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

Surround suppression and RF expansion.(A) A plot illustrating that cortical neurons show surround suppression and expansion of CRF size at low contrast (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 1a from [37]). (B) The size tuning curve of a simulated sparse coding model neuron at various contrast levels (“c” stands for contrast, with lighter curves representing lower contrast). The model neuron exhibits two characteristic behaviors reported in the electrophysiology literature: suppression with increasing stimulus size and an increase in the optimal stimulus size with lower contrast. The maximum of each tuning curve is marked by an arrow. (C) Physiologically measured distribution of surround suppression index (SI) in cat V1 (data replotted from [36], Figure 2A), illustrating that most cells do not exhibit significant surround suppression and the SI distribution is relatively uniform among suppressive cells. (D) The SI distribution for the model cells, illustrating the same qualitative properties as the distribution in (C). (E) Distribution of the SI difference (SI) between low and high contrast levels in macaque V1 (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 6b from [37]). The mean difference is 0.06, demonstrating that on average the SI for a cell is contrast invariant. (F) The distribution of SI for the sparse coding model cells. The mean difference is 0.02, also demonstrating contrast invariance in SI.
© Copyright Policy
Related In: Results  -  Collection

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

pcbi-1003191-g003: Surround suppression and RF expansion.(A) A plot illustrating that cortical neurons show surround suppression and expansion of CRF size at low contrast (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 1a from [37]). (B) The size tuning curve of a simulated sparse coding model neuron at various contrast levels (“c” stands for contrast, with lighter curves representing lower contrast). The model neuron exhibits two characteristic behaviors reported in the electrophysiology literature: suppression with increasing stimulus size and an increase in the optimal stimulus size with lower contrast. The maximum of each tuning curve is marked by an arrow. (C) Physiologically measured distribution of surround suppression index (SI) in cat V1 (data replotted from [36], Figure 2A), illustrating that most cells do not exhibit significant surround suppression and the SI distribution is relatively uniform among suppressive cells. (D) The SI distribution for the model cells, illustrating the same qualitative properties as the distribution in (C). (E) Distribution of the SI difference (SI) between low and high contrast levels in macaque V1 (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 6b from [37]). The mean difference is 0.06, demonstrating that on average the SI for a cell is contrast invariant. (F) The distribution of SI for the sparse coding model cells. The mean difference is 0.02, also demonstrating contrast invariance in SI.

Mentions: Similar to end-stopping, some V1 neurons also exhibit surround suppression where their response to a sinusoidal grating patch decreases as the patch size increases beyond the CRF. Additionally, the tuning curve for patch size often exhibits receptive field expansion at low contrast, meaning that the patch size achieving the maximum response increases at low contrast (Fig. 3A). As illustrated in the response of an example model cell shown in Fig. 3B, the sparse coding model can exhibit the same basic suppression and receptive field expansion properties observed in electrophysiology experiments. In addition, we note that the slight increase of response level (i.e., response rebound) at large stimulus size visible in Fig. 3B is also visible in Fig. 3A and discussed elsewhere [34].


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

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

Surround suppression and RF expansion.(A) A plot illustrating that cortical neurons show surround suppression and expansion of CRF size at low contrast (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 1a from [37]). (B) The size tuning curve of a simulated sparse coding model neuron at various contrast levels (“c” stands for contrast, with lighter curves representing lower contrast). The model neuron exhibits two characteristic behaviors reported in the electrophysiology literature: suppression with increasing stimulus size and an increase in the optimal stimulus size with lower contrast. The maximum of each tuning curve is marked by an arrow. (C) Physiologically measured distribution of surround suppression index (SI) in cat V1 (data replotted from [36], Figure 2A), illustrating that most cells do not exhibit significant surround suppression and the SI distribution is relatively uniform among suppressive cells. (D) The SI distribution for the model cells, illustrating the same qualitative properties as the distribution in (C). (E) Distribution of the SI difference (SI) between low and high contrast levels in macaque V1 (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 6b from [37]). The mean difference is 0.06, demonstrating that on average the SI for a cell is contrast invariant. (F) The distribution of SI for the sparse coding model cells. The mean difference is 0.02, also demonstrating contrast invariance in SI.
© Copyright Policy
Related In: Results  -  Collection

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

pcbi-1003191-g003: Surround suppression and RF expansion.(A) A plot illustrating that cortical neurons show surround suppression and expansion of CRF size at low contrast (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 1a from [37]). (B) The size tuning curve of a simulated sparse coding model neuron at various contrast levels (“c” stands for contrast, with lighter curves representing lower contrast). The model neuron exhibits two characteristic behaviors reported in the electrophysiology literature: suppression with increasing stimulus size and an increase in the optimal stimulus size with lower contrast. The maximum of each tuning curve is marked by an arrow. (C) Physiologically measured distribution of surround suppression index (SI) in cat V1 (data replotted from [36], Figure 2A), illustrating that most cells do not exhibit significant surround suppression and the SI distribution is relatively uniform among suppressive cells. (D) The SI distribution for the model cells, illustrating the same qualitative properties as the distribution in (C). (E) Distribution of the SI difference (SI) between low and high contrast levels in macaque V1 (reprinted by permission from Macmillan Publishers Ltd: Nature Neuroscience, Figure 6b from [37]). The mean difference is 0.06, demonstrating that on average the SI for a cell is contrast invariant. (F) The distribution of SI for the sparse coding model cells. The mean difference is 0.02, also demonstrating contrast invariance in SI.
Mentions: Similar to end-stopping, some V1 neurons also exhibit surround suppression where their response to a sinusoidal grating patch decreases as the patch size increases beyond the CRF. Additionally, the tuning curve for patch size often exhibits receptive field expansion at low contrast, meaning that the patch size achieving the maximum response increases at low contrast (Fig. 3A). As illustrated in the response of an example model cell shown in Fig. 3B, the sparse coding model can exhibit the same basic suppression and receptive field expansion properties observed in electrophysiology experiments. In addition, we note that the slight increase of response level (i.e., response rebound) at large stimulus size visible in Fig. 3B is also visible in Fig. 3A and discussed elsewhere [34].

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