Department or Program
Computational Theories of Mind try to model the mental economy of thoughts by using syntactic representations. Fixing the content of a representation in a physical system is a challenge. In particular, theories try to account for the ability to misrepresent the outside world, use robust representations, and opacity in content. Since these are generally work or language like representations seem to carry, language has dominated as the primary model for the mind. However, this model has not resulted in successful or biologically viable robots. In many cases, computational attempts to build a robot fail to accurately represent and track, change and background knowledge in given situations. In this thesis, I argue for a paradigm shift, following the ideas in Embodied Cognition. In particular, I argue that the brain is best understood as an organ that facilitates action. In this alternative paradigm, the weighted connections between neurons provide complex usage rules for input arrays. Unlike Computational Theories of Mind, Embodied Cognition maintains that representations are best understood as residing “in” the weights, ready for immediate use. Furthermore, I argue that the dynamic environment-body interaction of situated agents can account for complex behavior, while solving the frame problem and modeling the classical properties of representations.
Level of Access
Restricted: Campus/Bates Community Only Access
Date of Graduation
Bachelor of Arts
Kritschgau, Jurgen Desmond, "Context Dependent Representations: Solving the Problem of Content" (2016). Honors Theses. 183.
Number of Pages
Components of Thesis
1 pdf file
Available to Bates community via local IP address or Bates login.