David H. Jonassen

Examines constraint‑based discussion boards to scaffold preservice teachers’ online argumentation while solving ill‑structured, diagnosis–solution problems. Scaffolding led to more hypothesis generation/testing and richer problem‑space construction; certain epistemological beliefs were associated with problem‑solving performance.

With experienced practitioners, investigates how long‑term use of computer tools changes knowledge structures and problem‑solving processes in ill‑structured tasks. Compares performance with and without tools and discusses qualitative (of‑tools) and quantitative (with‑tools) effects on expert cognition.

Reports a seminar case study in which students used expert‑system shells to model metacognitive processes as runnable simulations of cognition. Building simulations fostered introspection, ownership, and meaning‑making; the article explains instructional benefits and logistics of using expert systems as a learning activity.

Compares skills required for solving well‑ versus ill‑structured problems in an open‑ended astronomy environment. Regression analyses showed well‑structured performance was predicted by domain knowledge and justification skills, whereas ill‑structured performance was additionally related to science attitudes and regulation of cognition.

In an experiment with undergraduates solving ill‑structured problems, access to a library of related stories influenced performance. The study details treatments (experimental, comparable, control) and reports effects on multiple‑choice and short‑answer measures, advancing design implications for case‑based scaffolds.

Proposes a meta‑theory of problem solving and a typology of problem types ranging from well‑ to ill‑structured. Describes how differences in structuredness, domain specificity, and complexity, as well as individual differences, necessitate varied instructional supports for problem‑solving competence across contexts.

Argues that activity theory offers an analytic lens for designing constructivist learning environments (CLEs). By modeling subjects, tools, rules, community, and division of labor within an activity system, designers can align learning tasks with authentic performance contexts and support knowledge construction emerging from purposeful activity.

Introduces the Mindtools perspective: general‑purpose software and applications serve as knowledge representation formalisms that prompt analysis, reflection, and collaboration. The paper categorizes Mindtools (semantic organization, dynamic modeling, interpretation, knowledge construction, and conversation tools) and explains how each scaffolds distinct forms of reasoning.

Differentiates well‑structured from ill‑structured problems and proposes models both for how learners solve these problems and for designing instruction to support them. The framework grounds well‑structured problem instruction in information‑processing theory, and ill‑structured problem instruction in constructivist and situated cognition perspectives.

Argues that computers should function as cognitive tools that learners use to represent, interpret, and construct knowledge. Rather than delivering information, technologies are repurposed to support processes such as modeling, mapping, and information organization, thereby engaging learners in deeper critical thinking and meaning making across domains.