Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.555818
Title: Anisotropic behaviour when judging shapes in motion
Author: Magnussen, Camilla MacGregor
Awarding Body: Glasgow Caledonian University
Current Institution: Glasgow Caledonian University
Date of Award: 2011
Availability of Full Text:
Access through EThOS:
Abstract:
Detecting objects in motion is a fundamental task that the visual system executes in everyday life in order for us to navigate safely through -- and interact with -- the environment. In order to make appropriate responses to moving objects, such as avoiding a moving branch or catching a ball, the visual system needs to precisely calculate the path an object is travelling. To develop computational models for the determination of such motion, we should first determine how accurately the visual system judges the direction of motion of simple shapes. Tracking of 2D features is, in theory, suffIcient to determine the direction of motion of objects in the fronto-parallel plane. However, large perceptual deviations have been observed for single lines containing features when they are in translation along non-cardinal axes. I have found that, surprisingly, such perceptual deviations also exist for progressively more complex shapes with multiple features such as crosses, squares, rectangles and octagons for translation along non-cardinal axes. When motion is on cardinal axes, typically no or only small perceptual deviations are observed. It is proposed that the perceptual deviations are linked with attraction towards static cues. These cues are the symmetry axes of the object and cardinal reference axes. When these axes are not aligned with the direction of motion, observers perceive motion biased towards them giving rise to substantial perceptual biases. I then investigated if these perceptual deviations would vanish or give rise to a non-linear trajectory if the static object axes were made dynamic: for example by changing orientation whiles the object translates. Therefore, several combinations of rotations and translations were tested. Surprisingly, observers perceived a curved trajectory for both cardinal and non-cardinal axes of translation as long as the orientation of the object axis was close to the axis of translation. This suggests that an object in rotation and translation is wrongly perceived to "slide" on a curved trajectory despite the fact that it is actually rotating and translating along a straight line. This illusion is novel and might be a consequence of the fact that objects in nature often travel along their intrinsic axes of symmetry. A visual system that is confronted with making best use of often ambiguous signals may occasionally make incorrect assignments of motion along trajectories that are closer to the object's orientation than they actually are. . Next attention was turned from motion of rigid objects to investigating non- rigidly moving objects. In natural scenes objects often move non-rigidly with different features belonging to the same object moving in different directions. Such non-rigid motions include those of a swimming jellyfish or a beating heart. This study employed a simple stimulus of three (invisible) apertures with a line segment behind the central one and terminated line segments behind the other two which produces the percept of rigidly or non-rigidly moving lines depending on the terminator velocities. It was discovered that when the terminators were moving in similar directions (rigid or marginally non- rigid motions) the central segment was captured and its velocity given by the vector sum of the velocities of the two terminators. In these circumstances, the motion signal of the central segment did not appear to play a significant role. On the other hand, the segment showed resistance to capture and its motion signal did play a significant role when: (1) ·th~· terminators were moving in opposite directions (e.g. one up and to the right and one down and to the right) giving the percept of a contracting or expanding line, (2) the terminators were moving close to the normal of the line segment, (3) when only one terminator was present. Overall these findings suggest that the visual system computes a vector summation of terminator motions, with the motion signal of the central segment not playing a substantial role when the segment is captured but when capture is reduced the motion signal of the segment should appear as an integral part of the motion computation. A simple model, based on the experimental observations, could account for the data and provides a promising, unified computational approach for signal integration across space in the presence of rigid and non-rigid motion. Abstract 11.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.555818  DOI: Not available
Share: