Associative Computation and Associative Prediction

Abstract

We introduce a biologically and psychologically plausible neuronal model which could explain how pictorial reasoning and learning is carried out by the human brains. This biologically inspired model sheds some light on how some problem solving abilities might actually be performed by the human brain using neural cell assemblies by forming a chain of associations. The model uses picture representation rather than symbolic representation to perform problem solving. The computational task concerning problem solving corresponds to the manipulation of pictures. A computation is performed with the aid of associations by the transformation from an initial state represented as a picture to a desired state represented as a picture. Picture representation allows for the presence of noise and also enables learning from examples. The solved problems are reused to speed up the search for related or similar problems. Either an observer chooses relevant examples or the model learns by experience of failures and successes. The learning from examples is demonstrated by empirical experiments in block world and on a robot in a labyrinth. It is shown that learning improves the behaviour of the model in a statistically significant manner.