Purchase this article with an account.
Patrick Spröte, Roland Fleming; Bending the truth: Generative models of shape for inferring transformations. Journal of Vision 2014;14(10):892. doi: https://doi.org/10.1167/14.10.892.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We can usually tell whether a tin can on the ground has been crushed or kicked, or if a cookie on a plate has been bitten. In general, humans readily make judgments about the generative processes and transformations that have been applied to objects. Conversely, our ability to recognize objects is often robust across such shape transformations: we can still identify the can even though it has been dented. This ability to determine and discount the causal history of objects suggests the visual system may separate the observed shape of an object into original (untransformed) elements plus the transformations that were applied to it. However, almost no empirical work has investigated to what extent we can extract information about shape transformations and which structural and configural information is used to determine causal history. We conducted two experiments in which we sought to shed light on these questions, using bending as an example transformation. In Experiment 1, we investigated whether subjects could detect and asymmetrically match the degree of transformation (here bending) applied to parametrically generated random 3D shapes. Subjects adjusted the degree of bend applied to a standard object until it appeared as bent as the test object, which had a different 3D orientation. In Experiment 2, observers had to identify the individual objects from Experiment 1 across different transformations. Subjects saw bent versions of each shape and had to identify the corresponding object from a set of untransformed 3D shapes. From our two studies, we conclude that subjects extract information about certain transformations applied to shapes, while ignoring other differences. Our results therefore provide first evidence for scission of a shapes representation into its causes a base shape and a transformation applied to it.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only