How do we reason with imprecise spatial descriptions? Do reasoners typically prefer one conclusion (over another) consistent with the imprecise descriptions? Based on empirical findings we are able to give a positive answer for the second question for spatial reasoning with cardinal direction relations. Analyzing further the pattern of the preferred conclusion reflects the idea of informativeness of the description. In consequence, we briefly explain heuristics and present a Bayesian model representing subjective belief of the reasoner.