Representation and Computation in Cognitive Models

• Published in: Topics in cognitive science Vol. 9; no. 3; pp. 694 - 718
• Main Authors: Forbus, Kenneth D., Liang, Chen, Rabkina, Irina
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.07.2017
• Subjects: Analogy; Cognition; Cognitive models; Cognitive Science; Computational modeling; Humans; Learning; Machine learning; Models, Theoretical; Relational representations; Representation; Symbolic modeling; Learning; Relational representations; Symbolic modeling; Computational modeling; Machine learning; Analogy; Representation
• ISSN: 1756-8757; 1756-8765
• DOI: 10.1111/tops.12277

One of the central issues in cognitive science is the nature of human representations. We argue that symbolic representations are essential for capturing human cognitive capabilities. We start by examining some common misconceptions found in discussions of representations and models. Next we examine evidence that symbolic representations are essential for capturing human cognitive capabilities, drawing on the analogy literature. Then we examine fundamental limitations of feature vectors and other distributed representations that, despite their recent successes on various practical problems, suggest that they are insufficient to capture many aspects of human cognition. After that, we describe the implications for cognitive architecture of our view that analogy is central, and we speculate on roles for hybrid approaches. We close with an analogy that might help bridge the gap. This paper makes the case that recent advances in models using distributed representations do not obviate the need for symbolic representations in computational systems that aim for human‐level cognitive performance. The authors describe several limitations of distributed representations and then explore the power and efficiency of symbolic models. They suggest that there may be value in hybrid systems that make use of the strengths of distributed representations as well as the power of structured representations.