서브메뉴
검색
A New Approach to Algorithm-Oriented Visual Psychophysics
A New Approach to Algorithm-Oriented Visual Psychophysics
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211152134
- ISBN
- 9798384023043
- DDC
- 616
- 서명/저자
- A New Approach to Algorithm-Oriented Visual Psychophysics
- 발행사항
- [Sl] : University of Pennsylvania, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 108 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-02, Section: B.
- 주기사항
- Advisor: Burge, Johannes.
- 학위논문주기
- Thesis (Ph.D.)--University of Pennsylvania, 2024.
- 초록/해제
- 요약The goal of visual neuroscience is to understand how visual systems are able to reconstruct scenes, infer scene properties, and make inferences about the natural environment from a pair of retinal images. Some approaches use psychophysical methods, which allow for controlled sensory presentation and perceptual measurement. Theory holds that through principled application of psychophysics and perceptual modelling, the mechanisms of the visual system can be uncovered. The most common approach has been to measure and model behavioral responses to simple stimuli. However, contemporary neuroscience continues to reveal a more holistic and complex visual system than initially expected. In this work, I argue that these findings suggest behavioral responses to simple stimuli may not as directly correspond to specific visual processing mechanisms, nor provide as powerful an assessment of mechanistic models as originally assumed. Algorithmic theory provides mathematical support for why this is the case and provides a prescription for how psychophysics and modelling can more efficiently explore the space of visual mechanisms. Here, I propose the use of natural or naturalistic images and repeated-measure experimental design to meet this prescription. Not only do natural images provide a diversity of stimuli-an effective substrate for exploring the space of visual mechanisms-but they are also relevant to the overarching goals of vision science-to understand how vision works in the real world. Last, I present the empirically-based portion of this work that acts as as a proof of concept for this framework. In this study I investigate human stereo-depth discrimination performance using naturalistic images and repeated-measures design. I develop broadly applicable methods that can be used to make powerful model assessments based upon the diversity of stimuli. Additionally, the methods and procedures developed provide highly interpretable results: these results show how natural image variability in luminance-patterns and depth-profiles limit human stereo-depth discrimination.
- 일반주제명
- Neurosciences
- 일반주제명
- Psychology
- 일반주제명
- Ecology
- 키워드
- Computation
- 키워드
- Internal noise
- 키워드
- Natural images
- 키워드
- Psychophysics
- 키워드
- Stereopsis
- 기타저자
- University of Pennsylvania Neuroscience
- 기본자료저록
- Dissertations Abstracts International. 86-02B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017163094
■00520250211152134
■006m o d
■007cr#unu||||||||
■020 ▼a9798384023043
■035 ▼a(MiAaPQ)AAI31483970
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a616
■1001 ▼aWhite, David Nathan.
■24512▼aA New Approach to Algorithm-Oriented Visual Psychophysics
■260 ▼a[Sl]▼bUniversity of Pennsylvania▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a108 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-02, Section: B.
■500 ▼aAdvisor: Burge, Johannes.
■5021 ▼aThesis (Ph.D.)--University of Pennsylvania, 2024.
■520 ▼aThe goal of visual neuroscience is to understand how visual systems are able to reconstruct scenes, infer scene properties, and make inferences about the natural environment from a pair of retinal images. Some approaches use psychophysical methods, which allow for controlled sensory presentation and perceptual measurement. Theory holds that through principled application of psychophysics and perceptual modelling, the mechanisms of the visual system can be uncovered. The most common approach has been to measure and model behavioral responses to simple stimuli. However, contemporary neuroscience continues to reveal a more holistic and complex visual system than initially expected. In this work, I argue that these findings suggest behavioral responses to simple stimuli may not as directly correspond to specific visual processing mechanisms, nor provide as powerful an assessment of mechanistic models as originally assumed. Algorithmic theory provides mathematical support for why this is the case and provides a prescription for how psychophysics and modelling can more efficiently explore the space of visual mechanisms. Here, I propose the use of natural or naturalistic images and repeated-measure experimental design to meet this prescription. Not only do natural images provide a diversity of stimuli-an effective substrate for exploring the space of visual mechanisms-but they are also relevant to the overarching goals of vision science-to understand how vision works in the real world. Last, I present the empirically-based portion of this work that acts as as a proof of concept for this framework. In this study I investigate human stereo-depth discrimination performance using naturalistic images and repeated-measures design. I develop broadly applicable methods that can be used to make powerful model assessments based upon the diversity of stimuli. Additionally, the methods and procedures developed provide highly interpretable results: these results show how natural image variability in luminance-patterns and depth-profiles limit human stereo-depth discrimination.
■590 ▼aSchool code: 0175.
■650 4▼aNeurosciences
■650 4▼aPsychology
■650 4▼aEcology
■653 ▼aAlgorithmic theory
■653 ▼aComputation
■653 ▼aInternal noise
■653 ▼aNatural images
■653 ▼aPsychophysics
■653 ▼aStereopsis
■690 ▼a0317
■690 ▼a0621
■690 ▼a0329
■71020▼aUniversity of Pennsylvania▼bNeuroscience.
■7730 ▼tDissertations Abstracts International▼g86-02B.
■790 ▼a0175
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17163094▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
