Relative perception system and intelligence

Yujian Li

DOI: https://doi.org/10.1109/MFI.2014.6997674

2014-01-01

Abstract:Physical symbol systems, for general intelligent action, take a physical stimulus as one object in the human world, whereas biological intelligent systems may take it as fundamentally different objects. For eliminating this conflict, the relative perception system hypothesis is proposed as a novel position for the nature of intelligence, based on the principle of world's relativity. From the hypothesis, different relative perception systems, artificial or natural, may have very different understandings of the world, because they observe a relative world of physical objects for producing intelligent behavior. Intelligent behavior is a sequence of actions and operations that it can appropriately choose to achieve a task. Intelligence is an ability of a relative perception system to produce such sequences that may be computational or non-computational, and can be measured by the relative perception intelligence quotient, which is the weighted sum of scores in performing a selected set of tasks. Relative AI is advocated as a new paradigm for artificial intelligence, claiming that different kinds of relative perception systems may have very different understandings of the world, and no brain models can implement the human mind without human body.