Efficient coding of natural scene statistics predicts discrimination thresholds for grayscale textures.

Previously, in Hermundstad et al., 2014, we showed that when sampling is limiting, the efficient coding principle leads to a 'variance is salience' hypothesis, and that this hypothesis accounts for visual sensitivity to binary image statistics. Here, using extensive new psychophysical data and image analysis, we show that this hypothesis accounts for visual sensitivity to a large set of grayscale image statistics at a striking level of detail, and also identify the limits of the prediction. We define a 66-dimensional space of local grayscale light-intensity correlations, and measure the relevance of each direction to natural scenes. The 'variance is salience' hypothesis predicts that two-point correlations are most salient, and predicts their relative salience. We tested these predictions in a texture-segregation task using un-natural, synthetic textures. As predicted, correlations beyond second order are not salient, and predicted thresholds for over 300 second-order correlations match psychophysical thresholds closely (median fractional error <0.13).

Introduction

Neural circuits in the periphery of the visual (Laughlin, 1981; Atick and Redlich, 1990; van Hateren, 1992; Fairhall et al., 2001; Ratliff et al., 2010; Liu et al., 2009; Borghuis et al., 2008; Garrigan et al., 2010; Kuang et al., 2012), auditory (Schwartz and Simoncelli, 2001; Lewicki, 2002; Smith and Lewicki, 2006; Carlson et al., 2012), and perhaps also olfactory (Teşileanu et al., 2019) systems use limited resources efficiently to represent sensory information by adapting to the statistical structure of the environment (Sterling and Laughlin, 2015). There is some evidence that this sort of efficient coding might also occur more centrally, in the primary visual cortex (Olshausen and Field, 1996; Bell and Sejnowski, 1997; Vinje and Gallant, 2000; van Hateren and van der Schaaf, 1998) and perhaps also in the entorhinal cortex (Wei et al., 2015). Behaviorally, efficient coding implies that the threshold for perceiving a complex sensory cue, which depends on the collective behavior of many cells in a cortical circuit, should be set by its variance in the natural environment. However, the nature of this relationship depends on the regime in which the sensory system operates. Specifically, in conventional applications of efficient coding theory where sampling is abundant, high variance is predicted to be matched by high detection thresholds. The authors of Hermundstad et al., 2014 argued instead that texture perception occurs in a regime where sampling noise is the limiting factor. This leads to the opposite prediction (Tkacik et al., 2010; Doi and Lewicki, 2011; Hermundstad et al., 2014), namely that high variance should lead to a low detection threshold, summarized as variance is salience (Hermundstad et al., 2014). Tests of this prediction in Tkacik et al., 2010; Hermundstad et al., 2014 showed that it holds for the visual detection of simple black-and-white binary textures.

These binary textures, while informative about visual sensitivity, are a highly restricted set and do not capture many perceptually-salient properties of natural scenes. Moving to a complete description of visual textures, however, requires specifying the co-occurrence of all possible patterns of light across a visual image, and is generally intractable. One way to make this specification tractable is to construct a local and discretized grayscale texture space, in which luminance is drawn from a discrete set and correlations in luminance are only specified up to a given neighborhood size. For example, if we consider four spatially-contiguous squares (‘checks’) with binary intensities, there are patterns that can each occur with different probabilities in a given texture. Imposing translation symmetry constrains these 16 parameters, leading to a 10-dimensional space of textures (Hermundstad et al., 2014). This space can be explored by synthesizing artificial binary textures with prescribed combinations of parameters (Victor et al., 2005; Victor and Conte, 2012), and by analyzing the relative contributions of these parameters to the correlated structure of natural images (Tkacik et al., 2010; Hermundstad et al., 2014). Here, we generalized these synthesis and analysis methods to multiple gray levels and used this to probe efficient encoding of grayscale textures composed of correlated patterns of three luminance levels (dark, gray, light) specified within blocks of four contiguous checks. We chose to add only one intermediate gray level compared to the binary case because it is the simplest generalization of binary textures that allows us to explore perceptual sensitivity to grayscale textures. Because the number of possible visual patterns increases as a power law in the number of distinguishable luminance values, this generalization already yields a very high-dimensional space: for intensities constrained to discrete values, there are patterns of four checks with three gray levels, leading to a 66-dimensional space of textures after accounting for the constraints of translation invariance.

This grayscale texture space enabled us to probe and interpret the relationship between natural scene statistics and psychophysics in much greater detail than is possible with binary textures. In particular, the ‘variance is salience’ hypothesis qualitatively predicts that directions corresponding to two-point correlations will be most perceptually salient, and it quantitatively predicts detection thresholds in different directions of this salient part of the texture space. (Two coordinates corresponding to contrast, which are also highly salient, are zeroed out by the preprocessing in our natural image analysis; we therefore do not probe these directions.) We tested these predictions by asking observers to report the location of textured strips presented rapidly against a background of white noise, and we found detailed agreement with the theory. By further exploiting symmetries in the distribution of grayscale textures, we show that human behavior not only reflects the relative informativeness of natural visual textures, but it also parallels known invariances in natural scenes. Natural scenes also have a notable, previously studied, asymmetry between bright and dark (Ratliff et al., 2010; Tkacik et al., 2010) which is reflected in the anatomy and physiology of visual circuits, and in visual behavior (Ratliff et al., 2010; Tkacik et al., 2010; Zemon et al., 1988; Chubb et al., 2004; Jin et al., 2008; Komban et al., 2014; Kremkow et al., 2014). The asymmetry is rooted ultimately in the lognormal distribution and spatial correlation structure of light intensities in natural scenes (Ratliff et al., 2010; Tkačik et al., 2011). Our image processing pipeline (see Materials and methods) starts by taking the logarithm of the pixel intensities and removing the large-scale spatial correlations, and thus significantly reduces the dark-light asymmetry. The ternarization procedure that we then employ further diminishes the asymmetry, allowing us to focus on other aspects of natural scene statistics.

Results

Local textures with multiple gray levels

We define textures in terms of statistical correlations between luminance levels at nearby locations, generalizing the methods developed for binary images (Victor and Conte, 2012; Hermundstad et al., 2014) to three luminance levels. If we consider four ‘checks’ arranged in a square, the three luminance levels lead to possible patterns, and their frequency of occurrence in an image is equivalently parameterized by intensity correlations within the square. Thus there is an 81-dimensional space of ternary textures defined by correlations within square arrangements of checks. However, translation invariance constrains these 81 probabilities, reducing the number of independent statistics and thus the dimension of the texture space.

We can quantify the statistics of such textures in an image patch by gliding a block (a ‘glider’) over the patch and analyzing the luminance levels at different locations within this block (Figure 1A). At the most basic level, we can measure the luminance histogram at each of the four check locations in the glider. Check intensities can take three values (0, 1, or 2, for black, gray, or white), and the corresponding frequencies of occurrence must add to one, leaving two free parameters. If the histograms at each of the four locations within the glider were independent, this would lead to texture dimensions. However, because of translation invariance in natural images, the luminance histograms at each location must be the same, leaving only two independent dimensions of texture space from the single-check statistics.

Figure 1.

Ternary texture analysis.

(A) With three luminance levels there are possible check configurations for a block (histogram on the right). We parametrize the pairwise correlations within these blocks using modular sums or differences of luminance values at nearby locations, (see Materials and methods and Appendix 1 for details). This notation denotes the remainder after division by 3, so that for example and . The texture coordinates are defined by the probabilities , , with which equals its three possible values, 0, 1, or 2. These three probabilities must sum to 1, so there are only two independent coordinates for each triplet of probabilities. (B) The eight second-order groups (planes) of texture coordinates in the ternary case. A texture group is identified by the choice of orientation of the pair of checks for which the correlation is calculated, and by whether a sum or a difference of luminance values is used. The greek letter notation ( for the second-order planes) mirrors the notation used in Hermundstad et al., 2014. (C and D) Example texture groups (‘simple’ planes). The origin is the point , representing an unbiased random texture. The interior of the triangle shows the allowed range in the plane where all the probability values are non-negative. The vertices are the points where only one of the probabilities is nonzero. An example texture patch is shown for the origin, as well as for each of the vertices of the probability space.

Next, we can analyze the statistics of luminance levels at pairs of locations within the glider. Taking into account translation invariance, there are four ways to position these pairs (Figure 1B), each with a different orientation. For each orientation, we can calculate either the sum or the difference of the luminance values and at the two locations. This yields eight possible texture groups (Figure 1B). Within each group, we build texture coordinates by counting the fraction of occurrences in which is equal to 0, 1, or 2, up to a multiple of 3; that is, we are building a histogram of  modulo 3, here denoted . The appearance of the modulo function here is a consequence of using a Fourier transform in the space of luminance levels, which is convenient for incorporating translation invariance into our coordinate system (see Materials and methods and Appendix 1). The fractions of values of equal to 0, 1, or 2 must add up to 1, so that each texture group is characterized by two independent coordinates, that is, a plane in texture space. These 8 planes constitute 16 independent dimensions of the texture space, in addition to the two dimensions needed to capture the histogram statistics at individual locations.

We can similarly analyze the joint statistics of luminance levels at three checks within the glider, or all four together. There are four ways to position 3-check gliders within the square, and, for each of these positions, eight parameters are needed to describe their occurrence frequencies, once the first- and second-order statistics have been fixed. This leads to third-order parameters. For configurations of all four checks, 16 parameters are required, once first-, second-, and third-order parameters are fixed. These parameters, in addition to the 18 parameters described above, lead to a 66-dimensional texture space. This provides a complete parameterization of the configurations with three gray levels. See Methods and Appendix 1 for a detailed derivation and a generalization to higher numbers of gray levels.

In order to probe human psychophysical sensitivity to visual textures, we need an algorithm for generating texture patches at different locations in texture space. To do so, we use an approach that generalizes the methods from Victor and Conte, 2012. Briefly, we randomly populate entries of the first row and/or column of the texture patch, and we then sequentially fill the rest of the entries in the patch using a Markov process that is designed to sample from a maximum-entropy distribution in which some of the texture coordinates are fixed (see Methods and Appendix 2 for details). Examples of texture patches obtained by co-varying coordinates within a single texture group are shown in Figure 1C and D. Examples of textures obtained by co-varying coordinates in two texture groups are shown in Figure 2. We refer to the first case as ‘simple’ planes, and the second as ‘mixed’ planes.

Figure 2.

Examples of textures from all the mixed planes for which psychophysics data are available.

Each patch is obtained by choosing coordinates in two different texture groups: for instance, a point in the plane (row 1, column 2) corresponds to choosing the probabilities that and . Apart from these constraints, the texture is generated to maximize entropy (see Materials and methods and Appendix 2). The center of the coordinate system in all planes corresponds to an unbiased texture (i.e., the probability for each direction is 1/3), while a mixed-plane coordinate equal to one corresponds to full saturation (i.e., the probability for that direction is 1). The dashed lines indicate the directions in texture space along which the illustrated patches were generated. The patches within a given plane are drawn at a constant distance from the center, but the precise amount of texture saturation varies, according to the largest saturation that could be generated in each direction. Note that along some directions, the maximum saturation is limited by the way in which the texture coordinates are defined, or by the texture synthesis procedure (see Appendix 2).

When applying these methods to the analysis of natural images, we bin luminance values to produce equal numbers of black, white, and gray checks (details below), thus equalizing the previously studied brightness statistics in scenes (e.g., Zemon et al., 1988; Chubb et al., 2004; Jin et al., 2008; Ratliff et al., 2010; Tkacik et al., 2010; Tkačik et al., 2011; Komban et al., 2014; Kremkow et al., 2014). This procedure allowed us to focus on higher-order correlations. In previous work, we found that median-binarized natural images show the highest variability in pairwise correlations, and that observers are correspondingly most sensitive to variations in these statistics (Tkačik et al., 2011; Hermundstad et al., 2014). In view of this, we focused our attention on pairwise statistics. For three gray levels, these comprise a 16-dimensional ‘salient’ subspace of the overall texture space.

Natural image statistics predict perceptual thresholds

Predictions from natural scene statistics

The ‘variance is salience’ hypothesis from Hermundstad et al., 2014 predicts that the most salient directions in texture space (for which detection thresholds are low) will be those along which there is the highest variance across image patches, while the least salient directions (for which detection thresholds are high) will be those with lowest variance. To test these predictions, we first mapped natural image patches to their corresponding locations within the texture space (as in Hermundstad et al., 2014). We then computed the inverse of the standard deviation of the natural image distribution along each direction, and we used this as our prediction of detection thresholds. The procedure is sketched in Figure 3 and described in more detail in Methods. We found that natural images have much higher variance in the second-order coordinate planes than in the third- and fourth-order planes. This predicted that textures exhibiting variability in the second-order planes would be most salient to observers, and thus would be amenable to a quantitative comparison between theory and behavior. Thus, we computed the predicted detection thresholds in four single and twenty-two mixed second-order coordinate planes, and we scaled this set of thresholds by a single overall scaling factor that was chosen to best match the behavioral measurements (blue dots in Figure 4C and Figure 5).

Figure 3.

Preprocessing of natural images.

(A) Images (which use a logarithmic encoding for luminance) are first downsampled by a factor and split into square patches of size . The ensemble of patches is whitened by applying a filter that removes the average pairwise correlations (see panel C), and finally ternarized after histogram equalization (see panel D). (B) Blurry images are identified by fitting a two-component Gaussian mixture to the full distribution of image textures (shown in light gray). This is shown here in a particular projection involving a second-order direction () and a fourth-order one . The texture analysis is restricted to the component with higher contrast, which is shown in black on the plot. Note that a value of 1/3 on each axis corresponds to the origin of the texture space. (C) Power spectrum before and after filtering an image from the dataset. (D) Images are ternarized such that within each patch a third of the checks are converted to black, a third to gray, and a third to white. The processing pipeline illustrated here extends the analysis of Hermundstad et al., 2014 to multiple gray levels.

Figure 4.

Experimental setup and results in second-order simple planes.

(A) Psychophysical trials used a four-alternative forced-choice task in which the subjects identified the location of a strip sampled from a different texture on top of a background texture. (B) The subject’s performance in terms of fraction of correct answers was fit with a Weibull function and the threshold was identified at the mid-point between chance and perfect performance. Note that if the subject’s performance never reaches the mid-point on any of the trials, this procedure may extrapolate a threshold that falls outside the valid range for the coordinate system (see, e.g., the points outside the triangles in panel C). This signifies a low-sensitivity direction of texture space. (C) Measured thresholds (red crosses with pink error bars; the error bars are in most cases smaller than the symbol sizes) and predicted thresholds (blue dots) in second-order simple planes. Thresholds were predicted to be inversely proportional to the standard deviation observed in each texture direction in natural images. The plotted results used downsampling factor and patch size . A single scaling factor for all planes was used to match to the psychophysics. The orange and green dashed lines show the effect of two symmetry transformations on the texture statistics (see text).

Figure 5.

The match between measured (red crosses and error bars) and predicted (blue dots) thresholds in 22 mixed planes.

Each plot corresponds to conditions in which the coordinates in two different texture groups are specified, according to the axis labels. For instance, column two in row one is the plane; the two coordinates correspond to choosing the probabilities that and . As in Figure 2, the center of the coordinate system in these planes corresponds to an unbiased texture (i.e., the probability for each direction is 1/3), while a coordinate equal to 1—indicated by the gray dotted circle—corresponds to full saturation (i.e., the probability for that direction is 1).

Psychophysical measurements

To measure the sensitivity of human subjects to different kinds of textures, we used a four-alternative forced-choice paradigm following Hermundstad et al., 2014; Victor and Conte, 2012; Victor et al., 2013; Victor et al., 2015. Subjects were briefly shown an array in which a rectangular strip positioned near the left, right, top, or bottom edge was defined by a texture difference: either the strip was structured and the background was unstructured, or the background was structured and the strip was unstructured. Structured patterns were constructed using the texture generation method described above, and unstructured patterns were generated by randomly and independently drawing black, gray, or white checks with equal probability. The texture analysis procedure in Figure 3 included a whitening step that removed long-range correlations in natural images, and the local textures generated to test the predictions psychophysically also lacked these correlations. This is appropriate, because, during natural vision, dynamic stimuli engage fixational eye movements that whiten the visual input (Rucci and Victor, 2015). Spatial filtering has also been thought to play a role in whitening (Atick and Redlich, 1990), but in vitro experiments (Simmons et al., 2013 found that adaptive spatiotemporal receptive field processing did not by itself whiten the retinal output, but rather served to maintain a similar degree of correlation across stimulus conditions). We infer from these studies that short visual stimuli like our 120 ms presentations should be pre-whitened, to make up for the absence of fixational eye movements that produce whitening in natural, continuous viewing conditions.

Subjects were asked to indicate the position of the differently textured strip within the array (Figure 4A). Thresholds were obtained by finding the value of a texture coordinate for which the subjects’ performance was halfway between chance and perfect (Figure 4B; see Materials and methods for details). For the second-order planes, subjects were highly consistent in their relative sensitivity to different directions in texture space, with a single scaling factor accounting for a majority of the inter-subject variability (see Appendix 6). The subject-average thresholds in the second order planes are shown in Figure 4C and Figure 5 (red crosses and error bars). As predicted by the natural scene analysis, sensitivity in the third and fourth-order planes was low; in fact, detection thresholds could not be reliably measured in most directions beyond second order (Appendix 6).

Variance predicts salience

Predicted detection thresholds were in excellent agreement with those measured experimentally (Figure 4C and Figure 5), with a median absolute log error of about 0.13. Put differently, 50% of the measurements have relative errors below 13%, since in this regime, the log error is very well approximated by relative error (see Materials and methods and Appendix 4). This match is unlikely to be due to chance—a permutation test yields for the hypothesis that all measured thresholds were drawn independently from a single distribution that did not depend on the texture direction in which they were measured (see Materials and methods and Appendix 4 for details and further statistical tests; 95% of the 10,000 permutation samples exhibited median absolute log errors in the range [0.24, 0.31]). The comparison between theory and experiment was tested in 12 directions in each of 26 single and mixed planes, for a total of 311 different thresholds (the measurement uncertainty in one direction in the mixed plane was too large, and we discarded that datapoint); a single scaling factor was used to align these two sets of measurements. Note that these measurements are not fully independent: the natural image predictions within each plane lie on an ellipse by construction; the psychophysical thresholds are measured independently at each point but are generally well-approximated by ellipses. Even taking this into account, the match between the predictions and the data is unlikely to be due to chance (; 95% range for the median absolute log error [0.16, 0.22]; see Appendix 4).

However, not all thresholds are accurately predicted from natural images. While some of the mismatches seem random and occur for directions in texture space where the experimental data have large variability (pink error bars in Figure 4C and Figure 5), there are also systematic discrepancies. Natural-image predictions tend to underestimate the threshold in the simple planes (median log prediction error −0.090) and overestimate the threshold in mixed planes (median log prediction error +0.008). That is, in human observers, detection of simultaneously-present multiple correlations is disproportionally better than predicted from the detection thresholds for single correlations, to a mild degree (see Appendix 5 for details).

A second observation is that prediction errors tend to be larger for the sum-correlations (such as ) than for the difference-correlations (such as ), independent of the direction of the error. This may be a consequence of the way that modular arithmetic and gray-level discretization interact, leading to a kind of non-robustness of the sum-correlations. Specifically, the most prominent feature of the patterns induced by the sum-correlations is that there is a single gray level that occurs in runs (e.g., leads to runs of black checks, leads to runs of white checks, and leads to runs of gray checks; see Figure 1C). On the other hand, for difference correlations, leads to runs of all gray levels, while and lead to ‘mini-gradients’ that cycle between white, gray, and black (Figure 1D). Thus, the sum-correlations are subject to the particular assignment of gray levels and modulus while the difference-correlations rely on relationships that hold independent of these choices.

Independent of these trends, the natural-image predictions tend to underestimate the thresholds in directions with very low variance even while they match the thresholds in directions with high variance (see Appendix 5). This suggests the need to go beyond the linear efficient-coding model employed here. A simple generalization that interpolates between existing analytically solvable models (Hermundstad et al., 2014) involves a power-law transformation of the natural-image variances, . We use a default exponent of throughout the text. The exponent that best matches our data is close but probably not equal to 1 (the 95% credible interval is [0.81, 0.98]; see Appendix 4), suggesting that a weak power-law nonlinearity might be involved. The inferred range for also confirms that the measured thresholds are not independent of the predicted ones (which would have mapped to ).

In sum, the modest systematic discrepancies between the natural-image predictions and the measured thresholds indicate that the efficient-coding model has limitations, and the observed mismatches can guide future studies that go beyond this model. More generally, we do not expect the predictions of efficient coding to hold indefinitely: adapting to increasingly precise environmental statistics must ultimately become infeasible, both because of insufficient sampling, and because of the growing computational cost required for adaptation. Whether, and to what extent, these issues are at play is a subject for future work.

These results are robust to several variations in our analysis procedure. We obtain similar results when we either vary the sub-sampling factor and patch size , modify the way in which we ternarize image patches, or analyze different image datasets (Figure 6A and Appendix 3). Eliminating downsampling completely (choosing ) does lead to slightly larger mismatches between predicted and measured thresholds (first three distributions on the left in Figure 6A) as expected from Hermundstad et al., 2014, a finding that we attribute to artifacts arising from imperfect demosaicing of the camera’s filter array output.

Figure 6.

Robustness of results and effects of symmetry transformations.

(A) The difference between the natural logarithms of the measured and predicted thresholds (red crosses and blue dots, respectively, in Figures 4C and 5) is approximately independent of the downsampling ratio and patch size used in preprocessing. The labels on the x-axis are in the format , with the violin plot and label in blue representing the analysis that we focused on in the rest of the paper. Each violin plot in the figure shows a kernel density estimate for the distribution of prediction errors for the 311 second-order single- and mixed-plane threshold measurements available in the psychophysics. The boxes show the 25th and 75th percentiles, and the lines indicate the medians. (B) Change in the natural logarithms of predicted (blue) or measured (red) thresholds following a symmetry transformation. Symmetry transformations that leave the natural image predictions unchanged also leave the psychophysical measurements unchanged. (See text for the special case of the exch(B,W) transformation.) The visualization style is the same as in panel A, except boxes and medians are not shown. The transformations starting with exch correspond to exchanges between gray levels; e.g., exch(B,W) exchanges black and white. lrFlip and udFlip are left-right and up-down geometric flips, respectively, while rot90 and rot180 are geometric rotations by the respective number of degrees (clockwise).

Invariances in psychophysics recapitulate symmetries in natural images

The ‘variance is salience’ hypothesis can be further tested by asking whether symmetries of the natural distribution of textures are reflected in invariances of psychophysical thresholds. Binary texture coordinates (Hermundstad et al., 2014) are not affected by many of these symmetry transformations, and so a test requires textures containing at least three gray levels. For instance, reflecting a texture around the vertical axis has no effect on second-order statistics in the binary case, but it leads to a flip around the direction in the simple plane in ternary texture space (dashed green line in Figure 4C). To see this, recall that the coordinates in the plane are given by the probabilities that for the three possible values of (Figure 1). Under a left-right flip, the values and are exchanged, leading to . This means that the direction gets mapped to , the direction gets mapped to , and the direction remains unaffected. More details, and a generalization to additional symmetry transformations, can be found in Appendix 3. We find that the distribution of natural images is symmetric about the direction and is thus unaffected by this transformation, predicting that psychophysical thresholds should also be unaffected when textures are flipped about the vertical axis. This is indeed the case (Figure 6B). Similarly, the natural image distribution is symmetric under flips about the horizontal axis, and also under rotations by 90, 180, and 270 degrees, predicting perceptual invariances that are borne out in the psychophysical data (Figure 6B).

Reflecting a texture about the vertical axis also has an interesting effect on the plane: it not only flips the texture about the direction, but it also maps the texture onto the plane corresponding to the opposite diagonal orientation, . The fact that a flip about the direction is a symmetry of natural images is thus related to the fact that the diagonal pairwise correlations are the same regardless of the orientation of the diagonal. This fact was already observed in the binary analysis (Hermundstad et al., 2014), and is related to the invariance under 90-degree rotations observed here (Figure 6B).

It is important to note that these symmetries were not guaranteed to exist for either natural images or human psychophysics. Most of the textures that we are using are not themselves invariant under rotations (see the examples from Figure 1C,D). This means that invariances of predicted thresholds arise from symmetries in the overall shape of the distribution of natural textures. Similarly, had observed thresholds been unrelated to natural texture statistics, we could have found a discrepancy between the symmetries observed in natural images and those observed in human perception. As an example, the up and down directions differ in meaning, as do vertical and horizontal directions. A system that preserves these semantic differences would not be invariant under flips and rotations. The fact that the psychophysical thresholds are, in fact, invariant under precisely those transformations that leave the natural image distribution unchanged supports the idea that this is an adaptation to symmetries present in the natural visual world.

Natural images also have a well-known asymmetry between bright and dark contrasts (Ratliff et al., 2010; Tkacik et al., 2010) that is reflected in the anatomy and physiology of visual circuits, and in visual behavior (Ratliff et al., 2010; Tkacik et al., 2010; Zemon et al., 1988; Chubb et al., 2004; Jin et al., 2008; Komban et al., 2014; Kremkow et al., 2014). Our psychophysical data also show a bright/dark asymmetry. For instance, in Figure 4C, the threshold contour is not symmetric under the exchange of black and white checks, which has the effect of reflecting thresholds about the upper-left axis in the plane (dashed orange line in the figure). Such bright-dark asymmetries lead to a wide distribution of relative changes in detection threshold upon the exchange of black and white (red violin plot in Figure 6B for the exch(B,W) transformation). Our natural image analysis does not show this asymmetry (blue violin plot in Figure 6B for the exch(B,W) transformation) because of our preprocessing, which follows previous work (Hermundstad et al., 2014). As observed in Ratliff et al., 2010, the bright-dark asymmetry rests on two main characteristics of natural images: a skewed distribution of light intensities such that the mean intensity is larger than the median, and a power-law spectrum of spatial correlations. Both of these are reduced or removed by our preprocessing pipeline, which starts by taking the logarithm of intensity values and thereby reduces the skewness in the intensity distribution, and continues with a whitening stage that removes the overall spatial correlation spectrum seen in natural images. The final ternarization step additionally reduces any remaining dark-bright asymmetry, since we ensure that each of the three gray levels occurs in equal proportions in the preprocessed patches. This explains why we do not see this asymmetry in our natural-image analysis.

Discussion

The efficient coding hypothesis posits that sensory systems are adapted to maximize information about natural sensory stimuli. In this work, we provided a rigorous quantitative test of this hypothesis in the context of visual processing of textures in a regime dominated by sampling noise. To this end, we extended the study of binary texture perception to grayscale images that capture a broader range of correlations to which the brain could conceivably adapt. We first generalized the definition of textures based on local multi-point correlations to accommodate multiple luminance levels. We then constructed algorithms for generating these textures, and we used these in our behavioral studies. By separately analyzing the distribution of textures across an ensemble of natural images, we showed that psychophysical thresholds can be predicted in remarkable detail based on the statistics of natural scenes. By further exploiting symmetry transformations that have non-trivial effects on ternary (but not binary) texture statistics, we provided a novel test of efficient coding and therein demonstrated that visually-guided behavior shows the same invariances as the distribution of natural textures. Overall, this work strengthens and refines the idea that the brain is adapted to efficiently encode visual texture information.

The methodology developed here can be used to address many hypotheses about visual perception. For example, if a specific set of textures was hypothesized to be particularly ethologically relevant, this set could be measured and compared against ‘irrelevant’ textures of equal signal-to-noise ratio. Because our hypothesis treats every dimension of texture space equally—the symmetry only broken by properties of the natural environment—we leveraged the rapidly increasing dimensionality of grayscale texture space to more stringently test the efficient coding hypothesis. In this vein, our construction can be generalized to larger numbers of gray levels and correlations over greater distances. However, the ability of neural systems to adapt to such correlations must ultimately be limited because, as texture complexity grows, it will eventually become impossible for the brain to collect sufficient statistics to determine optimal sensitivities. Even were it possible to accumulate these statistics, adapting to them might not be worth the computational cost of detecting and processing long-range correlations between many intensity values. Understanding the limits of texture adaptation will teach us about the cost-benefit tradeoffs of efficient coding in sensory cortex, in analogy with recently identified cost-benefit tradeoffs in optimal inference (Tavoni et al., 2019). And indeed, although our predictions are in excellent agreement with the data in most cases, we find a few systematic differences that may already be giving us a glimpse of these limits.

Materials and methods

Code and data

The code and data used to generate all of the results in the paper can be found on GitHub (RRID:SCR_002630), at https://github.com/ttesileanu/TextureAnalysis (Tesileanu et al., 2020; copy archived at https://github.com/elifesciences-publications/TextureAnalysis).

Definition of texture space

A texture is defined here by the statistical properties of blocks of checks, each of which takes the value 0, 1, or 2, corresponding to the three luminance levels (black, gray, or white; see Appendix 1 for a generalization to more gray levels). The probabilities for all the possible configurations of such blocks form an overcomplete coordinate system because the statistical properties of textures are independent of position. To build a non-redundant parametrization of texture space, we use a construction based on a discrete Fourier transform (see Appendix 1). Starting with the luminance values , of checks in a texture block (arranged as in Figure 1A), we define the coordinates which are equal to the fraction of locations where the linear combination has remainder equal to after division by three (the number of gray levels). In the case of three gray levels, the coefficients can be +1, −1, or 0.

Each set of coefficients identifies a texture group, and within each texture group we have three probability values, one for each value of . Since the probabilities sum up to 1, each texture group can be represented as a plane, and more specifically, as a triangle in a plane, since the probabilities are also non-negative. This is the representation shown in Figure 1C,D and used in subsequent figures. For compactness of notation, when referring to the coefficients , we write + and − instead of +1 and −1, and omit coefficients that are 0, e.g., instead of . We also use (rather than the generic symbol ) for 1-point correlations, for 2-point correlations, for 3-point correlations, and for 4-point correlation, matching the notation used in the binary case (Victor and Conte, 2012; Hermundstad et al., 2014). For instance, is the plane identified by the linear combination .

Texture analysis and synthesis

Finding the location in texture space that matches the statistics of a given image patch is straightforward given the definition above: we simply glide a block over the image patch and count the fraction of locations where the combination takes each of its possible values modulo three, for each texture group identified by the coefficients . (As a technical detail, we glide the smallest shape that contains non-zero coefficients. For example, for , we glide a region instead of a one. This differs from gliding the block for all orders only through edge effects, and thus the difference decreases as the patch size increases.)

In order to generate a patch that corresponds to a given location in texture space, we use a maximum entropy construction that is an extension of the methods from Victor and Conte, 2012. There are efficient algorithms based on 2d Markov models that can generate all single-group textures, namely, textures in which only the probabilities within a single texture group deviate from (1/3, 1/3, 1/3). For textures involving several groups, the construction is more involved, and in fact some combinations of texture coordinates cannot be achieved in a real texture. This restriction applies as textures become progressively more ‘saturated’, that is, near the boundary of the space. In contrast, near the origin of the space all combinations can be achieved via the ‘donut’ construction for mixing textures, described in Victor and Conte, 2012. Details are provided in Appendix 2.

Visual stimulus design

The psychophysical task is adapted from Victor et al., 2005, and requires that the subject identify the location of a -check target within a -check array. The target is positioned near one of the four sides of the square array (chosen at random), with an 8-check margin. Target and background are distinguished by the texture used to color the checks: one is always the i.i.d. (unbiased) texture with three gray levels; the other is a texture specified by one or two of the coordinates defined in the text. In half of the trials, the target is structured and the background is i.i.d.; in the other half of the trials, the target is i.i.d. and the background is structured. To determine psychophysical sensitivity in a specific direction in the space of image statistics, we proceed as follows Hermundstad et al., 2014; Victor and Conte, 2012; Victor et al., 2013; Victor et al., 2015. We measure subject performance in this 4-alternative forced-choice task across a range of ‘texture contrasts’, that is, distances from the origin in the direction of interest. Fraction correct, as a function of texture contrast, is fit to a Weibull function, and threshold is taken as the texture contrast corresponding to a fraction correct of 0.625, that is, halfway between chance (0.25) and ceiling (1.0). Typically, 12 different directions in one plane of stimulus space are studied in a randomly interleaved fashion. Each of these 12 directions is sampled at 3 values of texture contrast, chosen in pilot experiments to yield performance between chance and ceiling. These trials are organized into 15 blocks of 288 trials each (a total of 4320 trials), so that each direction is sampled 360 times.

Visual stimulus display

Stimuli, as described above, were presented on a mean-gray background for 120 ms, followed by a mask consisting of an array of i.i.d. checks, each half the size of the stimulus checks. The display size was deg; viewing distance was 103 cm. Each of the array stimulus checks consisted of hardware pixels, and measured min. The display device was an LCD monitor with a refresh rate of 100 Hz, driven by a Cambridge Research ViSaGe system. The monitor was calibrated with a photometer prior to each day of data collection to ensure that the luminance of the gray checks was halfway between that of the black checks (<0.1) and white checks (23 cd/m2).

Psychophysics subjects

Subjects were normal volunteers (three male, three female), ages 20 to 57, with visual acuities, corrected if necessary, of 20/20 or better. Of the six subjects, MC is an experienced psychophysical observer with thousands of hours of experience; the other subjects (SR, NM, WC, ZA, JWB) had approximately 10 (JWB), 40 (NM, WC, ZA) or 100 (SR) hours of experience at the start of the study, as subjects in related experiments. MC is an author. NM, WC, and ZA were naïve to the purposes of the experiment.

This work was carried out with the subjects’ informed consent, and in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and the approval of the Institutional Review Board of Weill Cornell.

Psychophysics averaging

The average thresholds used in the main text were calculated by using the geometric mean of the subject thresholds, after applying a per-subject scaling factor chosen to best align the overall sensitivities of all the subjects; these multipliers ranged from 0.855 to 1.15. Rescaling a single consensus set of thresholds in this fashion accounted for 98.8% of the variance of individual thresholds across subjects. The average error bars were calculated by taking the root-mean-squared of the per-subject error bars in log space (determined from a bootstrap resampling of the Weibull-function fits, as in Victor et al., 2005), and then exponentiating.

Natural image preprocessing

Images were taken from the UPenn Natural Image Database (Tkačik et al., 2011) and preprocessed as shown in Figure 3A (see Materials and methods for details). Starting with a logarithmic encoding of luminance, we downsampled each image by averaging over blocks of pixels to reduce potential camera sampling artifacts. We then split the images into non-overlapping patches of size , filtered the patches to remove the average pairwise correlation expected in natural images (van Hateren, 1992), and finally ternarized patches to produce equal numbers of black, gray, and white checks (Figures 3A, C and D). For most figures shown in the main text, we used and . Each patch was then analyzed in terms of its texture content and mapped to a point in ternary texture space following the procedure described in the main text. Finally, to avoid biases due to blurring artifacts, we fit a two-Gaussian mixture model to the texture distribution and used this to separate in-focus from blurred patches (Figure 3B; Hermundstad et al., 2014; details in Materials and methods).

Whitening of natural image patches

To generate the whitening filter that we used to remove average pairwise correlations, we started with the same preprocessing steps as for the texture analysis, up to and including the splitting into non-overlapping patches (first three steps in Figure 3A). We then took the average over all the patches of the power spectrum, which was obtained by taking the magnitude-squared of the 2d Fourier transform. Taking the reciprocal square root of each value in the resulting matrix yielded the Fourier transform of the filtering matrix.

Removal of blurred patches in natural images

Following the procedure from Hermundstad et al., 2014, we fit a Gaussian mixture model with non-shared covariance matrices to the distribution of natural images in order to identify patches that are out of focus or motion blurred. This assigned each image patch to one of two multivariate Gaussian distributions. To identify which mixture component contained the sharper patches, we chose the component that had the higher median value of a measure of sharpness based on a Laplacian filter. Specifically, each patch was normalized so that its median luminance was set to 1, then convolved with the matrix . The sharpness of the patch was calculated as the median absolute value over the pixels of the convolution result. This analysis was performed before any of the preprocessing steps, including the transformation to log intensities, and was restricted to the pixels that did not border image edges. There was thus no need to make assumptions regarding pixel values outside images.

Efficient coding calculations

Threshold predictions from natural image statistics were obtained as in Hermundstad et al., 2014. We fit a multivariate Gaussian to the distribution of texture patches after removing the blurry component, and used the inverse of the standard deviation in the texture direction of interest as a prediction for the psychophysical threshold in that direction. The overall scale of the predictions is not fixed by this procedure. We chose the scaling so as to minimize the error between the measurements and the predictions , , where are the measurement uncertainties in log space. This scaling factor was our single fitting parameter.

Calculating mismatch

In Figure 6 we show several comparisons of two sets of thresholds and . These are either the set of measured thresholds and the set of natural image predictions for specific preprocessing options; or sets of either measured or predicted thresholds before and after the action of a symmetry transformation. We measure mismatch by the difference between the natural logarithms of the two quantities, , which is approximately equal to the relative error when the mismatches are not too large (). In panel A of the figure all 311 measured values and the corresponding predictions were used. For panel B, this set was restricted in two ways. First, for we ignored the measurements for which we did not have psychophysics data for the transformed direction. And second, we ignored directions on which the transformation acted trivially (see Appendix 3).

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "Sensitivity to grayscale textures is adapted to natural scene statistics" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Timothy Behrens as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this manuscript, Tesileanu and co-authors describe a new set of psychophysical experiments and analyses that ask how perceptual salience is related to natural scene variability, specifically looking at the statistics of pairs of ternary pixel values. They find strong concordance between perceptual threshold and the variability of statistics in natural scenes, so that lower detection thresholds exist for the more variable directions in their defined statistical space. This work is solid and follows nicely on their prior work published in eLife, making it a good Research Advance.

Essential revisions:

1) Can something be said about which directions showed the biggest differences between psychophysics and natural scenes? In the Discussion, it's stated that it's hard to be perfect and low variance dimensions are difficult. But is there any more specific significance to which patch patterns deviate most from this variance hypothesis?

2) Please expand on how to interpret these results in light of the whitening process, which removed the average pairwise correlations from the image patches. The idea is that deviations from the mean 2-point correlations result in saliency, but is it troubling that the psychophysical stimulus itself did not have the mean 2-point correlations that exist in natural scenes? How would the visual system be exquisitely sensitive to deviations in natural scenes, but still work identically even in the absence of naturalistic pairwise correlations in the psychophysical stimulus? It seems as though this absence of naturalistic pairwise correlations would mean that the observer's visual system is pretty far from where it might reside under realistic natural viewing conditions. Yet the variability still predicts salience. Is this notable or is this easily explained?

3) Enhance the clarity of the Results section:

The theoretical framework set up here is very nice, but challenging to grasp for a naive reader. Many details are left to the appendix that could be moved to the main text (as a summary) to improve the clarity of the presentation. On the other hand, there are some details in the Results that detract from the clarity of the presentation that could be moved to the Materials and methods. Please consider these points and revise the Results section accordingly.

Essential revisions:

1) Can something be said about which directions showed the biggest differences between psychophysics and natural scenes? In the Discussion, it's stated that it's hard to be perfect and low variance dimensions are difficult. But is there any more specific significance to which patch patterns deviate most from this variance hypothesis?

We thank the reviewers for raising this point, and pursuing it reveals two interesting trends. This analysis is detailed in a new Appendix and summarized in the text (see Appendix 5, and subsection “Variance predicts salience”).

First, predictions in simple planes tend to underestimate actual thresholds, while predictions in mixed planes tend to overestimate actual thresholds (effect size of about 9%). As we elaborate in the appendix, this suggests that the brain puts more resources into analyzing mixed-plane correlations than our efficient-coding model would predict.

A second observation is that prediction errors are, on average, almost three times larger in directions defined by modular sums (like 𝛽++) than those defined by modular differences (like 𝛽+-), namely, 19% vs. 7%. As explained in the new material, we suggest that this reflects a kind of non-robustness (e.g., sensitivity to discretization) associated with the modular-sum correlations that distinguishes them from the modular-difference correlations.

2) Please expand on how to interpret these results in light of the whitening process, which removed the average pairwise correlations from the image patches. The idea is that deviations from the mean 2-point correlations result in saliency, but is it troubling that the psychophysical stimulus itself did not have the mean 2-point correlations that exist in natural scenes? How would the visual system be exquisitely sensitive to deviations in natural scenes, but still work identically even in the absence of naturalistic pairwise correlations in the psychophysical stimulus? It seems as though this absence of naturalistic pairwise correlations would mean that the observer's visual system is pretty far from where it might reside under realistic natural viewing conditions. Yet the variability still predicts salience. Is this notable or is this easily explained?

During natural vision, a substantial contribution to whitening of the visual input comes from fixational eye movements (Rucci and Victor, 2015). Spatial filtering has also been thought to play a role (Atick and Redlich, 1990), but in vitro experiments (Simmons et al., 2013) found that adaptive spatiotemporal receptive field processing did not by itself whiten the retinal output, but rather served to maintain a similar degree of correlation across stimulus conditions. We infer from these studies that short visual stimuli like our 120ms presentations should be pre-whitened, to make up for the absence of fixational eye movements that produce whitening in natural, continuous viewing conditions. We have now commented on this in the subsection entitled “Psychophysical Measurements”.

3) Enhance the clarity of the Results section:

The theoretical framework set up here is very nice, but challenging to grasp for a naive reader. Many details are left to the appendix that could be moved to the main text (as a summary) to improve the clarity of the presentation. On the other hand, there are some details in the Results that detract from the clarity of the presentation that could be moved to the Materials and methods. Please consider these points and revise the Results section accordingly.

We made several changes to enhance clarity and streamline the main text:

– We outlined the procedure used for generating texture patches in the first subsection of the Results (subsection “Local textures with multiple gray levels”)

– We expanded the legend for Figure 2

– We moved the preprocessing details from the Results section to the Materials and methods (subsection “Local textures with multiple gray levels”)

– We updated the legend for Figure 5 to match the similar explanation from Figure 2’s legend

– We added an example of how we can infer the effects of a symmetry transformation on our threshold prediction (subsection “Invariances in psychophysics recapitulate symmetries in natural images”)

We also simplified the statistical analysis (not specifically requested, but we think this helps):

– We replaced the relative error / relative change measures that we had used in Figure 6 by log errors, for consistency with the rest of the paper

– We replaced the root-mean-square measure D that we were using for the statistical tests with a median, again for consistency and for more robustness to outliers (Appendix 4)

– We added summary statistics and 95% confidence intervals in addition to p-values in the Results section (subsection “Variance predicts salience”)

Appendix 4—table 1.

Results from statistical tests comparing the match between measured and predicted thresholds to chance.

The left column gives the preprocessing parameters (the downsampling factor) and (the patch size) in the format . For each of the permutation tests, a -value and the shortest interval containing 95% of the values obtained in 10,000 samples is given (the 95% highest-density interval, or HDI). Similarly, for the exponent estimation, we include the shortest interval containing 95% of the posterior density for each of the two model parameters (the 95% highest posterior-density interval, or HPDI).

		1. Permutation #1		2. Permutation #2		3. exponent estimation
(l)3-8			D		D	η	σ
N×R	D_actual	p	[95% HDI]	p	[95% HDI]	[95% HPDI]	[95% HPDI]
1×32	0.20	<10−4	[0.26,0.33]	0.0041	[0.21,0.27]	[0.58,0.75]	[0.22,0.26]
1×48	0.20	<10−4	[0.28,0.35]	0.0020	[0.22,0.29]	[0.52,0.67]	[0.22,0.25]
1×64	0.21	<10−4	[0.29,0.37]	0.0010	[0.23,0.31]	[0.50,0.63]	[0.21,0.25]
2×32	0.13	<10−4	[0.24,0.31]	<10−4	[0.16,0.22]	[0.81,0.98]	[0.18,0.22]
2×48	0.15	<10−4	[0.26,0.33]	<10−4	[0.18,0.23]	[0.71,0.85]	[0.18,0.21]
2×64	0.16	<10−4	[0.27,0.34]	<10−4	[0.18,0.24]	[0.66,0.79]	[0.18,0.22]
4×32	0.12	<10−4	[0.24,0.30]	<10−4	[0.15,0.20]	[0.87,1.04]	[0.18,0.22]
4×48	0.14	<10−4	[0.26,0.33]	0.0003	[0.16,0.22]	[0.74,0.89]	[0.18,0.21]
4×64	0.15	<10−4	[0.26,0.34]	0.0002	[0.16,0.23]	[0.69,0.83]	[0.18,0.21]

Appendix 6—table 1.

Results from statistical tests comparing the match between measured and predicted thresholds to chance when using the van Hateren natural image database (van Hateren and van der Schaaf, 1998).

The left column gives the preprocessing parameters (the downsampling factor) and (the patch size) in the format . For each of the permutation tests, a -value and the shortest interval containing 95% of the values obtained in 10,000 samples is given (the 95% highest-density interval, or HDI). Similarly, for the exponent estimation, we include the shortest interval containing 95% of the posterior density for each of the two model parameters (the 95% highest posterior-density interval, or HPDI).

		1. Permutation #1		2. Permutation #2		3. exponent estimation
(l)3-8			D		D	η	σ
N×R	D_actual	p	[95% HDI]	p	[95% HDI]	[95% HPDI]	[95% HPDI]
1×32	0.22	<10−4	[0.27,0.34]	0.0081	[0.23,0.31]	[0.45,0.63]	[0.24,0.28]
1×48	0.23	<10−4	[0.30,0.37]	0.0013	[0.26,0.34]	[0.41,0.57]	[0.24,0.28]
1×64	0.24	<10−4	[0.30,0.38]	0.0020	[0.26,0.34]	[0.40,0.54]	[0.24,0.27]
2×32	0.16	<10−4	[0.26,0.32]	0.0001	[0.19,0.25]	[0.67,0.84]	[0.21,0.24]
2×48	0.18	<10−4	[0.27,0.34]	0.0003	[0.20,0.27]	[0.58,0.73]	[0.21,0.24]
2×64	0.19	<10−4	[0.28,0.36]	0.0002	[0.21,0.28]	[0.54,0.67]	[0.21,0.24]
4×32	0.14	<10−4	[0.25,0.32]	0.0002	[0.17,0.24]	[0.73,0.90]	[0.20,0.23]
4×48	0.15	<10−4	[0.27,0.34]	<10−4	[0.18,0.26]	[0.64,0.78]	[0.19,0.23]
4×64	0.16	<10−4	[0.28,0.36]	<10−4	[0.20,0.26]	[0.59,0.73]	[0.19,0.23]