Purchase this article with an account.
Philip J. Kellman, Patrick B. Garrigan, Don Kalar, Thomas F. Shipley; Good continuation and relatability: Related but distinct principles. Journal of Vision 2003;3(9):120. doi: 10.1167/3.9.120.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Factor of Good Curve (commonly known as good continuation) was proposed by Wertheimer (1921) in the context of segmentation of visible areas into objects or parts. Relatability (Kellman & Shipley, 1991) expresses the geometric conditions for interpolation of contours across gaps, in occluded and illusory contours. Many investigators equate these principles.
In this talk, we will argue that, although they both enforce certain smoothness properties, good continuation and relatability are distinct principles, each required to understand visual segmentation and grouping phenomena. As it is usually defined only by examples, we first put forth a formal definition of good continuation: Continuity vs. segmentation depends on the presence or absence of zero-order and first-order contour discontinuities. We will present data and demonstrations supporting this definition.
Relatability — which governs contour interpolation across gaps — differs in 5 ways from good continuation. Zero- and first-order discontinuities segment visible arrays (good continuation) but are prerequisites for unit formation via relatability. Relatable edges must be connectable by a monotonic curve; also, they cannot bend through more than 90 deg. We will present demonstrations and data showing that good continuation follows neither the monotonicity nor 90 deg constraints. Finally, breaches of good continuation produce either separate objects or parts, whereas failures of relatability (in the absence of completion by surface spreading) produce distinct objects. Good continuation and relatability express related but distinct laws, applicable to different issues of perceptual organization.
This PDF is available to Subscribers Only