Limits...
Information and perception of meaningful patterns.

Del Viva MM, Punzi G, Benedetti D - PLoS ONE (2013)

Bottom Line: In this work we formulate a model of early vision based on a pattern-filtering architecture, partly inspired by high-speed digital data reduction in experimental high-energy physics (HEP).This allows a much stronger data reduction than models based just on redundancy reduction.These results suggest that the limitedness of computing resources can play an important role in shaping the nature of perception, by determining what is perceived as "meaningful features" in the input data.

View Article: PubMed Central - PubMed

Affiliation: NEUROFARBA Dipartimento di Neuroscienze, Psicologia, Area del Farmaco e Salute del Bambino Sezione di Psicologia, Università di Firenze, Firenze, Italy. Michela@in.cnr.it

ABSTRACT
The visual system needs to extract the most important elements of the external world from a large flux of information in a short time for survival purposes. It is widely believed that in performing this task, it operates a strong data reduction at an early stage, by creating a compact summary of relevant information that can be handled by further levels of processing. In this work we formulate a model of early vision based on a pattern-filtering architecture, partly inspired by high-speed digital data reduction in experimental high-energy physics (HEP). This allows a much stronger data reduction than models based just on redundancy reduction. We show that optimizing this model for best information preservation under tight constraints on computational resources yields surprisingly specific a-priori predictions for the shape of biologically plausible features, and for experimental observations on fast extraction of salient visual features by human observers. Interestingly, applying the same optimized model to HEP data acquisition systems based on pattern-filtering architectures leads to specific a-priori predictions for the relevant data patterns that these devices extract from their inputs. These results suggest that the limitedness of computing resources can play an important role in shaping the nature of perception, by determining what is perceived as "meaningful features" in the input data.

Show MeSH
Human contrast sensitivity to visual patterns vs. model predictions.a, Probability distribution of the 512 possible 3×3 1-bit pixel patterns (grey histogram). The curves are the model selection functions (eq. 1) for W = 0.05 and two different values of N.(green: N = 50; blue: N = 15). Green and blue histograms are the probability distributions of corresponding selected patterns. Their actual bandwidth occupancies (∫f(p)>cpδ(p)dp) turn out to be slightly lower (respectively 0.025 and 0.015) than the imposed limit W. Cyan and yellow histograms are the distributions of low-probability patterns used in our measurements. b,c, Visualization of the pattern sets shown in (a), in green and blue respectively. d, Visualization of the lowest-probability patterns (discarded by our approach due to large storage occupation). e, Visualization of the highest-probability patterns (discarded due to large bandwidth occupation). f, Averaged sensitivity for detection of the patterns as a function of their probability, measured on three human subjects (different colors). Errors are determined by the fit (see Methods). The results of pairwise statistical comparisons (z tests, N = 100) amongst sensitivities plotted in (f) are:
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3716808&req=5

pone-0069154-g003: Human contrast sensitivity to visual patterns vs. model predictions.a, Probability distribution of the 512 possible 3×3 1-bit pixel patterns (grey histogram). The curves are the model selection functions (eq. 1) for W = 0.05 and two different values of N.(green: N = 50; blue: N = 15). Green and blue histograms are the probability distributions of corresponding selected patterns. Their actual bandwidth occupancies (∫f(p)>cpδ(p)dp) turn out to be slightly lower (respectively 0.025 and 0.015) than the imposed limit W. Cyan and yellow histograms are the distributions of low-probability patterns used in our measurements. b,c, Visualization of the pattern sets shown in (a), in green and blue respectively. d, Visualization of the lowest-probability patterns (discarded by our approach due to large storage occupation). e, Visualization of the highest-probability patterns (discarded due to large bandwidth occupation). f, Averaged sensitivity for detection of the patterns as a function of their probability, measured on three human subjects (different colors). Errors are determined by the fit (see Methods). The results of pairwise statistical comparisons (z tests, N = 100) amongst sensitivities plotted in (f) are:

Mentions: oWe then evaluated the probability distribution of the patterns in a set of natural images, and extracted the optimal set of patterns as per our recipe (fig. 3a). For this purpose we used a public database of 560 calibrated natural pictures [23] (see examples in fig. 4a). Each image (768×576 pixel) was digitized to 1-bit luminance (black/white), by setting the threshold at its median luminance value (fig. 4b).


Information and perception of meaningful patterns.

Del Viva MM, Punzi G, Benedetti D - PLoS ONE (2013)

Human contrast sensitivity to visual patterns vs. model predictions.a, Probability distribution of the 512 possible 3×3 1-bit pixel patterns (grey histogram). The curves are the model selection functions (eq. 1) for W = 0.05 and two different values of N.(green: N = 50; blue: N = 15). Green and blue histograms are the probability distributions of corresponding selected patterns. Their actual bandwidth occupancies (∫f(p)>cpδ(p)dp) turn out to be slightly lower (respectively 0.025 and 0.015) than the imposed limit W. Cyan and yellow histograms are the distributions of low-probability patterns used in our measurements. b,c, Visualization of the pattern sets shown in (a), in green and blue respectively. d, Visualization of the lowest-probability patterns (discarded by our approach due to large storage occupation). e, Visualization of the highest-probability patterns (discarded due to large bandwidth occupation). f, Averaged sensitivity for detection of the patterns as a function of their probability, measured on three human subjects (different colors). Errors are determined by the fit (see Methods). The results of pairwise statistical comparisons (z tests, N = 100) amongst sensitivities plotted in (f) are:
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3716808&req=5

pone-0069154-g003: Human contrast sensitivity to visual patterns vs. model predictions.a, Probability distribution of the 512 possible 3×3 1-bit pixel patterns (grey histogram). The curves are the model selection functions (eq. 1) for W = 0.05 and two different values of N.(green: N = 50; blue: N = 15). Green and blue histograms are the probability distributions of corresponding selected patterns. Their actual bandwidth occupancies (∫f(p)>cpδ(p)dp) turn out to be slightly lower (respectively 0.025 and 0.015) than the imposed limit W. Cyan and yellow histograms are the distributions of low-probability patterns used in our measurements. b,c, Visualization of the pattern sets shown in (a), in green and blue respectively. d, Visualization of the lowest-probability patterns (discarded by our approach due to large storage occupation). e, Visualization of the highest-probability patterns (discarded due to large bandwidth occupation). f, Averaged sensitivity for detection of the patterns as a function of their probability, measured on three human subjects (different colors). Errors are determined by the fit (see Methods). The results of pairwise statistical comparisons (z tests, N = 100) amongst sensitivities plotted in (f) are:
Mentions: oWe then evaluated the probability distribution of the patterns in a set of natural images, and extracted the optimal set of patterns as per our recipe (fig. 3a). For this purpose we used a public database of 560 calibrated natural pictures [23] (see examples in fig. 4a). Each image (768×576 pixel) was digitized to 1-bit luminance (black/white), by setting the threshold at its median luminance value (fig. 4b).

Bottom Line: In this work we formulate a model of early vision based on a pattern-filtering architecture, partly inspired by high-speed digital data reduction in experimental high-energy physics (HEP).This allows a much stronger data reduction than models based just on redundancy reduction.These results suggest that the limitedness of computing resources can play an important role in shaping the nature of perception, by determining what is perceived as "meaningful features" in the input data.

View Article: PubMed Central - PubMed

Affiliation: NEUROFARBA Dipartimento di Neuroscienze, Psicologia, Area del Farmaco e Salute del Bambino Sezione di Psicologia, Università di Firenze, Firenze, Italy. Michela@in.cnr.it

ABSTRACT
The visual system needs to extract the most important elements of the external world from a large flux of information in a short time for survival purposes. It is widely believed that in performing this task, it operates a strong data reduction at an early stage, by creating a compact summary of relevant information that can be handled by further levels of processing. In this work we formulate a model of early vision based on a pattern-filtering architecture, partly inspired by high-speed digital data reduction in experimental high-energy physics (HEP). This allows a much stronger data reduction than models based just on redundancy reduction. We show that optimizing this model for best information preservation under tight constraints on computational resources yields surprisingly specific a-priori predictions for the shape of biologically plausible features, and for experimental observations on fast extraction of salient visual features by human observers. Interestingly, applying the same optimized model to HEP data acquisition systems based on pattern-filtering architectures leads to specific a-priori predictions for the relevant data patterns that these devices extract from their inputs. These results suggest that the limitedness of computing resources can play an important role in shaping the nature of perception, by determining what is perceived as "meaningful features" in the input data.

Show MeSH