The Incoherence of Heuristically Explaining Coherence
Iris van Rooij, Cory Wright · eScholarship (California Digital Library) · 2006
doing some occasional “theoretical housekeeping. ” In this poster, we highlight some conceptual confusions lurking in an important attempt at explaining the human ability for rational or coherent thought: Thagard and Verbeurgt’s computational-level model of humans ’ capacity for making reasonable and truth-conducive abductive inferences (1998; Thagard, 2000). T&V’s model assumes that humans make such inferences by computing a coherence function (fcoh), which takes as input representation networks and their pair-wise constraints and gives as output a partition into accepted (A) and rejected (R) elements that maximizes the weight of satisfied constraints. We argue that their proposal gives rise to at least three difficult problems. Being NP-Hard Results in a Dilemma T&V proved that fcoh is NP-hard. This result proves that