Complex cells decrease errors for the muller-lyer illusion in a model of the visual ventral stream

Astrid Zeman, Oliver Obst, Kevin R. Brooks

Research output: Contribution to journalArticleResearchpeer-review

Abstract

To improve robustness in object recognition, many artificial visual systems imitate the way in which the human visual cortex encodes object information as a hierarchical set of features. These systems are usually evaluated in terms of their ability to accurately categorize well-defined, unambiguous objects and scenes. In the real world, however, not all objects and scenes are presented clearly, with well-defined labels and interpretations. Visual illusions demonstrate a disparity between perception and objective reality, allowing psychophysicists to methodically manipulate stimuli and study our interpretation of the environment. One prominent effect, the Muller-Lyer illusion, is demonstrated when the perceived length of a line is contracted (or expanded) by the addition of arrowheads (or arrow-tails) to its ends. HMAX, a benchmark object recognition system, consistently produces a bias when classifying Muller-Lyer images. HMAX is a hierarchical, artificial neural network that imitates the "simple” and "complex” cell layers found in the visual ventral stream. In this study, we perform two experiments to explore the Muller-Lyer illusion in HMAX, asking: (1) How do simple vs. complex cell operations within HMAX affect illusory bias and precision? (2) How does varying the position of the figures in the input image affect classification using HMAX? In our first experiment, we assessed classification after traversing each layer of HMAX and found that in general, kernel operations performed by simple cells increase bias and uncertainty while max-pooling operations executed by complex cells decrease bias and uncertainty. In our second experiment, we increased variation in the positions of figures in the input images that reduced bias and uncertainty in HMAX. Our findings suggest that the Muller-Lyer illusion is exacerbated by the vulnerability of simple cell operations to positional fluctuations, but ameliorated by the robustness of complex cell responses to such variance.

LanguageEnglish
Article number112
Pages1-9
Number of pages9
JournalFrontiers in Computational Neuroscience
Volume8
DOIs
Publication statusPublished - 24 Sep 2014

Fingerprint

Uncertainty
Sagittaria
Benchmarking
Aptitude
Visual Cortex
Tail
Recognition (Psychology)

Bibliographical note

Copyright the Author/s. This Document is protected by copyright and was first published by Frontiers. All rights reserved. It is reproduced with permission.

Cite this

@article{6446516626a74e5abd6ca7650e8616a7,
title = "Complex cells decrease errors for the muller-lyer illusion in a model of the visual ventral stream",
abstract = "To improve robustness in object recognition, many artificial visual systems imitate the way in which the human visual cortex encodes object information as a hierarchical set of features. These systems are usually evaluated in terms of their ability to accurately categorize well-defined, unambiguous objects and scenes. In the real world, however, not all objects and scenes are presented clearly, with well-defined labels and interpretations. Visual illusions demonstrate a disparity between perception and objective reality, allowing psychophysicists to methodically manipulate stimuli and study our interpretation of the environment. One prominent effect, the Muller-Lyer illusion, is demonstrated when the perceived length of a line is contracted (or expanded) by the addition of arrowheads (or arrow-tails) to its ends. HMAX, a benchmark object recognition system, consistently produces a bias when classifying Muller-Lyer images. HMAX is a hierarchical, artificial neural network that imitates the {"}simple” and {"}complex” cell layers found in the visual ventral stream. In this study, we perform two experiments to explore the Muller-Lyer illusion in HMAX, asking: (1) How do simple vs. complex cell operations within HMAX affect illusory bias and precision? (2) How does varying the position of the figures in the input image affect classification using HMAX? In our first experiment, we assessed classification after traversing each layer of HMAX and found that in general, kernel operations performed by simple cells increase bias and uncertainty while max-pooling operations executed by complex cells decrease bias and uncertainty. In our second experiment, we increased variation in the positions of figures in the input images that reduced bias and uncertainty in HMAX. Our findings suggest that the Muller-Lyer illusion is exacerbated by the vulnerability of simple cell operations to positional fluctuations, but ameliorated by the robustness of complex cell responses to such variance.",
author = "Astrid Zeman and Oliver Obst and Brooks, {Kevin R.}",
note = "Copyright the Author/s. This Document is protected by copyright and was first published by Frontiers. All rights reserved. It is reproduced with permission.",
year = "2014",
month = "9",
day = "24",
doi = "10.3389/fncom.2014.00112",
language = "English",
volume = "8",
pages = "1--9",
journal = "Frontiers in Computational Neuroscience",
issn = "1662-5188",
publisher = "Frontiers Media S.A.",

}

Complex cells decrease errors for the muller-lyer illusion in a model of the visual ventral stream. / Zeman, Astrid; Obst, Oliver; Brooks, Kevin R.

In: Frontiers in Computational Neuroscience, Vol. 8, 112, 24.09.2014, p. 1-9.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Complex cells decrease errors for the muller-lyer illusion in a model of the visual ventral stream

AU - Zeman, Astrid

AU - Obst, Oliver

AU - Brooks, Kevin R.

N1 - Copyright the Author/s. This Document is protected by copyright and was first published by Frontiers. All rights reserved. It is reproduced with permission.

PY - 2014/9/24

Y1 - 2014/9/24

N2 - To improve robustness in object recognition, many artificial visual systems imitate the way in which the human visual cortex encodes object information as a hierarchical set of features. These systems are usually evaluated in terms of their ability to accurately categorize well-defined, unambiguous objects and scenes. In the real world, however, not all objects and scenes are presented clearly, with well-defined labels and interpretations. Visual illusions demonstrate a disparity between perception and objective reality, allowing psychophysicists to methodically manipulate stimuli and study our interpretation of the environment. One prominent effect, the Muller-Lyer illusion, is demonstrated when the perceived length of a line is contracted (or expanded) by the addition of arrowheads (or arrow-tails) to its ends. HMAX, a benchmark object recognition system, consistently produces a bias when classifying Muller-Lyer images. HMAX is a hierarchical, artificial neural network that imitates the "simple” and "complex” cell layers found in the visual ventral stream. In this study, we perform two experiments to explore the Muller-Lyer illusion in HMAX, asking: (1) How do simple vs. complex cell operations within HMAX affect illusory bias and precision? (2) How does varying the position of the figures in the input image affect classification using HMAX? In our first experiment, we assessed classification after traversing each layer of HMAX and found that in general, kernel operations performed by simple cells increase bias and uncertainty while max-pooling operations executed by complex cells decrease bias and uncertainty. In our second experiment, we increased variation in the positions of figures in the input images that reduced bias and uncertainty in HMAX. Our findings suggest that the Muller-Lyer illusion is exacerbated by the vulnerability of simple cell operations to positional fluctuations, but ameliorated by the robustness of complex cell responses to such variance.

AB - To improve robustness in object recognition, many artificial visual systems imitate the way in which the human visual cortex encodes object information as a hierarchical set of features. These systems are usually evaluated in terms of their ability to accurately categorize well-defined, unambiguous objects and scenes. In the real world, however, not all objects and scenes are presented clearly, with well-defined labels and interpretations. Visual illusions demonstrate a disparity between perception and objective reality, allowing psychophysicists to methodically manipulate stimuli and study our interpretation of the environment. One prominent effect, the Muller-Lyer illusion, is demonstrated when the perceived length of a line is contracted (or expanded) by the addition of arrowheads (or arrow-tails) to its ends. HMAX, a benchmark object recognition system, consistently produces a bias when classifying Muller-Lyer images. HMAX is a hierarchical, artificial neural network that imitates the "simple” and "complex” cell layers found in the visual ventral stream. In this study, we perform two experiments to explore the Muller-Lyer illusion in HMAX, asking: (1) How do simple vs. complex cell operations within HMAX affect illusory bias and precision? (2) How does varying the position of the figures in the input image affect classification using HMAX? In our first experiment, we assessed classification after traversing each layer of HMAX and found that in general, kernel operations performed by simple cells increase bias and uncertainty while max-pooling operations executed by complex cells decrease bias and uncertainty. In our second experiment, we increased variation in the positions of figures in the input images that reduced bias and uncertainty in HMAX. Our findings suggest that the Muller-Lyer illusion is exacerbated by the vulnerability of simple cell operations to positional fluctuations, but ameliorated by the robustness of complex cell responses to such variance.

UR - http://www.scopus.com/inward/record.url?scp=84907586915&partnerID=8YFLogxK

U2 - 10.3389/fncom.2014.00112

DO - 10.3389/fncom.2014.00112

M3 - Article

VL - 8

SP - 1

EP - 9

JO - Frontiers in Computational Neuroscience

T2 - Frontiers in Computational Neuroscience

JF - Frontiers in Computational Neuroscience

SN - 1662-5188

M1 - 112

ER -