MicroWorlds are tools that allow an unexperienced computer user to build programs within any domain. These tools consider that building a model of a certain domain, helps the learner understand the principles governing the domain. According to Papert, microworlds offer a safe incubator environment for children to learn mathematics and physics. In these worlds, children learn by taking what is new and relating it to something that they already know and by making the new their own. The children communicate with a computational "Turtle" by means of a very simple programming language: LOGO. By sending appropriate movement commands to the turtle, they have access to the principles of Euclidean geometry. By adding velocity and velocity commands, Newtonian mechanics come alive before their eyes. The authors of this method of learning claim it is effective, active, and interactive. While learning to understand the correct theories, the children formulate and discard many false theories, thus mirroring the way that children learn when they are young.
In order to build microworlds, we need a programming language the 'ordinary' person can use to manipulate programs. The design criteria of this language will be different than those of programming languages today. Understandability, usefulness for small tasks, and interactivity will be more important than formal specifications as efficiency, verifiability, and uniformity, for instance. Boxer is a language which addresses these issues, providing its own environment/metaphor for programming and simulations. It relies on the concepts of spatial metaphor and naive realism to address these criteria. In Boxer, the spatial metaphor is that everything is in a box, all variables are boxes with values in them. By entering a box, you enter the environment in that box. Boxes can contain values, programs, text, graphics, or other boxes. The naive realism is based on the notion of "what-you-see-is-what-you-get". Everything can be viewed, selected, and manipulated by the user. In summary, Boxer is put forth as a language in which interactive, reconstructible media can be created or manipulated by the 'ordinary' person. A physics text book in this format could contain a simulation of the principles in the text. The user/reader can run the simulation, manipulate it, and change it.
The development of microworlds for education draws from several, somehow convergent psychological theories. Piaget's ideas about cognitive development constitute one of the main sources. According to Piaget, knowledge acquisition can be viewed as a double process: assimilation and accommodation. Knowledge is not acquired by throwing unrelated pieces together into an empty bag. Each new piece of knowledge is adapted by the learner in order to build a coherent whole with his previous knowledge. This process was referred to by Piaget as assimilation. Sometimes however, the adaptations required to make the new piece of knowledge fit with the prior knowledge are so complex that the learner performs what Piaget called a "process of accommodation". Accommodation may consist of a representational change or even of a radical modification of prior beliefs. Piaget ideas about cognitive development have been very influential in the design of educational programs. In fact, the notion of knowledge grounding discussed below has its origins in Piaget. However, some of his initial claims were so vague that they allow for a varied class of interpretations (Papert's chapter on the issue is a clear example; see the class notes). More recently, cognitive scientists have produced significant advances in understanding how this assimilation process may take place. On one hand, it is known that prior beliefs about the domain of interest (encoded as implicit or explicit theories) affect the learning of new facts in several ways: by focusing the learnerÕs attention on some aspects of the learning data and not on others, by suggesting specific interaction mechanisms that can be checked against the learning materials or by providing the learner with a set of heuristics to guide the learning search. On the other hand, knowledge about a domain affects the acquisition of knowledge in different (related or not) domains, through a process of structural mapping typically known as analogical reasoning. The same process produces the side effect of generating a series of abstract knowledge units that facilitate future transfer even more by saving the need of regenerating the structural mapping. How is this reflected in the design of educational materials? First, concepts are not introduced suddenly at the highest level of complexity the learner needs to master. Instead, the subject matter is explored in detail by the designer, so that materials can be organized in a way that facilitates the learner's assimilation process. Second, analogies are encouraged in order to introduce the learner to completely novel domains (this has not been too frequent so far). Additionally, MicroWorlds stand on the 'constructivist' extreme: the whole thing is based on the notion of 'by trying to build a plane you get to understand what flying is.'
MicroWorlds are very useful for learning domains for which the learner already has intuitions, and domains the learner has direct access to. By building increasingly complex worlds and checking them against their intuitions or against real observations of the domain of interest, learners get to understand the functioning laws. Unfortunately, for fields like Electricity and Magnetism and Quantum Mechanics, the learner would not have initial intuitions to check the quality of the resulting models. In those circumstances, an alternative approach can be used. The instructor develops a series of simulations that work according to the laws the student is trying to understand. The student is left with some limited power to build things (like placing charges in different locations, etc.), and to run its designs. Sometimes, the simulation is packed as a game or some other task with a clearly motivating goal, such that, in the process of learning to achieve the goal (score points, etc.) the student becomes familiar with the environment. This option has some advantages over the ÒpureÓ microworld philosophy in that the student does not need to spend any time in becoming familiar with a new programming environment. Playing video games in an 'Electroworld' is not probably as effective in the long run as building 'Electroworlds', but it produces more immediate benefits for a short period of study time. However, it has been argue that the 'video-game' approach may constrain the knowledge of the student to the game itself, whereas no transfer is guaranteed with respect to the real domain of interest. An intermediate approach (see Guzdial's work for example) would require environments that allow the learner to have access to predefined objects, and leave up to him to use them as such or to explore the "insides" of the objects. Finally, although Microworld researchers claim to stay on the constructivist extreme, it is unclear whether these environments could work without instructor guidance in suggesting topics and adaptations.