Complex Probabilistic Inference
Samuel J. Gershman, Jeffrey M. Beck · 2017
Our understanding of probabilistic inference in the brain has progressed rapidly. However, there remains a big gap between the relatively simple probabilistic inference problems facing low-level sensory systems and the intractably complex problems facing high-level cognitive sys- tems. Psychologists have begun exploring cognitively plausible algorithms for approximately solving complex inference problems. We review recent attempts to connect these algorithmic accounts to neural circuit mechanisms, and argue that neural mechanisms for solving low-level sensory inference problems can be extended to tackle complex inference problems.