Unique contributions of skeletal structure to shape perception and object recognition Public

Ayzenberg, Vladislav (Summer 2020)

Permanent URL: https://etd.library.emory.edu/concern/etds/6q182m22z?locale=fr
Published

Abstract

With seemingly little effort, humans can both identify an object across large changes in orientation and extend category membership to novel exemplars. Most remarkably, humans can accomplish these feats with little experience, categorizing never-before-seen objects from as little as one training exemplar. Although researchers have long suggested that global shape information is crucial for robust object recognition, it is unknown how humans perceptually organize visual information to create global shape percepts and use these percepts to recognize objects. In the current dissertation, I used behavioral, neural, computational, and developmental methods to test the hypothesis that a model of structure known as the medial axis, or shape skeleton, can support both perceptual organization and object recognition. Moreover, I examined whether shape skeletons play a role in rapid object learning by testing whether they can support one-shot categorization in infants. Consistent with these hypotheses, I found that a skeletal model was predictive of adult participants’ object similarity and category judgments (Study 1). Moreover, neuroimaging of the adult visual system revealed that a skeletal model was predictive of the multivariate response in V3 and lateral occipital cortex (LO), regions implicated in perceptual organization and object recognition, respectively (Study 2). Finally, I found that 6- to 12-month-old infants, a population with little object experience, could categorize never-before-seen objects by their skeleton after seeing just one exemplar (Study 3). In all studies, the skeletal model best fit participants’ responses (behavioral and neural) across changes in image-level properties, contour, and when controlling for other state-of-the-art artificial neural networks. Taken together, these studies highlight the unique and privileged role of shape skeletons in perceptual organization, object recognition, and one-shot categorization.

Table of Contents

Abstract

1

Chapter 1: General Introduction

2

Chapter 2: Skeletal descriptions of shape provide unique perceptual information for object recognition

Introduction

9

Experiment 1

11

Experiment 2

16

Experiment 3

19

Discussion

24

Supplementary Materials

26

Chapter 3: A dual role for shape skeletons in human vision: Perceptual organization and object recognition

Introduction

40

Methods

41

Results

48

Discussion

54

Supplementary Materials

59

Study 4: The shape skeleton supports single exemplar categorization in infants

Introduction

64

Experiment 1

64

Experiment 2

68

Discussion

71

Supplementary Materials

72

Chapter 5: General Discussion

76

References

83

About this Dissertation

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Subfield / Discipline
Degree
Submission
Language
  • English
Research Field
Mot-clé
Committee Chair / Thesis Advisor
Committee Members
Dernière modification

Primary PDF

Supplemental Files