TITLE: Multiple representations and visual mental imagery in artificial cognitive systems
SPEAKER: David Peebles (Department of Psychology, University of Huddersfield)

ABSTRACT:
Human cognition is multi-representational.  It typically requires the coordination of different information representations, both external (e.g., text, tables, diagrams, maps, computer interfaces etc.) and within the mind (e.g., logical and probabilistic inference with abstract, amodal, propositional representations, perception based depictive representations, including visual mental imagery where transformations and comparisons are conducted in the "mind's eye").  This has important implications for researchers developing human level artificial agents or computational models of the human mind in that they must consider the types of data structures and reasoning processes required to use and integrate multiple representations, in particular those that underlie visual mental imagery.  In this talk I will make the case for taking multiple representations and visual mental imagery seriously and describe various options for representing visual-spatial information in cognitive architectures.  I will argue that in order to obtain human level cognition, artificial agents must incorporate multiple representational formats and meta-cognitive processes that operate on them.