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

Contrast invariant orientation tuning.(A) Contrast invariance of orientation tuning curves recorded in cat V1 (data replotted from [46], Figure 3A). Note that the width of the orientation tuning curve does not change with contrast. (B) Sparse coding model neuron that demonstrates the same invariance property. Lighter curves correspond to lower contrast (“c” denotes contrast level). (C) Distribution of the slope of tuning curve half-width vs. the contrast in ferret V1 (data replotted from [47], Figure 3B). The sharp distribution around 0 indicates that the tuning curve half-width is contrast invariant (mean value is 0.002). (D) Distribution of the half-width vs. the contrast slope in the sparse coding model cells (mean value is 0.032). The model cells clearly demonstrate contrast invariance of the tuning curve half-width, and an even tighter peak around zero slope than shown in (C).
© Copyright Policy
Related In: Results  -  Collection

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

pcbi-1003191-g007: Contrast invariant orientation tuning.(A) Contrast invariance of orientation tuning curves recorded in cat V1 (data replotted from [46], Figure 3A). Note that the width of the orientation tuning curve does not change with contrast. (B) Sparse coding model neuron that demonstrates the same invariance property. Lighter curves correspond to lower contrast (“c” denotes contrast level). (C) Distribution of the slope of tuning curve half-width vs. the contrast in ferret V1 (data replotted from [47], Figure 3B). The sharp distribution around 0 indicates that the tuning curve half-width is contrast invariant (mean value is 0.002). (D) Distribution of the half-width vs. the contrast slope in the sparse coding model cells (mean value is 0.032). The model cells clearly demonstrate contrast invariance of the tuning curve half-width, and an even tighter peak around zero slope than shown in (C).

Mentions: Even when the stimulation is confined to the CRF with no involvement of the surround, cells in V1 exhibit several nonlinear effects that cannot be explained by a canonical linear-nonlinear model [4]. One example of such an effect is the contrast invariance of orientation tuning for V1 cells. In a linear-nonlinear model based on CRFs, higher contrast stimuli evoke stronger responses that more readily exceed the spiking threshold, thus broadening the orientation tuning curve for higher contrast stimuli (the “iceberg effect” [45]). However, as reported in the cat physiology literature, the orientation tuning width is largely contrast invariant [46] as demonstrated in Fig. 7A. Cells from the sparse coding model can also display this contrast invariance in the width of their orientation tuning curves, as shown in Fig. 7B. This invariance can potentially be attributed to recurrent inhibition from competing cells at orientations where the target cell is not the most efficient description (e.g., ortho-oriented stimuli). Even though these competing cells may not have large overlap with the CRF of the target cell, as the contrast increases they will become more active and induce stronger inhibition, thereby narrowing the tuning width of the target cell compared to the low-contrast response. Indeed, compared to the predictions of a linear-nonlinear model (not shown), the tuning width from our model is much narrower.


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

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

Contrast invariant orientation tuning.(A) Contrast invariance of orientation tuning curves recorded in cat V1 (data replotted from [46], Figure 3A). Note that the width of the orientation tuning curve does not change with contrast. (B) Sparse coding model neuron that demonstrates the same invariance property. Lighter curves correspond to lower contrast (“c” denotes contrast level). (C) Distribution of the slope of tuning curve half-width vs. the contrast in ferret V1 (data replotted from [47], Figure 3B). The sharp distribution around 0 indicates that the tuning curve half-width is contrast invariant (mean value is 0.002). (D) Distribution of the half-width vs. the contrast slope in the sparse coding model cells (mean value is 0.032). The model cells clearly demonstrate contrast invariance of the tuning curve half-width, and an even tighter peak around zero slope than shown in (C).
© Copyright Policy
Related In: Results  -  Collection

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

pcbi-1003191-g007: Contrast invariant orientation tuning.(A) Contrast invariance of orientation tuning curves recorded in cat V1 (data replotted from [46], Figure 3A). Note that the width of the orientation tuning curve does not change with contrast. (B) Sparse coding model neuron that demonstrates the same invariance property. Lighter curves correspond to lower contrast (“c” denotes contrast level). (C) Distribution of the slope of tuning curve half-width vs. the contrast in ferret V1 (data replotted from [47], Figure 3B). The sharp distribution around 0 indicates that the tuning curve half-width is contrast invariant (mean value is 0.002). (D) Distribution of the half-width vs. the contrast slope in the sparse coding model cells (mean value is 0.032). The model cells clearly demonstrate contrast invariance of the tuning curve half-width, and an even tighter peak around zero slope than shown in (C).
Mentions: Even when the stimulation is confined to the CRF with no involvement of the surround, cells in V1 exhibit several nonlinear effects that cannot be explained by a canonical linear-nonlinear model [4]. One example of such an effect is the contrast invariance of orientation tuning for V1 cells. In a linear-nonlinear model based on CRFs, higher contrast stimuli evoke stronger responses that more readily exceed the spiking threshold, thus broadening the orientation tuning curve for higher contrast stimuli (the “iceberg effect” [45]). However, as reported in the cat physiology literature, the orientation tuning width is largely contrast invariant [46] as demonstrated in Fig. 7A. Cells from the sparse coding model can also display this contrast invariance in the width of their orientation tuning curves, as shown in Fig. 7B. This invariance can potentially be attributed to recurrent inhibition from competing cells at orientations where the target cell is not the most efficient description (e.g., ortho-oriented stimuli). Even though these competing cells may not have large overlap with the CRF of the target cell, as the contrast increases they will become more active and induce stronger inhibition, thereby narrowing the tuning width of the target cell compared to the low-contrast response. Indeed, compared to the predictions of a linear-nonlinear model (not shown), the tuning width from our model is much narrower.

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