Feature Inference and the Causal Structure of Categories: A Modification
In extension to the causal feature inference experiment conducted by Rehder and Burnett (Rehder & Burnett, 2005), this paper further studies the psychological representation of causal knowledge based on Bayesian networks, especially in regards to the causal Markov condition. The results of this experiment indicate that people do perceive causality according to the causal Markov condition, in that they take into account only the causal factor and discard all other information when inferring unobserved features in a machine. Given that the study was a modified version of Rehder and Burnett, which previously observed people’s violation of the causal Markov condition, the results suggest that at least one or more of the modifications made in this study had significantly influenced participants to represent causality according to the causal Markov condition. Exactly what modification made this change is unclear; nevertheless, the results provide direction for future research that will reveal additional insights regarding how people perceive, interpret, and interact with the world around them.
Keywords - Psychology, Psychology of thinking, Cognitive psychology, Human cognition, Social sciences, Human learning and education, Baysian network, Causal Markov condition, Perception of causality.