Sequences of visual and haptic exploration were obtained on surfaces of different curvature from human subjects. We then extracted regions of interest (ROI) from the data as a function of number of times a subject fixated on a certain location on object and amount of time spent on such each location. Simple models like a plane, cone, cylinder, paraboloid, hyperboloid, ellipsoid, simple-saddle and a monkey-saddle were generated. Gaussian curvature representation of each point on all the surfaces was pre-computed. The surfaces have been previously tested for haptic and visual realism and distinctness by human subjects in a separate experiment. Both visual and haptic rendering were subsequently used for exploration by human subjects to study whether there is a similarity between the visual ROI and haptic ROIs. Additionally, we wanted to see if there is a correlation between curvature values and the ROIs thus obtained. A multiple regression model was further developed to see if this data can be used to predict the visual exploration path using haptic curvature saliency measures.