DECATHLON USING THE PSYCHOLOGY EXPERIMENT BUILDING LANGUAGE (PEBL)

The Cognitive Decathlon is a proposed set of tasks that can be tested on both human and arti¯cially intelligent agents, and which constitutes a modern speci¯cation for the Turing Test. In this paper, a partial implementation of the Cognitive Decathlon is described using the Psy-chology Experiment Building Language (PEBL). The tasks focus not simply on generic human abilities, but on critical skills that highlight aspects of human performance that are at odds with common arti¯cial intelligence approaches. The di®erences between human and algorithmic behavior in such tasks can reveal properties of the human cognitive architecture, and production of similar behavior by arti¯cial systems can help constrain and validate biologically-inspired systems