Visual stress responses to static images are associated with symptoms of Persistent Postural Perceptual Dizziness (PPPD)

1.1Natural and unnatural visual scenes

Human visual systems are thought to be optimised to process scenes with the characteristics of the natural environment, which tends to contain a broad range of orientations and a predominance of low spatial frequencies [19, 49]. For example, a natural Welsh landscape, with gentle hills, clouds and foliage has the consistent structure of natural scenes, with big objects providing most of the contrast variation in the scene and small objects or details providing less contrast variation. This property is reflected in the amplitude spectrum of the spatial frequencies that compose the scene: as spatial frequency increases, amplitude tends to reduce in a characteristic way [19, 21].

Efficient information processing requires sparse coding [60]. Having more neural activity in the visual cortex is not necessarily a good thing for processing visual information. Efficient processing would be expected to entail relatively sparse activity. Natural scenes and images that share these properties require less neural energy to process –perhaps because they are the environments in which we evolved and developed to perceive [19, 20, 46, 60]. Images that deviate from these statistical properties appear to be processed less efficiently: they are discriminated less well [22, 35, 47], they are subordinate in binocular rivalry [3], and they tend to produce a larger neural response [28, 38].

Many of the environments that elicit PPPD symptoms appear to deviate dramatically from natural scene statistics. A supermarket, for example, is highly cluttered with small objects and contains many more cardinal (horizontal/vertical) than oblique orientations. We informally observed that these types of environment show similarities to images that are known to produce visual discomfort (also known as visual stress) in individuals without dizziness. These uncomfortable images contain a limited range of orientations and a predominance of mid-high spatial frequencies [18, 33, 45, 49]. Particularly uncomfortable are images with an excess of aligned orientations with contrast energy at spatial frequencies around 3 cycles/degree, which corresponds to the peak sensitivity of the visual system [4, 7].

In the present study, we asked patients with PPPD and a large general population sample to rate a set of images for the amount of visual discomfort they experienced when viewing them. It should be emphasised that all the images were static, so discomfort did not depend on image motion or simulated self-motion that would be expected to be problematic in PPPD through interaction with balance processing. Instead, discomfort to static images could indicate that individuals with PPPD respond to visual information differently –compared to individuals without PPPD. This could potentially exacerbate dizziness symptoms if vision is relied upon for postural control.

The static images had previously been rated as either low or high discomfort by general population participants, and this was supported by an analysis of their spectral content that measured the degree to which they deviated from natural scenes in terms of orientation distribution and spatial frequency content [49] (cf. below). We investigated whether patients with PPPD, and individuals in the general population with more PPPD symptoms, rated the ‘high-discomfort’ images as more uncomfortable than did those with fewer PPPD symptoms.

1.2Association with migraine

It has been known for some time that individuals with migraine report a general increase in visual discomfort to images that deviate from the statistical properties of natural scenes [25, 41, 59, 72, 73]. They also show heightened neural responses in the visual cortex when viewing these images, suggesting that visual processing areas may be hyper-responsive in migraine [9, 28].

Migraine is associated with PPPD [52, 63], so we would expect that individuals with both migraine and visually-induced dizziness symptoms might also report increased visual discomfort. Of greater interest is whether the relationship between PPPD symptoms and visual discomfort only exists when there is co-occurring migraine, or if it exists irrespective of migraine. Therefore, when assessing the association between PPPD symptoms and visual discomfort, we control for the presence of co-morbid migraine.

2.1Participants

Patient cohort: Thirty patients were recruited from the vestibular clinic at University Hospital Wales (UHW). All patients had received a diagnosis of PPPD from a clinical scientist in Audiology or a Consultant Audiovestibular Physician, following the ICVD criteria [63] and common tests to examine vestibular functioning, including Halmagyi bedside head thrust testing, Video Head Impulse testing (vHIT using Synapsys system, Synapsys Solutions Ltd, West Sussex, UK), Videonystagmography (typically saccades, pursuit, gaze using GN system) and (sometimes) caloric testing if deemed necessary. Some patients had additional vestibular conditions (see Table 1). The average age of participants was 44 (sd = 14.3, range 11–67), 60%were female.

General population cohort: Surveys were sent to 18,683 members of a community public health participant list in Wales, and we received approximately 2000 responses. Participants who reported a current diagnosis of any common vestibular-related conditions were excluded from all analyses (N = 193). Following this exclusion, we obtained 1858 responses for discomfort image ratings and migraine screening questionnaire. Of these, 1845 provided age and 1853 provided gender information. Of these, 1797 completed the visual vertigo analogue scale (VVAS) and 1435 completed the situational characteristics questionnaire (SCQ). A total of 1392 participants had a full set of responses on all measures. The average age of participants was 55 (range 18–88), 74%were female. No payment or compensation was offered to participants. All procedures were approved by the School of Psychology, Cardiff University, ethics committee.

2.2Materials

All aspects of the survey were delivered via Qualtrics (Provo, UT), an online survey tool.

Visual discomfort images: We selected a random subset of 20 images from Penacchio and Wilkins (2015), which had been rated as high discomfort (n = 10) or low discomfort (n = 10). Images were taken from three categories: photographs of buildings, abstract art, and geometric shapes. Examples are shown in Fig. 1. Spectral analysis from Penacchio and Wilkins (2015) showed that the high-discomfort images contain a predominance of medium-high spatial frequencies and a narrower range of orientations, while the low-discomfort images conformed more to the statistical properties of natural scenes. This was reflected by a greater deviation of the amplitude spectrum of high-discomfort images, than low-discomfort images, from the average amplitude spectrum of 2000 natural images (the mean ratio of deviation of the high-discomfort images in the Van Hateren and van der Schaaf database [69] is 32.2, sd = 28.5, compared a mean deviation ratio of 3.9, sd = 2.4, for low-discomfort images).

Fig. 1

Example low discomfort images (top row) and high discomfort images (bottom row) used in the study.

On a standard 22 inch monitor with a viewing distance of 60 cm, the images subtended 25°×15° visual angle. The images were imbedded in a Qualtrics questionnaire and were viewed on participants’ personal devices, so they were rendered at different sizes and resolutions across participants; we asked participants at the beginning of the questionnaire to use the device with the biggest screen available (e.g., tablet preferable over a phone). The majority of general population participants used a computer monitor or a tablet to view the images (computer = 53%, tablet = 24%, phone = 23%). The visual angle of the images may not have differed much between computers and tablets, since the latter are generally held closer. But the visual angle is likely to have been smaller for phones. Since this was a large population survey, we had to accept this source of variability.

Visual Vertigo Analogue Scale (VVAS): The VVAS [13] is a short questionnaire that asked respondents to indicate on a rating scale from 0–10 the amount of dizziness they experience in 9 different situations. These situations are known triggers for patients with PPPD and include walking down a supermarket aisle, walking across a patterned floor, and going to the cinema. The items are then averaged, and this average score is multiplied by 10. The total score an individual could achieve by rating all situations a 10 (maximum dizziness) is 100. Note that the original version of the VVAS asks patients to place a mark on a continuous line on a piece of paper. We had to adapt this for delivery online. In our version, participants were asked to place a virtual line between 0 and 10, delimitated by increments of 1. However, since we do not use the score of the VVAS in an absolute sense, but merely to correlate with visual discomfort, this difference in method is not important here. Furthermore, the scale worked well insofar as patients scored higher than members of the general population and internal consistency was excellent (Cronbach’s alpha above 0.9, [52]).

Situational Characteristics Questionnaire (SCQ): The SCQ [32] was originally developed as a measure of space and motion discomfort, however, this condition is now considered to fall under the umbrella term of PPPD [63]. The SCQ is a 20 item questionnaire that, like the VVAS, also asks about discomfort in situations that trigger visually-induced dizziness and discomfort. Situations are rated between 0 and 3 and scores are normalised by subtracting responses to paired situations that are not commonly associated with visually-induced dizziness. The final score is obtained by dividing the summed ratings across all items by the total number of items and then multiplying by 10, therefore, the maximum score that can be given is 30. The SCQ was an optional questionnaire in the survey and consequently we are missing data for 425 participants (total N = 1465). The SCQ in our sample had acceptable internal consistency with a Cronbach’s alpha of 0.79.

Migraine Screen Questionnaire (MS-Q): This is a five item screening tool developed to identity migraine based on the criteria of the International Headache Society [37]. Participants answer yes/no questions about headache episodes they experience, such as ‘Do you usually suffer from nausea when you have a headache?’ and ‘Does light or noise bother you when you have a headache?’. Participants must response ‘yes’ to four or more of the five questions to have a result of probable migraine. The MS-Q is found to have good internal consistency (Cronbach’s alpha = 0.82) and high sensitivity (0.82–0.93) and specificity (0.81–0.97) in both neurological clinics and primary care [36, 37].

2.3Analysis

Our measure of visual discomfort was the ‘discomfort index’, defined as the difference in ratings between the low and high discomfort images. A greater score indicates an aversion to images that deviate from natural scene statistics rather than a bias to use higher ratings for all images. To provide reassurance that participants were appropriately using the scale, we checked that the high discomfort images were consistently rated as producing more visual discomfort than the low discomfort images: patients, t (29) = 11, p < 0.001, d = 1.96; general population, t (1857) = 42, p < 0.001, d = 0.97). To compare patient scores to controls whilst controlling for age, gender and migraine, we used ANCOVA. To assess the correlation between discomfort scores and PPPD symptoms in the general population we used Spearman correlations. For each analysis we used the maximum number of participants with complete responses for the information needed for that analysis.

4.1Visual processing of static images

All the images were static, so discomfort did not depend on the kind of motion in a video or simulated self-motion that would be expected to interact with balance processing and would be expected to be problematic in PPPD. These findings were not explained by co-morbid migraine, which has a known association with visual discomfort [25, 41, 59, 72, 73] and PPPD [51, 63, 66]. Indeed, PPPD symptoms showed a significantly stronger relationship with visual discomfort than did migraine.

We believe that these findings constitute preliminary evidence that visual processing itself (not just reliance on, and integration of, visual signals for balance) might be atypical in individuals with PPPD. A possible hypothesis is that some visual cortices are prone to a type of over-activation that renders them inefficient at processing certain kinds of scenes –ones that deviate markedly from natural scene statistics. These scenes are already known to be processed less well on average than images with natural statistics [22, 35, 47]. There is some initial evidence that dizziness symptoms in PPPD might correlate with increased neural activity in the visual cortex [54] –although that study explored motion stimuli and not static images. What such over-activation means remains unknown, and the reason why some visual systems may be more prone to it would only be speculation at this time.

In migraine and in Meares-Irlen Syndrome (visual stress) it has been suggested that increased visual discomfort might be due to hyper-responsivity in the visual cortex, though again, what this means mechanistically remains unknown [9, 12, 17, 28, 71–73]. There is some indication of greater or wider response in visual cortex when viewing pattern-glare-inducing stimuli or simply reading words [8, 28, 29]. If the visual cortex is hyper-responsive, this could be associated with higher levels of neural noise and lower levels of inhibition, which results in poorer visual discrimination abilities [16, 42, 45]. In psychophysical experiments, individuals with migraine tend to show higher contrast sensitivity thresholds and motion discrimination thresholds [12, 57, 68], which could be driven by a lack of inhibition relative to excitation in the visual cortex [2, 34, 45]. Poorer contrast sensitivity in migraine is also related to poorer performance on motion processing tasks [58]. It is not known if individuals with PPPD also show higher contrast sensitivity thresholds or general difficulties with spatial and motion discrimination, and this would seem to be an important avenue for future research. Poorer visual discrimination, or more aversive responses to visual scenes, might render the visual signal a less reliable cue for postural control, which could especially destabilise PPPD patients.

4.2Relationship to moving images or self-movement

While our study explored discomfort in response to static, not moving, images, it remains the case that PPPD is normally associated with moving around environments or watching moving images on a screen. But symptoms tend not to occur for movement through all environments; some are worse than others. It seems that a combination of movement and cluttered scenes –such as supermarkets –are especially problematic. It seems plausible –though it is currently speculation - that hyper-activity to scenes with structured repetition deviating from natural scenes could make it especially difficult for a visual system to extract efficiently accurate self-movement signals. In this view, static deviation from natural scenes is merely uncomfortable; adding movement to such scenes introduces a further problem of integration across senses, and becomes potentially nauseating. If vestibular signals have also been compromised for some reason, or if vestibular processing networks are atypical [27, 30, 39, 40, 44, 48, 70, 75], this cumulation of factors may be what leads to debilitating PPPD.

In the future, it would be interesting to extend the approach of studying visual discomfort when viewing static images to the types of visual motion stimuli that individuals with PPPD are particularly sensitive to. It may be possible to build a computational model that captures the spatial frequency and orientation content of moving images and combines this with a measurement of the visual motion within the stimuli. This tool could be used to predict the types of stimuli that are most likely to trigger symptoms in individuals with PPPD or are more effective for visual desensitization.