The perception-reasoning interface poses a range of interesting questions. There is much evidence that we operate with mental images and mental maps that have analogous properties. This project uses cognitive systems methodology to study aspects of the phenomenon.
In the first phase (2016 — 2018), the question was: how can we as human beings generate mental images and concrete drawings from abstract information, and how can they be so flexible and context-dependent? To answer this question a cognitive system was built, for brevity called imaginer, which leverages a non-monotonic reasoning mechanism, called logical lateration. Logical lateration is based on Context Logic (CL), a formal logic related to Fuzzy Logic developed to handle perceptual/sensory inputs and to be very similar to natural language (NL), in particular with respect to contextualization. The aim of the latter is to make it possible to simply step over the translation between NL and a formal language and just interact with the system in a (very simple) fragment of English.
The result has potential applications in both GIS cartography and human-computer interaction, generally. The inherently context-dependent internal representation would allow for reflecting the human ability to re-conceptualize depending on context, to integrate and separate concerns and to create different visualizations of a (qualitative) knowledgebase.
The system is online at http://logical-lateration.appspot.com
Besides the foundational ramifications of this result and the many direct applications it has in geography, the most interesting application domain is presumably in ethical and explainable AI and particularly in ethical autonomous transportation systems. While machine learning corresponds to — with advanced computing and storage facilities: highly sophisticated — evolutionarily early forms of intelligence, a grounded logic allows the addition of grounded higher-level principles to these systems or, expressed in another way, to add perceptual grounding to logical systems. Machine learning systems and logical reasoning systems can thus be connected in a natural way.
In the second phase, the system will be extended to allow it to leverage logical lateration in more complex reasoning tasks. Besides concrete direct applications, this research may also shed new light on some older questions regarding higher cognition.
- J. D. Kralik, J. H. Lee, P. S. Rosenbloom, P. C. Jackson, S. L. Epstein, O. J. Romero, R. Sanz, O. Larue, H. R. Schmidtke, S. W. Lee, and K. McGreggor. Metacognition for a common model of cognition. Procedia Computer Science 145 (2018): 730-739. (Link)
- H. R. Schmidtke. A canvas for thought. Procedia Computer Science 145 (2018): 805-812. (Link)
- H. R. Schmidtke. A survey on verification strategies for intelligent transportation systems. Journal of Reliable Intelligent Environments, 4(4):211–224, 2018. (Link)
- H. R. Schmidtke. Logical lateration – a cognitive systems experiment towards a new approach to the grounding problem. Cognitive Systems Research, 52:896 – 908, 2018. (Link)
- H. R. Schmidtke. Geometric logics. In P. Blache, H. Christiansen, V. Dahl, D. Duchier, and J. Villadsen, editors, Constraints and Language, chapter 10, pages 219–233. Cambridge Scholars Publishing, 2014. (Download)
- H. R. Schmidtke. Contextual reasoning in context-aware systems. In Workshop Proceedings of the 8th International Conference on Intelligent Environments, page 82–93. IOS Press, 2012. (Download)
- H. R. Schmidtke, D. Hong, and W. Woo. Reasoning about models of context: A context-oriented logical language for knowledge-based context-aware applications. Revue d’Intelligence Artificielle, 22(5):589–608, 2008. (Download)