chemical reasoning employed in scientific argumentation in physical chemistry

Recent science education reform efforts have emphasized scientific practices in addition to scientific knowledge. Less work has been done at the tertiary level to consider students' engagement in scientific practices. In this work, we consider physical chemistry students' engagement in argumentation and construction of causal explanations. Students in two POGIL physical chemistry classrooms were videotaped as they engaged in discourse while solving thermodynamics problems. Videos were transcribed and transcripts were analyzed using the Toulmin Argument Pattern (TAP). Arguments were then characterized using the modes of reasoning in a learning progression on chemical thinking (CTLP) (Sevian and Talanquer, 2014). Results showed that students used primarily relational reasoning, in which no causal explanation is generated, rather a single relationship between variables was used to justify a claim. We discuss all types of reasoning present in students' arguments.

Ene-nucleic acids: a different paradigm to DNA chemistry

The chemical reasoning would suggest that Ene-nucleic acid precursors with constrained flexibility and selectivity could be the missing link between the prochiral-acyclic and chiral-cyclic structures.