Limits...
Visual nonclassical receptive field effects emerge from sparse coding in a dynamical system.

Zhu M, Rozell CJ - PLoS Comput. Biol. (2013)

Bottom Line: Models seeking to explain these non-classical receptive field (nCRF) effects in terms of circuit mechanisms, input-output descriptions, or individual visual tasks provide limited insight into the functional significance of these response properties, because they do not connect the full range of nCRF effects to optimal sensory coding strategies.Furthermore, when the population diversity of an nCRF effect has also been reported in the literature, we show that this model produces many of the same population characteristics.These results show that the sparse coding hypothesis, when coupled with a biophysically plausible implementation, can provide a unified high-level functional interpretation to many response properties that have generally been viewed through distinct mechanistic or phenomenological models.

View Article: PubMed Central - PubMed

Affiliation: Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States of America.

ABSTRACT
Extensive electrophysiology studies have shown that many V1 simple cells have nonlinear response properties to stimuli within their classical receptive field (CRF) and receive contextual influence from stimuli outside the CRF modulating the cell's response. Models seeking to explain these non-classical receptive field (nCRF) effects in terms of circuit mechanisms, input-output descriptions, or individual visual tasks provide limited insight into the functional significance of these response properties, because they do not connect the full range of nCRF effects to optimal sensory coding strategies. The (population) sparse coding hypothesis conjectures an optimal sensory coding approach where a neural population uses as few active units as possible to represent a stimulus. We demonstrate that a wide variety of nCRF effects are emergent properties of a single sparse coding model implemented in a neurally plausible network structure (requiring no parameter tuning to produce different effects). Specifically, we replicate a wide variety of nCRF electrophysiology experiments (e.g., end-stopping, surround suppression, contrast invariance of orientation tuning, cross-orientation suppression, etc.) on a dynamical system implementing sparse coding, showing that this model produces individual units that reproduce the canonical nCRF effects. Furthermore, when the population diversity of an nCRF effect has also been reported in the literature, we show that this model produces many of the same population characteristics. These results show that the sparse coding hypothesis, when coupled with a biophysically plausible implementation, can provide a unified high-level functional interpretation to many response properties that have generally been viewed through distinct mechanistic or phenomenological models.

Show MeSH

Related in: MedlinePlus

Size tuning peak at high vs. low contrast.(A) RF expansion of macaque V1 cells (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 3a from [37]). (B) RF expansion of sparse coding model cells. Most points lie above the diagonal, indicating that (on average) the optimal stimulus size is larger at lower contrasts and the cell demonstrates RF expansion.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3757072&req=5

pcbi-1003191-g004: Size tuning peak at high vs. low contrast.(A) RF expansion of macaque V1 cells (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 3a from [37]). (B) RF expansion of sparse coding model cells. Most points lie above the diagonal, indicating that (on average) the optimal stimulus size is larger at lower contrasts and the cell demonstrates RF expansion.

Mentions: In our implementation, a dictionary optimized for sparse coding with natural scenes was determined via unsupervised learning under sparsity constraints using whitened natural scenes as the training set (whitening is a first-order approximation of retinal processing). The learned dictionary was overcomplete with effective dictionary elements for the pixel image patches used as stimuli. The training set, whitening and learning rule were all exactly as in [14], while the sparse codes during training (i.e., solutions to (2)) were calculated using a standard software package [30] (for computational efficiency) with . We interpret these dictionary elements as the classical spatial receptive fields (CRFs) of the simulated neurons. This interpretation is supported by our own simulated receptive field mapping experiment (results not shown) using sparse dot stimuli, similar to previous studies (e.g., see Fig. 4b in [14]). The results demonstrated in this study are based on the responses of 72 units in this dictionary that had CRFs well-localized within the available image patch (shown in Fig. 1).


Visual nonclassical receptive field effects emerge from sparse coding in a dynamical system.

Zhu M, Rozell CJ - PLoS Comput. Biol. (2013)

Size tuning peak at high vs. low contrast.(A) RF expansion of macaque V1 cells (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 3a from [37]). (B) RF expansion of sparse coding model cells. Most points lie above the diagonal, indicating that (on average) the optimal stimulus size is larger at lower contrasts and the cell demonstrates RF expansion.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3757072&req=5

pcbi-1003191-g004: Size tuning peak at high vs. low contrast.(A) RF expansion of macaque V1 cells (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 3a from [37]). (B) RF expansion of sparse coding model cells. Most points lie above the diagonal, indicating that (on average) the optimal stimulus size is larger at lower contrasts and the cell demonstrates RF expansion.
Mentions: In our implementation, a dictionary optimized for sparse coding with natural scenes was determined via unsupervised learning under sparsity constraints using whitened natural scenes as the training set (whitening is a first-order approximation of retinal processing). The learned dictionary was overcomplete with effective dictionary elements for the pixel image patches used as stimuli. The training set, whitening and learning rule were all exactly as in [14], while the sparse codes during training (i.e., solutions to (2)) were calculated using a standard software package [30] (for computational efficiency) with . We interpret these dictionary elements as the classical spatial receptive fields (CRFs) of the simulated neurons. This interpretation is supported by our own simulated receptive field mapping experiment (results not shown) using sparse dot stimuli, similar to previous studies (e.g., see Fig. 4b in [14]). The results demonstrated in this study are based on the responses of 72 units in this dictionary that had CRFs well-localized within the available image patch (shown in Fig. 1).

Bottom Line: Models seeking to explain these non-classical receptive field (nCRF) effects in terms of circuit mechanisms, input-output descriptions, or individual visual tasks provide limited insight into the functional significance of these response properties, because they do not connect the full range of nCRF effects to optimal sensory coding strategies.Furthermore, when the population diversity of an nCRF effect has also been reported in the literature, we show that this model produces many of the same population characteristics.These results show that the sparse coding hypothesis, when coupled with a biophysically plausible implementation, can provide a unified high-level functional interpretation to many response properties that have generally been viewed through distinct mechanistic or phenomenological models.

View Article: PubMed Central - PubMed

Affiliation: Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States of America.

ABSTRACT
Extensive electrophysiology studies have shown that many V1 simple cells have nonlinear response properties to stimuli within their classical receptive field (CRF) and receive contextual influence from stimuli outside the CRF modulating the cell's response. Models seeking to explain these non-classical receptive field (nCRF) effects in terms of circuit mechanisms, input-output descriptions, or individual visual tasks provide limited insight into the functional significance of these response properties, because they do not connect the full range of nCRF effects to optimal sensory coding strategies. The (population) sparse coding hypothesis conjectures an optimal sensory coding approach where a neural population uses as few active units as possible to represent a stimulus. We demonstrate that a wide variety of nCRF effects are emergent properties of a single sparse coding model implemented in a neurally plausible network structure (requiring no parameter tuning to produce different effects). Specifically, we replicate a wide variety of nCRF electrophysiology experiments (e.g., end-stopping, surround suppression, contrast invariance of orientation tuning, cross-orientation suppression, etc.) on a dynamical system implementing sparse coding, showing that this model produces individual units that reproduce the canonical nCRF effects. Furthermore, when the population diversity of an nCRF effect has also been reported in the literature, we show that this model produces many of the same population characteristics. These results show that the sparse coding hypothesis, when coupled with a biophysically plausible implementation, can provide a unified high-level functional interpretation to many response properties that have generally been viewed through distinct mechanistic or phenomenological models.

Show MeSH
Related in: MedlinePlus