TYPE Original Research PUBLISHED 20 July 2023 DOI 10.3389/fnins.2023.1200661 A mechanistic account of visual OPEN ACCESS discomfort EDITED BY Christopher DiMattina, Florida Gulf Coast University, United States Olivier Penacchio *, Xavier Otazu 1,2, Arnold J. Wilkins 4 and 1,2,3 REVIEWED BY Sarah M. Haigh 5,6 Isamu Motoyoshi, 1 Department of Computer Science, Universitat Autònoma de Barcelona, Bellaterra, Spain, 2 Computer The University of Tokyo, Japan Vision Center, Universitat Autònoma de Barcelona, Bellaterra, Spain, 3 School of Psychology and Branka Spehar, Neuroscience, University of St Andrews, St Andrews, United Kingdom, 4 Department of Psychology, University of New South Wales, Australia University of Essex, Colchester, United Kingdom, 5 Department of Psychology, University of Nevada *CORRESPONDENCE Reno, Reno, NV, United States, 6 Institute for Neuroscience, University of Nevada Reno, Reno, NV, Olivier Penacchio United States op5@st-andrews.ac.uk RECEIVED 05 April 2023 ACCEPTED 27 June 2023 Much of the neural machinery of the early visual cortex, from the extraction PUBLISHED 20 July 2023 of local orientations to contextual modulations through lateral interactions, is CITATION thought to have developed to provide a sparse encoding of contour in natural Penacchio O, Otazu X, Wilkins AJ and scenes, allowing the brain to process efficiently most of the visual scenes we are Haigh SM (2023) A mechanistic account of exposed to. Certain visual stimuli, however, cause visual stress, a set of adverse visual discomfort. Front. Neurosci. 17:1200661. effects ranging from simple discomfort to migraine attacks, and epileptic seizures doi: 10.3389/fnins.2023.1200661 in the extreme, all phenomena linked with an excessive metabolic demand. The COPYRIGHT theory of efficient coding suggests a link between excessive metabolic demand © 2023 Penacchio, Otazu, Wilkins and Haigh. and images that deviate from natural statistics. Yet, the mechanisms linking energy This is an open-access article distributed under demand and image spatial content in discomfort remain elusive. Here, we used the terms of the Creative Commons Attribution License (CC BY). The use, distribution or theories of visual coding that link image spatial structure and brain activation to reproduction in other forums is permitted, characterize the response to images observers reported as uncomfortable in a provided the original author(s) and the biologically based neurodynamic model of the early visual cortex that included copyright owner(s) are credited and that the original publication in this journal is cited, in excitatory and inhibitory layers to implement contextual influences. We found accordance with accepted academic practice. three clear markers of aversive images: a larger overall activation in the model, a No use, distribution or reproduction is less sparse response, and a more unbalanced distribution of activity across spatial permitted which does not comply with these terms. orientations. When the ratio of excitation over inhibition was increased in the model, a phenomenon hypothesised to underlie interindividual differences in susceptibility to visual discomfort, the three markers of discomfort progressively shifted toward values typical of the response to uncomfortable stimuli. Overall, these findings propose a unifying mechanistic explanation for why there are differences between images and between observers, suggesting how visual input and idiosyncratic hyperexcitability give rise to abnormal brain responses that result in visual stress. KEYWORDS visual discomfort, visual stress, interindividual differences, natural scenes, efficient coding, hypermetabolism, urban scenes, computational modelling Introduction Some static images, particularly stripes, are consistently reported as aversive, causing headaches, eyestrain, perceptual distortions, hallucinatory colours or shapes, a series of bodily symptoms regrouped under the term visual stress (Wilkins, 1995). Stripes can also trigger seizures in individuals with photosensitive epilepsy (Wilkins et al., 1984; Radhakrishnan et al., 2005; Hermes et al., 2017). The current literature considers visual stress as resulting from unusually strong neural activity in the visual cortex (Haigh et al., 2013). Visual stress occurs in Frontiers in Neuroscience 01 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 a range of neurological conditions that are co-morbid with epilepsy images and to sine gratings, a class of images associated with visual (Wilkins et al., 2022), consistent with the theory that cortical discomfort. They found that the response of the model to natural hyperexcitability underlies visual stress. In migraine, hyperexcitability images was sparser than that to gratings, involving fewer “neurons” is also suggested by a large haemodynamic response to unpleasant and therefore consistent with the excessive metabolism associated sensory stimuli, a low phosphene threshold in response to transcranial with uncomfortable images. The approach, however, had three magnetic stimulation, and the use of antiepileptics in migraine limitations. (1) The study considered only two extreme classes of prophylaxis (Porciatti et al., 2000; Huang et al., 2003; Ferrari et al., stimuli, natural images, and gratings, and not more complex real-life 2015; Mulleners et al., 2015). In the non-clinical population, images scenes. (2) The model did not include contextual influences, i.e., the that trigger visual stress are generally associated with a large cortical way the activity of a neuron is modulated by the activity of haemodynamic response, as measured using functional magnetic neighbouring neurons (Fitzpatrick, 2000; Carandini et al., 2005). resonance imaging (Huang et al., 2003, 2011) and near infra-red Contextual modulations in the early visual system are fundamental to spectroscopy (Haigh et al., 2013, 2015; Le et al., 2017), or a large understand how the strength of activity of a neuron can be deeply electrophysiological response, for example, steady-state evoked modified depending on the spatial structure of the input outside its potentials (O'Hare, 2016; Haigh et al., 2019; Gentile and Aguirre, 2020; classical receptive field (Zipser et al., 1996; Vinje and Gallant, 2000). Lindquist et al., 2021) and event-related potentials (Haigh et al., 2019). (3) The model proposed in (Hibbard and O'Hare, 2015) could not To date, no general principle has explained the link between the account for differences between individuals, for example by spatial structure of an image and the visual stress associated with it, considering individual differences in the balance between cortical but the most satisfying account for the relationship between visual excitation and inhibition, a possible reason for increased susceptibility stress and enhanced cortical activity comes from a prevalent theory in to discomfort in some individuals (Adjamian et al., 2004). sensory neuroscience, that of efficient coding (Attneave, 1954; Barlow, In this study, we systematically analyzed the activity of a 1961). The theory of efficient coding states that sensory systems have biologically based neurodynamic model of the early visual cortex that evolved under the selective pressure to provide an efficient implements contextual influences in response to images from urban representation of the stimuli relevant for survival in an organism’s landscapes and abstract art. We considered theories of visual coding environment. The efficiency is expressed both in terms of the that link image spatial structure and brain activation in order to information transmitted and the metabolic cost of the neural processes investigate how different metrics of the model activity predicted involved. The theory has received strong empirical and theoretical observers’ self-reported visual discomfort. Our aim was twofold: (1) support for different sensory modalities (Olshausen and Field, 1996; To gain a mechanistic understanding of the differences in induced Lewicki, 2002; Machens et al., 2005). A large body of theoretical and discomfort between images, and (2) to investigate possible mechanistic empirical evidence has demonstrated an accord between neural explanations for individual differences in susceptibility to discomfort. encoding and regularities in the relevant natural environment (e.g., We first identified possible neural markers of discomfort. We next Olshausen and Field, 1996; Lewicki, 2002; Machens et al., 2005). An analyzed how these markers changed when the ratio of excitation over important regularity concerns how the correlation between the inhibition in the network was modified. This allowed us to test the luminance of two nearby locations in a natural scene decreases with putative role of disparities in balance between cortical excitation and distance. It is encapsulated in the Fourier power spectrum of natural inhibition in the mechanisms underlying inter-individual differences scenes, which decreases with spatial frequency f as 1 / f α , with an in susceptibility to visual discomfort. exponent α consistently found between 1.6 and 2.4 (Field, 1987; Tolhurst et al., 1992; Geisler, 2008). In agreement with the theory of efficient coding, discrimination performance is optimal for stimuli Materials and methods with such statistics (Knill et al., 1990; Parraga et al., 2000; Geisler et al., 2001). Similarly, measuring the deviation from 1 / f α provides a Stimuli, participants, and procedure reliable predictor of the visual discomfort a visual stimulus provokes (Fernandez and Wilkins, 2008; Juricevic et al., 2010; O'Hare and We considered four sets of stimuli previously used in the literature, Hibbard, 2011; Penacchio and Wilkins, 2015; Ogawa and Motoyoshi, two sets of photographs of urban architecture (building frontages) and 2020), and the way in which the chromaticity differences in a stimulus two sets of images of abstract art. depart from those in nature predicts visual discomfort (Penacchio et al., 2021). Architecture stimuli Whilst the theory of efficient coding provides a theoretical We used two sets, hereafter, Architecture 1 and 2, N = 74 images foundation for the understanding of visual stress, measuring in each, previously considered in a study on the contribution of spatial deviations with respect to the statistical regularities of natural scenes frequencies to visual discomfort for which the experiments were held can only provide functional models (Carandini and Heeger, 2012; Kay, in person (Penacchio and Wilkins, 2015). Participants [N = 10 in each 2018) of visual stress, namely models that describe a functional set, see (Penacchio and Wilkins, 2015)] viewed each colour 720 × relationship between input, here static images, and output, here 960-pixel images for an unlimited time on a calibrated LCD display observers’ self-reported discomfort when viewing the images. with a size of 20 cm x 27 cm (width x height) at 80 cm, representing a Understanding mechanistically how the spatial structures of static visual angle of 15°. They were instructed to rate the images for images cause discomfort therefore remains a major challenge. In a first discomfort on a Likert scale, from 1 for ‘not uncomfortable at all’ to 7 computational attempt to characterize the neural correlate of visual for ‘very uncomfortable’. We explicitly stated that we were interested discomfort using a mechanistic model (Hibbard and O'Hare, 2015), in discomfort, which we assumed to be distinct from unpleasantness, computed the activity of a bank of simple cells in response to natural ugliness, or un-preference. Frontiers in Neuroscience 02 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 Art stimuli The units in Component 1 were modelled using Gabor filters We considered two sets of images of abstract art from a study on the sensitive to four regularly spaced orientations (0°, 45°, 90° and 135°) contribution of variations in chromaticity to visual discomfort (Penacchio and 12 spatial frequencies, corresponding to 12 regularly spaced et al., 2021), Art 1 and 2, N = 50 images in each. For these sets, the wavelengths, ranging from 1 cycle per degree (cpd) to 10.8 cpd for the responses were collected online using Qualtrics, in agreement with experimental conditions of Architecture 1 and 2. The output of the COVID-19 protocols. Participants (Art 1, N = 53 after 5 were discarded convolution of these filters with the images was the input to the because they showed no variability in their responses, 37 female, 2 network (Component 2). non-binary, 14 male, mean age 20.0 years, SD age 2.0; Art 2, N = 79 after 1 The cortical columns (hypercolumns) in Component 2 densely was discarded on the same grounds as for Art 1, 59 female, 6 non-binary, sampled the input image (one hypercolumn per pixel, resulting in a 14 male, mean age 20 years, SD age 2.5) viewed each grey-level 512×512 total of 2562 hypercolumns) (Figure 1A). The population of pixel image for an unlimited duration and had to report their level of excitatory units within each hypercolumn included all 48 types of discomfort on a Likert scale (1, ‘no discomfort’, 2, ‘some discomfort’, 3, simple cells (4 orientations x 12 spatial scales). The hypercolumns ‘moderate discomfort’, 4, ‘uncomfortable, and 5, ‘very uncomfortable’). As also included a ‘mirror’ population of inhibitory units. Excitatory these two sets were rated online, we had no control on the viewing cells were connected to each other through monosynaptic excitatory conditions, and therefore no knowledge of the visual angle represented by connections, and disynaptically via inhibitory interneurons the stimuli for each participant. All participants were recruited from the [conforming to Dale’s law stating that neurons in the cortex are University of Nevada, Reno, and consented to the study electronically. either excitatory or inhibitory (Eccles, 1976)]. The architecture and They obtained a course credit for their time. All verified that they had connectivity of the network has been presented in previous work (Li, normal or corrected to normal vision and none reported a diagnosed 1999, 2002; Penacchio et al., 2013) and mimics the biological psychiatric or neurological condition. The protocol was approved by the architecture responsible for contextual modulations of the activity Institutional Review Board at the University of Nevada, Reno (333057), of a neuron by stimulation of its non-classical receptive field (nCRF). and was conducted in accordance with the Declaration of Helsinki. It is based on lateral connections that simulate the connectivity in Supplementary File “Raw data” (Penacchio et al., 2023) provides mammal and primate early visual cortex (Knierim and Vanessen, a visualisation of all the ratings of all observers for the four sets of 1992; Kapadia et al., 1995; Weliky et al., 1995). In particular, the stimuli. While the sets of urban scenes rated for discomfort consisted connectivity is set up such that: (i) mutual monosynaptic excitation of colour images, these images were modestly coloured and colour is strong between neighbouring units sensitive to similar spatial differences did not explain variance in discomfort (Penacchio et al., frequencies and to orientations similar to the direction formed by 2021; see Electronic Supplementary Material). these units [a cortical feature at the basis of contour enhancement (Li, 1999)], and, (ii) inhibition is strong between neighbouring units sensitive to orientations perpendicular to the orientation these units Experiments form [a property thought to be at the basis of iso-orientation suppression (Li, 1999)]. Overall, the activity of a unit is driven by the Our study consisted of two experiments. In Experiment 1, input within its CRF and modulated by contextual influence in its we looked for possible markers of visual discomfort in the activity of nCRF. Modulation can be suppressive or facilitatory, depending of a dynamic model of the visual system. Therefore, we contrasted the strength of horizontal recruitment, but tends to be more participants’ self-reported discomfort when viewing an image to suppressive for higher input levels (see Li, 1999) in agreement with different metrics of the activity of a model of the early visual system empirical observations (Hirsch and Gilbert, 1991; Weliky et al., in response to this image (see Figure 1). We found such markers, and 1995; Cavanaugh et al., 2002). so in Experiment 2 we explored whether and how they would change The architecture of Component 2 has been previously shown to when the ratio of excitation over inhibition was altered in the model. be able to account for figure-ground segmentation, bottom-up saliency, Since we could not measure intra-cortical inhibition, Experiment 2 contour grouping, and brightness induction (Zhaoping and May, 2007; was entirely computational. Zhang et al., 2012; Penacchio et al., 2013; Zhaoping and Zhe, 2015; Berga and Otazu, 2020, 2022). We also verified that simulating the nCRF of units with naturalistic stimuli increased the sparseness of their response Model of the early visual cortex (Vinje and Gallant, 2000; Haider et al., 2010; Supplementary Material S2; Penacchio et al., 2023). Model architecture Importantly, as our main purpose in this paper was to understand The model was made of two components. Component 1 consisted whether a biologically realistic model of the early visual cortex processes of units whose receptive fields (RFs) mimic those of simple cells in the uncomfortable images differently from other images, we did not fit any primary visual cortex (V1) and have been shown to fit well with parameters in any of the two components of the model but used the physiological data (Serre and Riesenhuber, 2004; Serre et al., 2007). previously published parameters [Component 1, (Serre and Riesenhuber, Component 2 was a network that takes the output of Component 1 2004, Serre et al., 2007); Component 2 (Li, 1999; Penacchio et al., 2013)]. and implements contextual influences. It consisted of a firing-rate Supplementary Material S1 gives a full description of the two components neurodynamic network made of units that emulate (‘glutamatergic’) of the model including the equations that govern the dynamic of the excitatory and (‘GABAergic’) inhibitory neurons connected as a network. Note that our choice for the size of the input images (256 × 256) recurrent network that emulates the connectivity of the visual cortex imposed an upper limit to the spatial frequencies handled by the model underpinning contextual influences (Wilson and Cowan, 1972; Li, (the highest frequency was 10.8 cpd). This choice reflected both a 1999, 2002; Dayan and Abbott, 2001; Figure 1A). constraint on the computational load for processing all the images with Frontiers in Neuroscience 03 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 FIGURE 1 Schematics of the experiment. (A) We processed images from four sets of stimuli (Architecture 1 and Architecture 2, N=74 each, Art 1 and Art 2, N=50 each) using a neurodynamic model of the early visual cortex. This model consisted of a layer of excitatory units and a layer of inhibitory units scattered in a grid of hypercolumns organised retinotopically, each including units sensitive to luminance edges with different spatial orientations and spatial frequencies. The hypercolumns were interconnected through excitatory-excitatory, excitatory-inhibitory, and inhibitory-inhibitory connections following a biologically plausible pattern of lateral connections. For each image, we recorded the firing rates of all the excitatory units in the model over several temporal iterations of the model, leading to (B) the model population response to the image, i.e., vectors of non-negative numbers. (C) Observers reported perceived discomfort when viewing each image in one of the four sets of stimuli by rating each stimulus on a Likert scale in which the lowest value meant ‘not uncomfortable at all’ and the highest ‘very uncomfortable to look at’. (D) We then regressed (here, for illustration, group average of) reported visual discomfort against different metrics of the model population response. The metrics were chosen to reflect three main hypotheses on the neural correlate of visual discomfort (see section Metrics of model activity, Rationale for types of metrics considered). (E) To analyse the contribution to visual discomfort of different spatial frequencies, we also regressed reported visual discomfort against the same metrics applied to the subset of units in the model sensitive to a given spatial frequency (‘frequency channels’, see Methods). The subpanels only show four of the twelve channels. all the possible values of gain (see below) and robust evidence from the 2019) rgb2gray function (Art 1 and 2), and finally down-sampled to literature that the frequencies that contribute most to discomfort are 256 × 256 pixel images using Matlab’s function ‘imresize’ with a nearest below 10 cpd (Wilkins et al., 1984; O'Hare and Hibbard, 2011). neighbour algorithm. To avoid edge effects, the images were padded using a standard mirroring process of width 28 pixels, resulting in 312 Model dynamic and output × 312-pixel images input to the model. Only the 256 × 256 central The dynamic of the model was simulated using a discrete time positions were considered for the model population response ( xisθ ( t ) ). implementation with an Euler integration scheme of the first order (Strogatz, 2015). The model assumes a periodic activity (‘steady state’) Excitation/inhibition balance after 3 to 4 membrane time constants after stimulus onset (Li, 1999; We manipulated the ratio of excitation over inhibition (E/I) in the Penacchio et al., 2013). For each image, the output of the model, model (e.g., modelling a lack of Gamma-aminobutyric acid (GABA) hereafter, model population response, was the vector ( xisθ ( t ) ) formed neurotransmitter) by decreasing the strength of activation of the by the firing rates of all the excitatory cells between the 4th and 20th inhibitory layer of the starting model (hereafter, reference model) using membrane time constants, where i ∈ [1, 256] is the hypercolumn 2 a multiplicative factor called gain. The gain varied between 1 (no position, s ∈ [112 , ] is the frequency channel, θ ∈ {0°,45°,90°,135°} is modification, reference model) to 0 (no inhibition at all) by steps of the orientation, and t ∈ {4,5,…,20} is the membrane time constant, −0.125. See Supplementary Material S1 for details. giving a vector of length ≈ 5.35 × 107 for each image (Figure 1B). All metrics to analyse the activity of the model were based on this vector. Rationale for using a neurodynamic model Input pre-processing Before being processed by the model, the images were cropped to We believe that many static models of early spatial vision that centred squares of maximal size, then converted to luminance using include units sensitive to different spatial frequencies and the measurement of the display’s R, G and B channels [Architecture 1 orientations, non-linearity and divisive normalization to and 2, see Penacchio and Wilkins (2015)], or Matlab’s (MATLAB, implement contrast gain-control (Heeger, 1992; Foley, 1994; Frontiers in Neuroscience 04 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 Cavanaugh et al., 2002; Kay et al., 2013; Schutt and Wichmann, activation level using alternative metrics (see 2017) would also be good model candidate to predict visual Supplementary Material S3; Penacchio et al., 2023). (2) The sparseness discomfort. We chose to use instead an excitatory-inhibitory of the model population response was measured as neurodynamic model based on Zhaoping Li’s seminal model (Li, ( ) S ( xisθ ( t ) ) = −∑ tanh xisθ ( t ) i ,θ , s ,t 2 (Hyvarinen, Hurri et al., 1999, 2002) for several reasons. First, this allowed us to have separate excitatory and inhibitory populations, and therefore to 2009). We also computed alternative measures of sparseness (see test putative hypotheses on the role played by the relative strength Supplementary Material S3; Penacchio et al., 2023). (3) Isotropy was of excitation and inhibition in visual discomfort, thought to play computed using the (Shannon) entropy H ist of the distribution of a central role in individual differences in susceptibility to visual response for each hypercolumn i, spatial frequency s and membrane discomfort. Second, this allowed us to inherit from (Li, 1999, ( ) time t , and taking the grand mean, H ( xisθ ( t ) ) = ∑ i , s ,t H ist / N , 2002) a realistic implementation of mutual influences between where N = 2562 × 12 × 16 is the number of terms in the sum, neighbouring units tuned to different spatial orientations. Third, and H ist = −∑θ = 0°, 45°,90°,135° pθ ( i,s,t ) log 2 ( pθ ( i,s,t ) ) is the including timing goes a step further toward more complex models entropy of the probability distribution obtained by normalising that in the future will allow a careful estimation of the amount of the vector (xis 0° ( t ) , xis 45° ( t ) , xis 90° ( t ) , xis135° ( t )) so that it sums activity in the gamma band, which are crucial to understand to one. Entropy measures isotropy in the sense that if all photosensitive epilepsy (e.g., Hermes et al., 2017 and the orientations are equally represented at ( i,s,t ), (i.e., see Discussion). xis 0° ( t ) = xis 45° ( t ) = xis 90° ( t ) = xis135° ( t ) , H ist assumes its maximal value, H ist = log 2 ( 4 ) = 2 . When the response to one orientation is greater than that for the others we have H ist < 2, and H ist = 0 if only Metrics of model activity a single orientation is represented (as for ‘stripes’). For analysing the frequency channels separately, we applied the same metrics to the Rationale for types of metrics considered subpopulation of units corresponding to each channel, i.e., ( xis0θ ( t ) ) Two main theories of visual coding link brain activation and with s0 being one of the 12 spatial frequencies (Figure 1E). Figure 2 image structure. The contrast energy theory states that activation illustrates the computations of the metrics and shows the stimulus in the primary cortex depends on local contrast at different with the minimum and maximum value of each metric in set spatial frequencies, independently of the spatial arrangement of Architecture 1 as well as the relevant aspect of the network activity the luminance edges producing this contrast (Movshon et al., [see Supplementary File “all markers’ values” (Penacchio et al., 2023) 1978; Albrecht and Hamilton, 1982; Rieger et al., 2013). The for an illustration of the markers values for all the stimuli in theory of sparse coding, by contrast, supposes that the early Architecture 1]. visual system has adapted to process the specific arrangements and redundancy of local contrast features found in natural scenes with a reduced number of neurons strongly active simultaneously Natural images (Atick and Redlich, 1992; Field, 1994; Olshausen and Field, 1996, 2004; Vinje and Gallant, 2002; Hyvarinen et al., 2009), saving To estimate the distribution of isotropy of the model response for metabolic energy. We therefore considered metrics of the model natural images we randomly selected 100 images of the van Hateren’s population activity in response to each image ( xisθ ( t ) ) related to database of calibrated natural images (van Hateren and van der Schaaf, (1) the model population activation level (‘total amount of firing 1998), processed them as done with the stimuli of the experiments and of all units when processing the image’) and (2) the sparseness of extracted this metric for each input image. the model response (‘to what extent the encoding of the image was carried out by a small number of active units across retinotopic space and time’). We also considered a metric based on (3) the Statistical analysis isotropy of the model response (‘how much activity was evenly distributed across orientations in the model hypercolumns’). Our We used linear mixed effects (multilevel) models to analyse rationale was that periodic patterns such as stripes, causing a possible correlations between metrics and participants’ self- concentration of neural activity in a subpopulation of cortical reported discomfort in Experiment 1. In all models, the metrics cells sensitive to the same orientation, are strong inducers of were considered as fixed effects and participant identity was pattern-sensitive epilepsy and visual discomfort (Wilkins et al., considered as a random effect (i.e., with random intercept, and, 1979, 1984; Meldrum and Wilkins, 1984; O'Hare and Hibbard, when possible, random slope). This allowed us to estimate the 2011; Hermes et al., 2017). effect of possible markers of discomfort in the general population while ignoring the specificities of each participant, including Metrics implementation specific viewing conditions and apparatus for the sets rated (1) The model activation level, similar to a measure of contrast online, Art 1 and 2. The models were fitted in R (R Development energy, was measured using the L1-norm of the model population Core Team, 2020) using the function lmer from the package lme4 response, E (= ( xisθ (t )))  = ( xisθ (t ))  1 ∑i,θ ,s,t xisθ (t ) , where (Bates et al., 2014). All the metrics were normalised to a mean of ( i,θ ,s,t ) runs over all the hypercolumns, orientations, frequencies 0 and a SD of 0.5 (e.g., Gelman and Hill, 2006). We selected the and membrane time constants considered. As the choice of a models using log likelihood, AIC information criterion, and particular norm over others is arbitrary, we also measured the likelihood ratio (BIC is reported for information). For hypothesis Frontiers in Neuroscience 05 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 FIGURE 2 Illustration of the metrics used as markers of visual discomfort. (A) Activation level of model population response: (from left to right, first panel) Image in the set Architecture 1 with the lowest average activation level, (second) corresponding heatmap of activation summed over all membrane times, frequency channels and orientations, (third) image with the highest average activation level, and (fourth) corresponding heatmap of activation summed as in the second panel; in the heatmaps, the yellower the colour the higher activation. (B) Sparseness of model population response: (first panel) Image with the highest value for the sparseness metric in the same set as in panel (A), (second, black curve) corresponding histogram of firing rates and (grey curve) histogram for the image in the third panel for comparison, (third) image with the lowest value for the sparseness metric, and (fourth, black curve) corresponding histogram of firing rates. (C) Isotropy metric: (first panel) Image with the highest level of the metric in the same set, (second) heatmap of isotropy averaged across all membrane time and frequency channels and (inset) example of distribution of responses across orientations with isotropy equal to the average for the whole image (1.708), (third) Image with the lowest level of the metric, and (fourth) heatmap of isotropy averaged as in panel two and (inset) example of distribution with isotropy equal to the average of the whole image (0.696). testing we used χ2-distributions whose degree of freedom were were associated with more discomfort (Figure 3A, full model, the differences in degrees of freedom between the models to Architecture 1: χ2 = 107, df = 3, p < 10−15; Architecture 2: χ2 = 157, df = 3, be compared. Following recommended practice (Meteyard and p < 10−15; Art 1: χ2 = 99, df = 3, p < 10−15; Art 2: χ2 = 78, df = 3, p < 10−15). Davies, 2020), all the details about the models tested and adopted We also found a relationship between the sparseness of encoding of are provided in Supplementary Material, section an image and its associated discomfort. Images inducing a less sparse Statistical Inference. (‘denser’) encoding caused more discomfort (Figure 3B, Architecture 1: χ2 = 72, df = 3, p < 10−14; Architecture 2: χ2 = 134, df = 3, p < 10−15; Art 1: χ2 = 77, df = 3, p < 10−15; Art 2: χ2 = 43, df = 3, p < 10−8). Anisotropy of Results the model response was clearly associated with discomfort for the two sets showing urban landscapes (Figure 3C, Architecture 1: χ2 = 72, Experiment 1: markers of visual discomfort df = 3, p < 10−14; Architecture 2: χ2 = 122, df = 3, p < 10−15). However, that was not the case for the first set of abstract art (Figure 2C, Art 1: The three types of metrics envisioned, activation level, sparseness, p = 0.35), and the relationship was inverted for the second set of and isotropy, were markers of visual discomfort. We found an effect abstract art, in which images encoded with a more isotropic activity of population activation level on observers’ judgements for the four were associated with more discomfort (Figure 3C, Art 2: χ2 = 169, sets of images. Images that elicited a greater activation in the model df = 3, p < 10−15). To investigate the low predictive power of isotropy Frontiers in Neuroscience 06 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 FIGURE 3 Correlations between average reported visual discomfort against the three main metrics of model population activity, namely (A) model population response activation level, (B) sparseness of the model response, and (C) isotropy in the model response for the four sets of stimuli (from left to right column, Architecture 1, N=74, Architecture 2, N=74, Art 1, N=50, and Art 2, N=50). Each point represents the Pearson’s correlation coefficient for one frequency channel of the model (dots, for the 12 frequency channels of the model) or for the whole model (triangle). The value of p of each regression is colour-coded with a level of blue (the darker, the lower the value of p), or with yellow for p-values above the reference threshold 0.05. Each inset shows the raw data for the regression in the case of the whole population (triangle). All metrics were normalised to a mean value of 0 and standard deviation of 0.5 (see Methods). for the two sets of art, we compared the distributions of this metric for in natural images, values which were associated with discomfort in the Art 1 and 2 to that found in natural images. We found that these sets of urban scenes Architecture 1 and 2. distributions greatly overlapped those for natural images, as illustrated Table 1 shows the Pearson correlation coefficients of the in Figures 4C,D, left panels (Jensen-Shannon divergence with natural regressions between metrics and discomfort ratings in Figure 3 (see images, 0.265 for Art 1, 0.233 for Art 2), in contrast with the wider inset in each panel). distribution and bigger separation with respect to natural scenes in the As previous research has shown that the discomfort an image case or the architectural facades (Figures 4A,B, Jensen-Shannon elicits strongly depends on its frequency content (Wilkins et al., 1984; divergence, 0.855 for Architecture 1, 0.810 for Architecture 2). O'Hare and Hibbard, 2011; Hermes et al., 2017), we also considered Accordingly, only a small proportion of the stimuli in Art 1 and 2 had the frequency channels in the model separately (Figure 3, “frequency isotropy values that fell outside of the distribution of isotropy found channels” in all panels). For the two sets of architecture, we found Frontiers in Neuroscience 07 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 FIGURE 4 Comparison of the distribution of isotropy in each set with that in nature and regression of isotropy against average reported discomfort for (A) the set Architecture 1, (B) Architecture 2, (C) Art 1, and (D) Art 2. In each panel, the left plot shows the distribution of isotropy in the model for all stimuli in the set (grey, left of the central axis) and distribution for a set of natural images (green, right of central axis). The dots show the raw values of the metric for each stimulus (Architecture 1, N=74, Architecture 2, N=74, Art 1, N=50, Art 2, N=50, and natural images, N=100), the box plot show the first (bottom) and third (top) quartile, the notches show the 95% confidence interval for the median, and the star and circle show the mean of the distributions. The right plots in each panel show the regression of isotropy against average reported discomfort (grey) as well as the mean value of the metric for the set of natural images. TABLE 1 Pearson correlation coefficients between the three types of metrics and average reported discomfort for the four sets of stimuli. Architecture 1 Architecture 2 Art 1 Art2 Activation rE = 0.57, p < 10−7 rE = 0.65, p < 10−9 rE = 0.38, p = 0.0065 rE = 0.31, p = 0.029 ci = [0.40, 0.71] ci = [0.49, 0.76] ci = [0.11, 0.60] ci = [0.03, 0.54] Sparseness rS = − 0.56, p < 10−6 rS = − 0.61, p < 10−8 rS = − 0.37, p = 0.0086 rS = −0.22, ns, p = 0.12; ci = [−0.70, −0.38] ci = [−0.73, −0.44] ci = [−0.59, −0.10] ci = [−0.47, 0.06] Isotropy rH = − 0.53, p < 10 −6 rH = − 0.57, p < 10 −7 rH = −0.13, ns, p = 0.38 rH = 0.51, p = 0.00017 ci = [−0.68, −0rE.35] ci = [−0.71, −0.40] ci = [−0.39, 0.16] ci = [0.27, 0.69] The values in bold correspond to the correlations that are both significant (under the 0.05 value) and for which the effect’s polarity agrees with the hypothesis of discomfort associated with an overload of activity. that visual discomfort was best predicted by the three metrics applied The three metrics were correlated to each other within each set to the channels tuned to spatial frequencies within the range of stimuli (see Supplementary Figures S2–S5, 1.5–6 cpd, as shown by the peaks of correlation around these Supplementary Material S5.1 and Figures S6–S9). However, for the frequencies in Figures 3A–C for Architecture 1 and 2 (see sets of architectural images the best linear combination of the three Supplementary Material S1 for a derivation of the correspondence metrics was a better predictor of visual discomfort than any of the between frequency channels in the model and visual angle in the metrics taken separately (Supplementary Material S5), leading to experimental conditions for Architecture 1 and 2). These frequencies Pearson correlation coefficients of rESH = 0.63 (p < 10−8; ci = [0.47, are those at which human sensitivity is maximal (Campbell and 0.75]) for Architecture 1, rESH = 0.67 (p < 10−10; ci = [0.52, 0.78]) for Robson, 1968). By contrast, for the images of art the strength of the Architecture 2, explaining, respectively, 40 and 45% of the variance correlations between discomfort and the metrics were higher for the in judgements for discomfort. Considering the three metrics channels tuned to the highest frequencies (Figures 3A–C, Art 1 and together did not provide a better predictor for the sets Art 1 and 2 2). We believe this difference to reflect the variability in spatial when the metrics were computed from the full model activity. frequencies caused by the uncontrolled nature of the online format However, for the frequency channel tuned to the highest spatial in Art 1 and 2. frequencies (Figure 3), a linear combination of the metrics predicted Frontiers in Neuroscience 08 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 better observers’ ratings than any of the metrics in isolation (see model. To establish this threshold, for each of the three metrics, Supplementary Material S5.2), with Pearson correlation coefficients we used the linear regression of discomfort judgement against the of rESH = 0.39 (p = 0.0047; ci = [0.13, 0.61]) for Art 1, rESH = 0.52 metric using best linear mixed models from Experiment 1 to find the (p < 10−3; ci = [0.28, 0.70]) for Art 2. Finally, whilst these two metrics metric value that corresponded to the 15% higher percentile of are generally correlated, considering activation alongside discomfort ratings. We next computed the number of images beyond sparseness, or vice versa, strongly increased model fit for this threshold for each value of the gain. We found a systematic Architecture 1 and Art 2, suggesting a certain degree of increase in the number of images beyond this threshold for decreasing independence between them (see Supplementary Material S5.3 and values of gain (see Supplementary Figure S14). theoretical considerations therein). To illustrate how the distribution of firing within hypercolumns changed as the strength of inhibition was decreased, we followed the distribution of the firing activity across orientation at each retinotopic Experiment 2: impact of excitation/ location (averaged over all membrane times and frequency channels) inhibition balance on the markers of when the gain of inhibition was decreased. For each image, a “winner- discomfort takes-all” process took place in the model’s hypercolumns in which all the activity concentrated in a single orientation for low values of When inhibition was decreased in the model, the markers of inhibition, as if only a single orientation was present in the input discomfort identified in Experiment 1 systematically shifted toward stimulus (see Supplementary Figure S15). In other words, as the values associated with increased discomfort. The shift was toward strength of inhibition was lowered the activity of the model became higher activation level (Architecture 1, used for the sake of illustration, akin to the activity in response to especially uncomfortable stimuli Figure 5A), a less sparse encoding (Figure 5B), and a decreased such as stripes. isotropy in the model response (Figure 5C). These differences with the reference model (gain = 1, top right distribution) reached significance for all metrics (the highest gain leading to a different distribution was Discussion gain = 0.375 for the activation level, Kolmogorov–Smirnov test, D = 0.297, p = 0.0027, and for sparseness, D = 0.378, value of p <10−4, In this study, we identified possible markers of visual discomfort and gain = 0.875 for D = 0.351, p = 0.00019), showing that the effect of in the response of a biologically plausible model of the early visual reducing inhibition was stronger for isotropy than for the two other cortex. Informed by theories in visual neuroscience that link brain markers. This analysis revealed a similar shift toward values associated activation and image spatial content, and by a growing body of work with visual discomfort for the three other sets of stimuli, apart from showing an association between visual discomfort and larger two metrics, activation level and sparseness, for Art 1 (see metabolic and electrophysiological response, we derived three types Supplementary Figures S11–S13). of measures that were candidates for characterising the model’s To evaluate how the changes in the metrics when decreasing response to uncomfortable images. We found that the level of inhibition would translate into predicted discomfort, we determined, activation, the sparseness, and the isotropy of the model population for each value of gain, the number of images with a metric value above response to images from urban scenes and abstract art were good the threshold associated with 15% most discomfort in the original predictors of the visual discomfort experienced by observers when FIGURE 5 Changes in markers of visual discomfort when the balance of excitation over inhibition is modified. Distributions of (A) activation, (B) sparseness, and (C) isotropy metrics for all the stimuli in Architecture 1 and increasing values of gain for the inhibitory layer. The gain ranged from 0, i.e., no inhibitory activity in the model (top left, light grey distribution), to 1, i.e., reference model (top right, blue distribution), in steps of 0.125. Differences between distributions and the distribution for the reference model were tested using two-sample Kolmogorov–Smirnov tests; p-values are colour coded as in Figure 3. See Supplementary Figures S11–S13 for the equivalent distributions for sets Architecture 2, Art 1 and 2. Frontiers in Neuroscience 09 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 viewing the images. Our results therefore provide new insights into 2013), greater alpha suppression (Haigh et al., 2018), greater steady- how overload in the visual system may lead to discomfort by state evoked potentials (Lindquist et al., 2021), and larger event- suggesting three possible non-mutually exclusive mechanisms: 1. related potentials (ERPs) (Haigh et al., 2019). Individuals with Discomfort may arise because the overall amount of activity of the migraine also reported greater discomfort and larger ERPs compared neurons in the network, or activation level, is too high; 2. Discomfort to headache-free individuals (Haigh et al., 2019), further supporting may arise because too many neurons have a high activity, i.e., the the link between discomfort and strong neural responses. encoding is not sparse enough; 3. Discomfort may arise when activity To our knowledge, the only modelling attempts to understand in the hypercolumns is concentrated at a given orientation, i.e., when the neural basis of discomfort so far is (Hibbard and O'Hare, the isotropy in the network response is low. 2015), in which the authors explored a putative relationship With our second experiment, we showed that the same three between discomfort and sparse coding by computing the kurtosis mechanisms may also shed light into interindividual differences in of the distribution of responses of a set of Gabor units tuned to susceptibility to visual discomfort. To this end, we took advantage of several spatial frequencies and orientations similar to the first the separation between excitatory and inhibitory populations in component of our model. The response to natural images was neurodynamic models based on Wilson and Cowan formalism much sparser than that to sine gratings. Our study extends these (Wilson and Cowan, 1972, 1973), which does not exist in models findings using a model that considers contextual modulations of based on divisive normalization (Heeger, 1992; Foley, 1994; neurons’ activity outside their CRF, carried out through lateral Cavanaugh et al., 2002; Kay et al., 2013; Schutt and Wichmann, 2017), connections, and to images of real urban scenes and abstract art in order to modify the balance between excitation and inhibition. within a continuum of levels of reported discomfort, instead of When the balance took higher ratios, e.g., because of a deficit of the two theoretically extreme corpuses of images formed by GABA neurotransmitter availability, the three markers of discomfort natural images and sine gratings. In future works, considering shifted toward values associated with increased visual discomfort, larger, more diverse datasets, for example including interior namely a higher activation, a lower sparseness, and a decreased scenes, would provide insights into the generalizability of our isotropy in the model population response. The modelling therefore results to other facets of daily life. predicts that more and more stimuli will lead to discomfort as the A prevailing thesis in vision science is that the evolution of availability of inhibition lowers. This first mechanistic account for the visual system has been driven by the selective pressure to individual difference in susceptibility to visual discomfort corroborate provide an encoding of natural stimuli that is efficient, both in previous findings of a link between GABA concentration and brain terms of transmission of information and metabolism (Attneave, activation. There is evidence for variation in GABA levels within the 1954; Barlow, 1961). For example, there is strong evidence normal population (Cai et al., 2012). Resting GABA concentration supporting a sparse representation of natural images in the visual predicts peak gamma frequency and fMRI amplitude in response to cortex (Vinje and Gallant, 2000, 2002; Weliky et al., 2003; Haider visual stimulation in humans (Muthukumaraswamy et al., 2009) in et al., 2010). Efficient coding predicts that natural stimuli are response to striped patterns (Adjamian et al., 2004; comfortable to view, which is possibly a basis for the restorative Muthukumaraswamy et al., 2009). Intriguingly, our modelling makes effect of nature (Menzel and Reese, 2022). However, and in a link between reduced inhibition and stripes or gratings, visually apparent contradiction with the efficient coding hypothesis, uncomfortable stimuli par excellence (Wilkins et al., 1984; Hermes several studies have shown an association between deviation et al., 2017): As inhibition was reduced in the model, the isotropy of from the statistics of natural images and decreased BOLD the model response, which measures the balance of activity across response. In (Isherwood et al., 2017), synthetic noise images with spatial orientations, was progressively lost in a winner-takes-all 1 / f α spectra triggered a stronger BOLD fMRI response in the process in which a single orientation would take all the activity, as in early visual cortex for a slope closer to 1 / f than for steeper or the presence of stripes. Together, our experiments support a unified shallower slopes. In (Olman et al., 2004), the BOLD fMRI view of visual discomfort in which core properties of the early visual response to natural images was stronger than to their whitened system – cells sensitive to oriented luminance edges at different spatial counterpart. In Rieger et al. (2013), randomising phases in orientations and frequencies coupled through a pattern of excitatory natural images, thereby removing edge information to which the and inhibitory connections that implement contextual influences – visual system is putatively adapted, had no impact on population partially account for both image-wise differences in induced metabolic cost as measured by blood oxygenation level dependent discomfort and individual differences in susceptibility to (BOLD) response. visual discomfort. The question then is, how to reconcile these findings and the Although a link between visual discomfort and an exceptionally theory of efficient coding? Although the theory predicts that strong neural response has long been proposed (Wilkins et al., 1984), natural stimuli are efficiently processed, it does not follow empirical evidence has only emerged recently. Le et al. (2017) used logically that deviations with respect to nature cause inefficient near infrared spectroscopy to measure the oxyhaemoglobin response neural processing. Indeed, efficient coding entails that, from posterior areas of the cortex. The amplitude of the response to considering all possible images, natural images are included in images of buildings increased with the extent to which the image the subset of images processed efficiently, along with some statistics deviated from 1 / f α . When focusing on colour, striped non-natural images. And indeed, this inclusion is strict: patterns comprising two colours were more uncomfortable when the non-natural images with a reduced dynamic range of contrast or chromaticity differences were large (e.g., red and blue) compared to artificial square-wave gratings with very low frequencies are small (pink and purple). The large chromaticity differences also processed typically. To explore this question computationally, elicited a larger haemodynamic response in visual cortex (Haigh et al., we contrasted the activation of our model against two measures Frontiers in Neuroscience 10 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 of deviations with respect to 1 / f : one that sums deviation in hypermetabolic response as in discomfort and migraine (Haigh et al., amplitude contrast with respect to natural images at any 2012), but also a temporal synchronization in this response (Binnie frequency and spatial orientation in the full two-dimensional et al., 1985). There is strong evidence of an association between amplitude spectrum (Penacchio and Wilkins, 2015), and the seizure generation and oscillations in the gamma band (30–80 Hz) slope of the amplitude spectrum (e.g., Tolhurst et al., 1992), as (Hermes et al., 2017). Our firing-rate model is based on Wilson- used in Olman et al. (2004) and Isherwood et al. (2017). We found Cowan equations (Wilson and Cowan, 1972; Wilson and Cowan, strong correlations between activation and deviation with respect 1973). Such equations are not able to generate the fast oscillations to the full two-dimensional spectrum of natural images needed for monitoring activity in the gamma band (Devalle et al., (Penacchio and Wilkins, 2015) for all sets but Art 2, but no 2017). Lastly, the neural hyperexcitability underlying differences in correlation with the spectral slope (Supplementary Material S9, susceptibility to visual discomfort may not simply rest on a higher Table and Figures S16–S19). This computational finding is in ratio of glutamatergic excitation over GABAergic inhibition, but on agreement with our former empirical measurements in (Le et al., more complex interconnections between several types of 2017), in which we found an association between increased neurotransmitters (Saxena et al., 2021). A natural progression of this metabolism and higher two-dimensional deviation with respect work is to augment the modelling by adopting more elaborate to 1 / f (Penacchio and Wilkins, 2015). Overall, this suggests that equations that allow the generation of fast oscillations (Devalle et al., BOLD fMRI responses are more tuned to the amount of contrast 2017) or by using spiking networks, and to consider a bigger diversity at different spatial scales and regions than to the spectral slope of neuron and neurotransmitter types (Shaw et al., 2020). (Field and Brady, 1997; Isherwood et al., 2017), a dependence This study demonstrates the ability for a model of the early visual better captured by measuring deviations in the full system to account for two fundamental aspects of visual discomfort, two-dimensional spectrum (Penacchio and Wilkins, 2015). This image-wise differences in induced discomfort and individual calls for a theoretical and empirical characterisation of the differences in susceptibility to visual discomfort. Three markers of deviations from the statistics of natural scenes that lead to model activity predicted more than 40 percent of the variance in inefficient processing. observers’ reported discomfort from complex, real-life images. More Nevertheless, several studies have shown that visual discomfort is efforts to integrate idiosyncratic differences linked to hyperexcitability predicted by the spatial distribution of luminance contrast (Fernandez in the modelling approach are now central to generate tools that and Wilkins, 2008; Juricevic et al., 2010; O'Hare and Hibbard, 2011; make individual predictions. Penacchio and Wilkins, 2015) or the extent to which the difference in chromaticity (Penacchio et al., 2021) deviates from that found in nature. In our study, the third marker of discomfort, isotropy, shows a similar Data availability statement relationship. In the two sets in which this metric had a large range of values (i.e., urban landscapes), how the isotropy of the encoding on an The datasets presented in this study can be found in online image differed from the typical values for natural images predicted how repositories. The names of the repository/repositories (Penacchio et al., uncomfortable to look at it was. The study therefore confirms that 2023) and accession number(s) can be found in the article/ discomfort can be triggered by some types of deviation from nature while Supplementary material. giving new insights into how image spatial content (or excessive excitation/inhibition ratio) harnesses this deviation. More generally, by offering three distinct, if partly correlated, Ethics statement mechanisms by which discomfort arises, the study provides a direct and intuitive account for the propensity of simple, laboratory stimuli The studies involving human participants were reviewed and based on gratings to provoke discomfort, headaches or seizures in approved by Institutional Review Board at the University of Nevada, photosensitive epilepsy (see Wilkins, 1995; Hermes et al., 2017). The Reno (333057). The patients/participants provided their written likelihood of causing discomfort increases with the size or the contrast informed consent to participate in this study. of the gratings (as does the model activation level) and increases if the gratings’ frequency is closer to three cpd (the predictive power of all three markers is higher around this frequency for the experiments in Author contributions which the visual angle of the stimuli was controlled). This likelihood decreases when superimposing gratings with different orientations OP: conceptualization, visualisation, and writing – original draft (plaid patterns), therefore increasing the isotropy of the model preparation. OP and XO: data curation, methodology, and software. response (Wilkins, 1995). OP, XO, and AW: formal analysis. OP, XO, SH, and AW: investigation. Nonetheless, the modelling approach is subject to important OP, XO, and SH: Resources. OP, AW, SM, and XO: writing – review limitations. As it does not include the machinery of colour processing, and editing. All authors contributed to the article and approved the the model cannot account for the contribution of chromatic submitted version. information to discomfort (Haigh et al., 2013; Penacchio et al., 2021; O'Hare et al., 2023). More importantly, while visual stress, migraine, and photosensitive epilepsy have in common a heightening of Funding metabolic and neural response, there is the caveat that our model may not be able to capture the temporal aspect of epilepsy. This publication is part of the R+D+I grant PID2020- Epileptogenic stimuli not only trigger a hyperneuronal and 118254RB-I00 financed by MCIN/AEI/10.13039/501100011033, Frontiers in Neuroscience 11 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 by the Agencia de Gestió d’Ajuts Universitaris i de Recerca Publisher’s note (AGAUR) through 2021-SGR-01470, and the CERCA Programme/Generalitat de Catalunya. OP was funded by a Maria All claims expressed in this article are solely those of the Zambrano Fellowship for attraction of international talent for the authors and do not necessarily represent those of their affiliated requalification of the Spanish university system—NextGeneration organizations, or those of the publisher, the editors and the EU (ALRC). SH was funded by a pilot award from a NIH COBRE reviewers. Any product that may be evaluated in this article, or (PG20GM103650) and salary support from a NIH R15 AREA claim that may be made by its manufacturer, is not guaranteed or grant (MH122935). endorsed by the publisher. Conflict of interest Supplementary material The authors declare that the research was conducted in the The Supplementary material for this article can be found online absence of any commercial or financial relationships that could at: https://www.frontiersin.org/articles/10.3389/fnins.2023.1200661/ be construed as a potential conflict of interest. full#supplementary-material References Adjamian, P., Holliday, I. E., Barnes, G. R., Hillebrand, A., Hadjipapas, A., and Singh, K. D. Field, D. J. (1987). Relations between the statistics of natural images and the (2004). Induced visual illusions and gamma oscillations in human primary visual cortex. Eur. response properties of cortical cells. J. Opt. Soc. Am. A 4, 2379–2394. doi: 10.1364/ J. Neurosci. 20, 587–592. doi: 10.1111/j.1460-9568.2004.03495.x JOSAA.4.002379 Albrecht, D. G., and Hamilton, D. B. (1982). Striate cortex of monkey and cat - Field, D. J. (1994). What is the goal of sensory coding. Neural Comput. 6, 559–601. contrast response function. J. Neurophysiol. 48, 217–237. doi: 10.1152/jn.1982.48.1.217 doi: 10.1162/neco.1994.6.4.559 Atick, J. J., and Redlich, A. N. (1992). What does the retina know about natural scenes. Field, D. J., and Brady, N. (1997). Visual sensitivity, blur and the sources of variability Neural Comput. 4, 196–210. doi: 10.1162/neco.1992.4.2.196 in the amplitude spectra of natural scenes. Vis. Res. 37, 3367–3383. doi: 10.1016/ S0042-6989(97)00181-8 Attneave, F. (1954). Some informational aspects of visual perception. Psychol. Rev. 61, 183–193. doi: 10.1037/h0054663 Fitzpatrick, D. (2000). Seeing beyond the receptive field in primary visual cortex. Curr. Opin. Neurobiol. 10, 438–443. doi: 10.1016/S0959-4388(00)00113-6 Barlow, H. B. (1961). Possible principles underlying the transformations of sensory messages. Cambridge, MA, USA. MIT Press. Foley, J. M. (1994). Human luminance pattern-vision mechanisms – masking experiments require a new model. J. Opt. Soc. Am. A Opt. Image Sci. Vis. 11, 1710–1719. Bates, D., Maechler, M., Bolker, B., and Walker, S. (2014). lme4: Linear mixed-effects doi: 10.1364/JOSAA.11.001710 models using Eigen and S4. R package version 1.0-6. Available at: http://CRAN.R-project. org/package=lme4 Geisler, W. S. (2008). Visual perception and the statistical properties of natural scenes. Annu. Rev. Psychol. 59, 167–192. doi: 10.1146/annurev.psych.58.110405.085632 Berga, D., and Otazu, X. (2020). Modeling bottom-up and top-down attention with a neurodynamic model of V1. Neurocomputing 417, 270–289. doi: 10.1016/j. Geisler, W. S., Perry, J. S., Super, B. J., and Gallogly, D. P. (2001). Edge co-occurrence neucom.2020.07.047 in natural images predicts contour grouping performance. Vis. Res. 41, 711–724. doi: 10.1016/S0042-6989(00)00277-7 Berga, D., and Otazu, X. (2022). A Neurodynamic model of saliency prediction in V1. Neural Comput. 34, 378–414. doi: 10.1162/neco_a_01464 Gelman, A., and Hill, J. (2006). Data analysis using regression and multilevel/ hierarchical models. Cambridge, Cambridge University Press Binnie, C. D., Findlay, J., and Wilkins, A. J. (1985). Mechanisms of epileptogenesis in photosensitive epilepsy implied by the effects of moving patterns. Electroencephalogr. Gentile, C. P., and Aguirre, G. K. (2020). A neural correlate of visual discomfort from Clin. Neurophysiol. 61, 1–6. doi: 10.1016/0013-4694(85)91065-X flicker. J. Vis. 20:10. doi: 10.1167/jov.20.7.11 Cai, K., Nanga, R. P. R., Lamprou, L., Schinstine, C., Elliott, M., Hariharan, H., et al. Haider, B., Krause, M. R., Duque, A., Yu, Y., Touryan, J., Mazer, J. A., et al. (2010). (2012). The impact of gabapentin administration on brain GABA and glutamate Synaptic and network mechanisms of sparse and reliable visual cortical activity during concentrations: a 7T H-1-MRS study. Neuropsychopharmacology 37, 2764–2771. doi: nonclassical receptive Field stimulation. Neuron 65, 107–121. doi: 10.1016/j. 10.1038/npp.2012.142 neuron.2009.12.005 Campbell, F. W., and Robson, J. G. (1968). Application of Fourier analysis to visibility Haigh, S. M., Barningham, L., Berntsen, M., Coutts, L. V., Hobbs, E. S. T., Irabor, J., of gratings. J. Physiol. 197, 551–566. doi: 10.1113/jphysiol.1968.sp008574 et al. (2013). Discomfort and the cortical haemodynamic response to coloured gratings. Carandini, M., Demb, J. B., Mante, V., Tolhurst, D. J., Dan, Y., Olshausen, B. A., et al. Vis. Res. 89, 47–53. doi: 10.1016/j.visres.2013.07.003 (2005). Do we know what the early visual system does? J. Neurosci. 25, 10577–10597. Haigh, S. M., Chamanzar, A., Grover, P., and Behrmann, M. (2019). Cortical hyper- doi: 10.1523/JNEUROSCI.3726-05.2005 excitability in migraine in response to chromatic patterns. Headache 59, 1773–1787. doi: Carandini, M., and Heeger, D. J. (2012). Normalization as a canonical neural 10.1111/head.13620 computation. Nat. Rev. Neurosci. 13, 51–62. doi: 10.1038/nrn3136 Haigh, S. M., Cooper, N. R., and Wilkins, A. J. (2015). Cortical excitability and the Cavanaugh, J. R., Bair, W., and Movshon, J. A. (2002). Nature and interaction of signals shape of the haemodynamic response. NeuroImage 111, 379–384. doi: 10.1016/j. from the receptive field center and surround in macaque V1 neurons. J. Neurophysiol. neuroimage.2015.02.034 88, 2530–2546. doi: 10.1152/jn.00692.2001 Haigh, S. M., Cooper, N. R., and Wilkins, A. J. (2018). Chromaticity separation and Dayan, P., and Abbott, L. (2001). Theoretical neurosciences: Computational and the alpha response. Neuropsychologia 108, 1–5. doi: 10.1016/j. mathematical modeling of neural systems. Cambridge, Massachussets. neuropsychologia.2017.11.020 Devalle, F., Roxin, A., and Montbrio, E. (2017). Firing rate equations require a spike Haigh, S. M., Karanovic, O., Wilkinson, F., and Wilkins, A. J. (2012). Cortical synchrony mechanism to correctly describe fast oscillations in inhibitory networks. hyperexcitability in migraine and aversion to patterns. Cephalalgia 32, 236–240. doi: PLoS Comput. Biol. 13:21. doi: 10.1371/journal.pcbi.1005881 10.1177/0333102411433301 Eccles, J. (1976). Electrical to chemical transmission in the central nervous-system. Heeger, D. J. (1992). Normalization of cell responses in cat striate cortex. Vis. Neurosci. Notes Rec. R. Soc. Lond. 30, 219–230. 9, 181–197. doi: 10.1017/S0952523800009640 Fernandez, D., and Wilkins, A. J. (2008). Uncomfortable images in art and nature. Hermes, D., Trenite, D., and Winawer, J. (2017). Gamma oscillations and Perception 37, 1098–1113. doi: 10.1068/p5814 photosensitive epilepsy. Curr. Biol. 27, R336–R338. doi: 10.1016/j.cub.2017.03.076 Ferrari, M. D., Klever, R. R., Terwindt, G. M., Ayata, C., and van den Maagdenberg, A. Hibbard, P. B., and O'Hare, L. (2015). Uncomfortable images produce non-sparse (2015). Migraine pathophysiology: lessons from mouse models and human genetics. responses in a model of primary visual cortex. R. Soc. Open Sci. 2:140535. doi: 10.1098/ Lancet Neurol. 14, 65–80. doi: 10.1016/S1474-4422(14)70220-0 rsos.140535 Frontiers in Neuroscience 12 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 Hirsch, J. A., and Gilbert, C. D. (1991). Synaptic physiology of horizontal connections in Olman, C. A., Ugurbil, K., Schrater, P., and Kersten, D. (2004). BOLD fMRI and the cats visual-cortex. J. Neurosci. 11, 1800–1809. doi: 10.1523/JNEUROSCI.11-06-01800.1991 psychophysical measurements of contrast response to broadband images. Vis. Res. 44, 669–683. doi: 10.1016/j.visres.2003.10.022 Huang, J., Cooper, T. G., Satana, B., Kaufman, D. I., and Cao, Y. (2003). Visual distortion provoked by a stimulus in migraine associated with hyperneuronal activity. Olshausen, B. A., and Field, D. J. (1996). Emergence of simple-cell receptive field properties Headache 43, 664–671. doi: 10.1046/j.1526-4610.2003.03110.x by learning a sparse code for natural images. Nature 381, 607–609. doi: 10.1038/381607a0 Huang, J., Zong, X. P., Wilkins, A., Jenkins, B., Bozoki, A., and Cao, Y. (2011). fMRI Olshausen, B. A., and Field, D. J. (2004). Sparse coding of sensory inputs. Curr. Opin. evidence that precision ophthalmic tints reduce cortical hyperactivation in migraine. Neurobiol. 14, 481–487. doi: 10.1016/j.conb.2004.07.007 Cephalalgia 31, 925–936. doi: 10.1177/0333102411409076 Parraga, C. A., Troscianko, T., and Tolhurst, D. J. (2000). The human visual system is Hyvarinen, A., Hurri, J., and Hoyer, P. O. (2009). Natural image statistics a probabilistic optimised for processing the spatial information in natural visual images. Curr. Biol. 10, approach to early computational vision introduction. Dordrecht, Springer-Verlag London. 35–38. doi: 10.1016/S0960-9822(99)00262-6 Isherwood, Z. J., Schira, M. M., and Spehar, B. (2017). The tuning of human visual Penacchio, O., Haigh, S. M., Ross, X., Ferguson, R., and Wilkins, A. J. (2021). Visual cortex to variations in the 1/f(alpha) amplitude spectra and fractal properties of discomfort and variations in chromaticity in art and nature. Front. Neurosci. 15:711064. synthetic noise images. NeuroImage 146, 642–657. doi: 10.1016/j.neuroimage.2016.10.013 doi: 10.3389/fnins.2021.711064 Juricevic, I., Land, L., Wilkins, A., and Webster, M. A. (2010). Visual discomfort and Penacchio, O., Otazu, X., and Dempere-Marco, L. (2013). A neurodynamical model natural image statistics. Perception 39, 884–899. doi: 10.1068/p6656 of brightness induction in V1. PLoS One 8:e64086. doi: 10.1371/journal.pone.0064086 Kapadia, M. K., Ito, M., Gilbert, C. D., and Westheimer, G. (1995). Improvement in Penacchio, O., Otazu, X., Wilkins, A. J., and Haigh, S. M. (2023). A mechanistic visual sensitivity by changes in local context – parallel studies in human observers and account of visual discomfort, Dryad, Dataset. doi: 10.5061/dryad.g79cnp5kw in V1 of alert monkeys. Neuron 15, 843–856. doi: 10.1016/0896-6273(95)90175-2 Penacchio, O., and Wilkins, A. J. (2015). Visual discomfort and the spatial distribution Kay, K. N. (2018). Principles for models of neural information processing. NeuroImage of Fourier energy. Vis. Res. 108, 1–7. doi: 10.1016/j.visres.2014.12.013 180, 101–109. doi: 10.1016/j.neuroimage.2017.08.016 Porciatti, V., Bonanni, P., Fiorentini, A., and Guerrini, R. (2000). Lack of cortical contrast Kay, K. N., Winawer, J., Rokem, A., Mezer, A., and Wandell, B. A. (2013). A two-stage gain control in human photosensitive epilepsy. Nat. Neurosci. 3, 259–263. doi: 10.1038/72972 cascade model of BOLD responses in human visual cortex. PLoS Comput. Biol. R Development Core Team (2020). R: a language and environment for statistical 9:e1003079. doi: 10.1371/journal.pcbi.1003079 computing Vienna, Austria R Foundation for Statistical Computing. Knierim, J. J., and Vanessen, D. C. (1992). Neuronal responses to static texture patterns Radhakrishnan, K., St Louis, E. K., Johnson, J. A., McClelland, R. L., in area-V1 of the alert macaque monkey. J. Neurophysiol. 67, 961–980. doi: 10.1152/ Westmoreland, B. F., and Klass, D. W. (2005). Pattern-sensitive enilepsy: electroclinical jn.1992.67.4.961 characteristics, natural history, and delineation of the epileptic syndrome. Epilepsia 46, Knill, D. C., Field, D., and Kersten, D. (1990). Human discrimination of fractal images. 48–58. doi: 10.1111/j.0013-9580.2005.26604.x J. Opt. Soc. Am. A Opt. Image Sci. Vis. 7, 1113–1123. doi: 10.1364/JOSAA.7.001113 Rieger, J. W., Gegenfurtner, K. R., Schalk, F., Koechy, N., Heinze, H. J., and Le, A. T. D., Payne, J., Clarke, C., Murphy, K. A., Prudenziati, F., Armsby, E., et al. Grueschow, M. (2013). BOLD responses in human V1 to local structure in natural (2017). Discomfort from urban scenes: metabolic consequences. Landsc. Urban Plan. scenes: implications for theories of visual coding. J. Vis. 13:19. doi: 10.1167/13.2.19 160, 61–68. doi: 10.1016/j.landurbplan.2016.12.003 Saxena, N., Muthukumaraswamy, S. D., Richmond, L., Babic, A., Singh, K. D., Lewicki, M. S. (2002). Efficient coding of natural sounds. Nat. Neurosci. 5, 356–363. Hall, J. E., et al. (2021). A comparison of GABA-ergic (propofol) and non-GABA- doi: 10.1038/nn831 ergic (dexmedetomidine) sedation on visual and motor cortical oscillations, using magnetoencephalography. NeuroImage 245:11. doi: 10.1016/j. Li, Z. P. (1999). Visual segmentation by contextual influences via intra-cortical interactions neuroimage.2021.118659 in the primary visual cortex. Network 10, 187–212. doi: 10.1088/0954-898X_10_2_305 Schutt, H. H., and Wichmann, F. A. (2017). An image-computable psychophysical Li, Z. P. (2002). A saliency map in primary visual cortex. Trends Cogn. Sci. 6, 9–16. spatial vision model. J. Vis. 17:35. doi: 10.1167/17.12.12 doi: 10.1016/S1364-6613(00)01817-9 Serre, T., Oliva, A., and Poggio, T. (2007). A feedforward architecture accounts for Lindquist, L. C., McIntire, G. R., and Haigh, S. M. (2021). The effects of visual rapid categorization. Proc. Natl. Acad. Sci. U. S. A. 104, 6424–6429. doi: 10.1073/ discomfort and chromaticity separation on neural processing during a visual task. Vis. pnas.0700622104 Res. 182, 27–35. doi: 10.1016/j.visres.2021.01.007 Serre, T., and Riesenhuber, M. (2004). Realistic modeling of simple and complex cell Machens, C. K., Gollisch, T., Kolesnikova, O., and Herz, A. V. M. (2005). Testing the tuning in the HMAX model, and implications for invariant object recognition in cortex, AI efficiency of sensory coding with optimal stimulus ensembles. Neuron 47, 447–456. doi: Memo 2004–017. CBCL Memo 239. Cambridge, MA, USA. MIT 239. 10.1016/j.neuron.2005.06.015 Shaw, A. D., Muthukumaraswamy, S. D., Saxena, N., Sumner, R. L., Adams, N. E., MATLAB (2019). MATLAB and statistics toolbox release 2019b, 9.7.0.1190202 Moran, R. J., et al. (2020). Generative modelling of the thalamo-cortical circuit (R2019b). Natick, Massachusetts, The MathWorks Inc. mechanisms underlying the neurophysiological effects of ketamine. NeuroImage 221:10. Meldrum, B. S., and Wilkins, A. J. (1984). Photosensitive epilepsy in man and the doi: 10.1016/j.neuroimage.2020.117189 baboon. Electrophysiology of epilepsy. Schwartzkroin, P.A., and Wheal, H.V., eds, Strogatz, S. (2015). Nonlinear dynamics and chaos: with applications to physics, biology, Academic Press, London chemistry, and engineering. Boulder, CO, Westview Press, a member of the Perseus Menzel, C., and Reese, G. (2022). Seeing nature from low to high levels: mechanisms Books Group underlying the restorative effects of viewing nature images. J. Environ. Psychol. 81:14. Tolhurst, D. J., Tadmor, Y., and Chao, T. (1992). Amplitude spectra of natural images. doi: 10.1016/j.jenvp.2022.101804 Ophthalmic Physiol. Opt. 12, 229–232. doi: 10.1111/j.1475-1313.1992.tb00296.x Meteyard, L., and Davies, R. A. I. (2020). Best practice guidance for linear mixed- van Hateren, J. H., and van der Schaaf, A. (1998). Independent component filters of effects models in psychological science. J. Mem. Lang. 112:22. doi: 10.1016/j. natural images compared with simple cells in primary visual cortex. Proc. R. Soc. B Biol. jml.2020.104092 Sci. 265, 359–366. doi: 10.1098/rspb.1998.0303 Movshon, J. A., Thompson, I. D., and Tolhurst, D. J. (1978). Spatial summation in Vinje, W. E., and Gallant, J. L. (2000). Sparse coding and decorrelation in primary receptive-fields of simple cells in cats striate cortex. J. Physiol. 283, 53–77. doi: 10.1113/ visual cortex during natural vision. Science 287, 1273–1276. doi: 10.1126/ jphysiol.1978.sp012488 science.287.5456.1273 Mulleners, W. M., McCrory, D. C., and Linde, M. (2015). Antiepileptics in migraine Vinje, W. E., and Gallant, J. L. (2002). Natural stimulation of the nonclassical receptive prophylaxis: an updated Cochrane review. Cephalalgia 35, 51–62. doi: field increases information transmission efficiency in V1. J. Neurosci. 22, 2904–2915. 10.1177/0333102414534325 doi: 10.1523/JNEUROSCI.22-07-02904.2002 Muthukumaraswamy, S. D., Edden, R. A. E., Jones, D. K., Swettenham, J. B., and Weliky, M., Fiser, J., Hunt, R. H., and Wagner, D. N. (2003). Coding of natural scenes Singh, K. D. (2009). Resting GABA concentration predicts peak gamma frequency and in primary visual cortex. Neuron 37, 703–718. doi: 10.1016/S0896-6273(03)00022-9 fMRI amplitude in response to visual stimulation in humans. Proc. Natl. Acad. Sci. U. S. A. 106, 8356–8361. doi: 10.1073/pnas.0900728106 Weliky, M., Kandler, K., Fitzpatrick, D., and Katz, L. C. (1995). Patterns of excitation and inhibition evoked by horizontal connections in visual cortex share a common relationship to Ogawa, N., and Motoyoshi, I. (2020). Differential effects of orientation and spatial- orientation columns. Neuron 15, 541–552. doi: 10.1016/0896-6273(95)90143-4 frequency spectra on visual unpleasantness. Front. Psychol. 11:1342. doi: 10.3389/ fpsyg.2020.01342 Wilkins, A. J. (1995). Visual Stress, Oxford Oxford University Press O'Hare, L. (2016). Steady-state VEP responses to uncomfortable stimuli. Eur. J. Wilkins, A. J., Darby, C. E., and Binnie, C. D. (1979). Neurophysiological aspects of Neurosci. 45, 410–422. doi: 10.1111/ejn.13479 pattern-sensitive epilepsy. Brain 102, 1–25. doi: 10.1093/brain/102.1.1 O'Hare, L., Goodwin, P., and Sharman, R. J. (2023). The relationship between visual Wilkins, A. J., Evans, B. J. W., and Plant, G. T. (2022). “Potential uses for precision discomfort and cortical excitability in cone-opponent stimuli. Brain Res. 1798:11. doi: tinted lenses in a neurology clinic” in touchREVIEWS in neurology (in press) 10.1016/j.brainres.2022.148142 Wilkins, A., Nimmosmith, I., Tait, A., McManus, C., Dellasala, S., Tilley, A., et al. O'Hare, L., and Hibbard, P. B. (2011). Spatial frequency and visual discomfort. Vis. (1984). A neurological basis for visual discomfort. Brain 107, 989–1017. doi: 10.1093/ Res. 51, 1767–1777. doi: 10.1016/j.visres.2011.06.002 brain/107.4.989 Frontiers in Neuroscience 13 frontiersin.org Penacchio et al. 10.3389/fnins.2023.1200661 Wilson, H. R., and Cowan, J. D. (1972). Excitatory and inhibitory interactions Zhaoping, L., and May, K. A. (2007). Psychophysical tests of the hypothesis of a in localized populations of model neurons. Biophys. J. 12, 1–24. doi: 10.1016/ bottom-up saliency map in primary visual cortex. PLoS Comput. Biol. 3, 616–633. doi: S0006-3495(72)86068-5 10.1371/journal.pcbi.0030062 Wilson, H. R., and Cowan, J. D. (1973). Mathematical theory of functional Zhaoping, L., and Zhe, L. (2015). Primary visual cortex as a saliency map: a parameter- dynamics of cortical and thalamic nervous tissue. Kybernetik 13, 55–80. doi: free prediction and its test by behavioral data. PLoS Comput. Biol. 11:e1004375. doi: 10.1007/BF00288786 10.1371/journal.pcbi.1004375 Zhang, X. L., Zhaoping, L., Zhou, T. G., and Fang, F. (2012). Neural activities in Zipser, K., Lamme, V. A. F., and Schiller, P. H. (1996). Contextual modulation in V1 create a bottom-up saliency map. Neuron 73, 183–192. doi: 10.1016/j. primary visual cortex. J. Neurosci. 16, 7376–7389. doi: 10.1523/ neuron.2011.10.035 JNEUROSCI.16-22-07376.1996 Frontiers in Neuroscience 14 frontiersin.org