Chemical representations play a vital part in the teaching and learning of chemistry. The aim of this research was to investigate students’ understanding of chemical representations and to ascertain the influence of chemical representations on students’ developing mental models of chemical phenomena. Three primary threads flowing through the thesis are models, representations and learning. Each thread was found to play a vital part in students’ learning of chemical content, in their learning of the scientific process and in their learning about the process of learning itself. This research with students from Year 8 to first year university level comprised four studies that provide comparisons between ages, abilities, learning settings and teaching and learning approaches. Students’ modelling ability was observed to develop and improve through instruction and practice and usually coincided with an improvement in their understanding of chemical concepts. While students were observed to actively use models to make predictions and test ideas, some were not aware of the predictive nature of models when asked about it. From the research, five characteristics of scientific models have been identified: scientific models as multiple representations, scientific models as exact replicas, scientific models as explanatory tools, how scientific models are used, and the dynamic nature of scientific models. A theoretical framework relating the four types of models - teaching, scientific, mental and expressed - and a typology of models that highlights the significant attributes of models, support the research results. The data showed that students’ ability to describe the role of the scientific model in the process of science improved with their increasing age and maturity. / The relationship between the three levels of chemical representation of matter - the macroscopic level, the sub-microscopic level and the symbolic level - revealed some complexities concerning the representational and theoretical qualities and the reality of each level. The research data showed that generally most students had a good understanding of the macroscopic and symbolic levels of chemical representation of matter. However, students’ understanding of the sub-microscopic level varied, with some students being able to spontaneously envisage the sub- microscopic view while for others their understanding of the sub-microscopic level of chemical representation was lacking. To make sense of the sub-microscopic level, students’ appreciation of the accuracy and detail of any scientific model, or representation upon which their mental model is built, depended on them being able to distinguish reality from representation, distinguish reality from theory, know what a representation is, understand the role of a representation in the process of science, and understand the role of a theory in the process of science. In considering learning, the importance of an individual’s modelling ability was examined alongside the role of chemical representations and models in providing clear and concise explanations. Examining the links forged between the three levels of chemical representation of matter provided an insight into how students were learning and understanding chemical concepts. Throughout this research, aspects of students’ metacognition and intention were identified as being closely related to their development of mental models. / The research identified numerous factors that influenced learning, including internal factors such as students’ prior chemical and mathematical knowledge, their modelling ability and use of chemical representations, motivation, metacognitive ability and time management as well as external factors such as organisation, assessment, teaching resources, getting feedback and good explanations. The choice of learning strategies by students and instructors appeared to be influenced by those factors that influenced learning. Feedback to students, in the form of discussion with classmates, online quizzes and help from instructors on their understanding was observed to be significant in promoting the learning process. Many first year university non-major chemistry students had difficulties understanding chemical concepts due to a limited background knowledge in chemistry and mathematics. Accordingly, greater emphasis at the macroscopic level of representation of matter with contextual references is recommended. The research results confirmed the theoretical construct for learning chemistry - the rising iceberg - that suggests all chemistry teaching begins at the macroscopic level, with the sub-microscopic and symbolic levels being introduced as needed. More of the iceberg becomes visible as the students’ mental model and depth of understanding increases. In a variety of situations, the changing status of a concept was observed as students’ understanding in terms of the intelligibility, plausibility and fruitfulness of a concept developed. / The research data supported four aspects of learning - epistemological, ontological, social affective and metacognitive - as being significant in the students’ learning and the development of their mental models. Many university students, who are mature and are experienced learners, exhibited strong rnetacognitive awareness and an intentional approach to learning. It is proposed that the intentional and metacognitive learning approaches and strategies could be used to encourage students to be more responsible for their own learning.

Les relations entre la structure des composés chimiques et leurs propriétés sont complexes et à haute dimension. Dans le processus de développement de médicaments, plusieurs proprié- tés d’un composé doivent souvent être optimisées simultanément, ce qui complique encore la tâche. Ce travail explore deux représentations des composés chimiques pour les tâches de prédiction des propriétés. L’objectif de ces représentations proposées est d’améliorer l’explicabilité afin de faciliter le processus d’optimisation des propriétés des composés. Pre- mièrement, nous décomposons l’algorithme ECFP (Extended connectivity Fingerprint) et le rendons plus simple pour la compréhension humaine. Nous remplaçons une fonction de hachage sujet aux collisions par une relation univoque de sous structure à bit. Nous consta- tons que ce changement ne se traduit pas par une meilleure performance prédictive d’un perceptron multicouche par rapport à l’ECFP. Toutefois, si la capacité du prédicteur est ra- menée à celle d’un prédicteur linéaire, ses performances sont meilleures que celles de l’ECFP. Deuxièmement, nous appliquons l’apprentissage automatique à l’analyse des paires molécu- laires appariées (MMPA), un paradigme de conception du développement de médicaments. La MMPA compare des paires de composés très similaires, dont la structure diffère par une modification sur un site. Nous formons des modèles de prédiction sur des paires de com- posés afin de prédire les différences d’activité. Nous utilisons des contraintes de similarité par paires comme MMPA, mais nous utilisons également des paires échantillonnées de façon aléatoire pour entraîner les modèles. Nous constatons que les modèles sont plus performants sur des paires choisies au hasard que sur des paires avec des contraintes de similarité strictes. Cependant, les meilleurs modèles par paires ne sont pas capables de battre les performances de prédiction du modèle simple de base. Ces deux études, RCFP et comparaisons par paires, visent à aborder la prédiction des propriétés d’une manière plus compréhensible. En utili- sant l’intuition et l’expérience des chimistes médicinaux dans le cadre de la modélisation prédictive, nous espérons encourager l’explicabilité en tant que composante nécessaire des modèles cheminformatiques prédictifs. / The relationships between the structure of chemical compounds and their properties are complex and high dimensional. In the drug development process, multiple properties of a compound often need to be optimized simultaneously, further complicating the task. This work explores two representations of chemical compounds for property prediction tasks. The goal of these suggested representations is improved explainability to better understand the compound property optimization process. First, we decompose the Extended Connectivity Fingerprint (ECFP) algorithm and make it more straightforward for human understanding. We replace a collision-prone hash function with a one-to-one substructure-to-bit relationship. We find that this change which does not translate to higher predictive performance of a multi- layer perceptron compared to ECFP. However, if the capacity of the predictor is lowered to that of a linear predictor, it does perform better than ECFP. Second, we apply machine learning to Matched Molecular Pair Analysis (MMPA), a drug development design paradigm. MMPA compares pairs of highly similar compounds, differing in structure by modification at one site. We train prediction models on pairs of compounds to predict differences in activity. We use pairwise similarity constraints like MMPA, but also use randomly sampled pairs to train the models. We find that models perform better on randomly chosen pairs than on pairs with strict similarity constraints. However, the best pairwise models are not able to beat the prediction performance of the simpler baseline single model. Both of these investigations, RCFP and pairwise comparisons, aim to approach property prediction in a more explainable way. By using intuition and experience of medicinal chemists within predictive modelling, we hope to encourage explainability as a necessary component of predictive cheminformatic models.