“School of Cognitive”

Back to Papers Home
Back to Papers of School of Cognitive

Paper   IPM / Cognitive / 8796
School of Cognitive Sciences
  Title:   biologically inspired framework for learning and abstract representation of attention control
  Author(s): 
1.  Hadi Shariatpanahi
2.  Majid Ahmadabadi
  Status:   Published
  Journal: Lecture Notes in Artificial Intelligent
  Vol.:  4840
  Year:  2007
  Pages:   307-324
  Supported by:  IPM
  Abstract:
The goal of this research is to develop a framework that enables artificial agents to learn active control of their attention as a means toward efficient planning, decision-making, and recognition. The proposed method is inspired by recent findings in psychology and neuroscience that give rise to the assumption that sequential firing of mirror neurons are connected with prediction, recognition, and planning. As these capabilities are connected to active perception, we hypothesize that simulated sequential mirror neurons can provide an abstract representation of learned attention control. The proposed framework consists of three phases. The first phase is designed for learning active control of attention using reinforcement learning. In the second phase, sequential concepts are extracted from the agent?s experience and sequential mirror neurons are generated. In the last phase the concepts represented by these sequential mirror neurons are employed for higher level motor-planning and control of attention, as well as recognition.

Download TeX format
back to top
scroll left or right