Rational and Effective Linking Across Conceptual Cases - without Rules

Human reasoning across experiential cases in episodic memory seems quite different from conventional artificial reasoning with conceptual representations by systematically manipulating them according to logical rules. One difference is that in humans linkages between particular experiences can apparently be made in a number of qualitatively different ways, forming recollective chains along different dimensions. For example, watching one movie may recall another which had a similar ending, cinematography, or common actors. It may also recall an otherwise unrelated movie which produced the same emotional response. These linkages do not appear to be economically or simply described by rules. Yet case-based reasoning systems could benefit from sequential indexing of this kind. A conceptual-graph-based FGP machine using a small database of intellectual property law cases could show how such “memory-walks” may be computed without rules.