This is a summary of Reconstructing Constructionism
Authors: Richard Noss, University College London; James Clayson, American University of Paris
Click here to check the original version.
This paper argues that constructionism, the learning theory launched by Seymour Papert around 1980, needs to reconnect with its original epistemological foundations if it’s going to have any lasting effect on education. The authors want to shift constructionism from a loose framework for action into a more precise set of tools for describing what learners actually do inside constructionist environments, so those descriptions can be tested and used to improve design.
Constructivism
Papert coined constructionism deliberately as distinct from constructivism. Constructivism describes the psychological basis of how people learn in general. Constructionism is narrower and more pedagogical: it’s a theory about how people learn best through building something external, reflecting on it, debugging it, and sharing it. The Logo programming language was the original vehicle for this idea, and its influence continues today through tools like Scratch, which has produced over 10 million student projects.
Despite that legacy, the authors say constructionism hasn’t produced fundamental change in how teaching and learning actually happen. They point to three reasons. First, not every educational system actually wants to foster independent thought, which is a political reality more than a theoretical one. Second, the research literature around constructionism hasn’t developed enough nuance to guide practice at scale. Third, and most importantly, constructivism and constructionism get conflated constantly, to the point where almost any teaching approach gets labeled constructivist regardless of whether it reflects Papert’s actual ideas.
Educational theories
The authors then dig into what kind of theory constructionism really is, building on a 2004 paper by Andy diSessa and Paul Cobb. DiSessa and Cobb sorted educational theories into categories, from grand theories like Piaget’s constructivism, which was never meant to guide design and mostly doesn’t, down to domain-specific instructional theories that make testable predictions about learning and how to trigger it.
Constructionism falls into a middle category diSessa and Cobb call frameworks for action. These frameworks help structure the design of learning environments, but they tend to blur the line between their scientific claims and the practical actions they recommend, which makes them hard to test rigorously.
Noss and Clayson use this framing to explain why constructionism has stalled. When you build something with digital tools, you build with things, not just abstract ideas, and the relationships between those things get expressed through the system of things itself, not through formal notation.
situated abstraction
The authors illustrate this with something they call situated abstraction: a pattern where students working on math or science problems clearly grasp the underlying relationships between variables, but express those relationships using the specific tools and objects they used to build the model rather than standard algebraic notation. This term, developed in their earlier work with Celia Hoyles, tries to name a real pattern of student behavior so it can be studied and used to guide design rather than dismissed as students failing to reach proper abstraction.
Six characteristics
The bulk of the paper lays out six characteristics the authors believe define a constructionist agenda:
- Modelling comes first. Learners build external structures through building, reflecting, and debugging, which in turn builds internal knowledge. This lets students use crucial mathematical or scientific concepts directly rather than being taught about them abstractly, echoing Papert’s Power Principle of learning powerful ideas through use.
- Accessibility comes second. Since students already live inside technological systems, from mobile phones to financial systems, they should get a real look at how models work and shape the world around them, not just use tools blindly.
- Layering is third. The authors build software with increasing layers of mathematical and scientific complexity, letting students choose how deep to dig into the reasoning behind the software’s feedback, or whether to edit and extend the models themselves.
- Tapping into youth culture is fourth. Math and science carry real social weight but often get dismissed by students as boring or irrelevant. The authors argue for designing environments where students actually need the content knowledge to reach goals they care about.
- Making knowledge visible through language is fifth. Students need a language, broadly defined, through which they can express and represent what they’re learning.
- Sustained collaborative engagement is sixth. Long-term collaborative projects, where students hold individual and collective responsibility and stay engaged with real content over time, produce better learning than short, isolated exercises.
Design
The paper closes on design. Every one of these six characteristics demands serious design investment, and the difficulty of that design work tends to be badly underestimated. A small number of designers end up shaping how millions of students and teachers experience technology in the classroom, often without making clear how the tools actually work or what assumptions are baked into them.
The authors argue this makes design itself, and its underlying epistemological assumptions, a central research problem, one that has to account for how learning communities shape and get shaped by the tools they use, especially as collaboration technology adds even more complexity to that relationship.