# Diagram Based Reasoning

## Final Report Summary - DBR (Diagram based reasoning)

1.1 Summary overview of the results
The aim of the present research project was the investigation into the capacity of diagrams to convey information and supporting reasoning and inferences. The objectives of the research project were the following: the evaluation of the cognitive advantages in the use of diagrams; a classification of diagrams.

First, different uses of diagrams to convey information were individuated. Instead of considering the extraction of any propositional content from them, we focus on the way in which they get manipulated to infer some new conclusion by means of them; this happens in continued interaction - and not in contrast - with language. Accordingly, a classification for diagrams was outlined. The classification proposed maps diagrams into two categories, according to their use. In each of these two categories, diagrams are then further distinguished according to the degree of correspondence between the space they display and the data structure they are intended to convey.

The first use is static: the user simply extracts some piece of information from the diagram and for this reason the diagram promotes her memory. Lists, tables, icons, scientific images are kinds of representations that serve as memory aid and therefore promote our mnemonic capacities. The second and more interesting use is dynamic: the user infers some new piece of information from the diagram, and does that by modifying - manipulating - it. Inside these two categories, diagrams are further distinguished according to the degree of correspondence between the particular diagrammatic format and the structure of data that should be conveyed by using that particular diagrammatic format. The correspondence can have degree 0 or at the opposite limit be very high; for example, linguistic descriptions have degree 0 and scientific images have a very high degree.

Secondly, an experiment was performed in order to evaluate the inferential power of the use of diagrams, in collaboration with another member of the research group, Alexis Ouspensky. We focused our study on astronomical phenomena that are within the reach of a naive theory, i.e. that involve naive representations, by requesting subjects to perform some elementary reasoning about the geometry of the corresponding situations. We focus on their visual-spatial reasoning about shadows. We chose to allow some of our subjects to make use of drawings. To our knowledge, these two elements have not been studied so far.

The results showed that drawings (graphs, sketches) improve the performance when answering questions involving visuo-spatial reasoning about non-manipulable objects. The use of images in education, therefore, should be seriously scrutinised.

1.3 Summary of the socioeconomic impacts of the project
The researcher has worked on a website (see http://diagrambasedreasoning.wordpress.com online), in order to improve the discussion about the subject of her research and to connect different people working on the same topic. The theory of diagrammatic reasoning presented here tries to negotiate a difficult line between computational accounts of the brain and situated accounts. A theory of diagrammatic reasoning is more than a model of how this ability can be designed or imitated by an artificial agent. Thus, if successful, it will have immediate and far-reaching consequences for teaching practices, including those that use information technologies, as it will provide a tool for deciding the likely effectiveness of those practices.