Session 5

Assessment Depth Cue Dynamics

Christopher Tyler and Spero Nicholas

Monocular depth cues are often designated as 'shape-from-X', where X denotes 'shading', 'motion', 'texture', etc. Depth cues in the real world generally come fully coordinated, with the various cues expressing the same depth structure, even though some may be spatially sparse. When one cue conveys different depth structure from the others, it will typically be downweighted. However, when the focus is on one cue to depth structure, with the others (including the cognitive information) specifying a flat plane, visual processing of the implied depth has a perceptible time course to full development. We measured the perceived depth dynamics for the single monocular cues to depth in a flat plane image by alternating with a stereoscopic image of the same depth structure having an exponential rise time towards its full amplitude. The exponential time constant and amplitude parameters of the comparison stereo dynamics were adjusted by the observers on a trial-by-trial basis to match the perceived dynamics of the monocular depth cues of shading, texture, motion and contrast. In practice, the perceived depth dynamics were complex, with time constants up to several seconds, and varied among cues. Implications for the Bayesian theory of cue reweighting based in reliability will be discussed.

Art and the brain: the view from dementia

Cosima Gretton

By examining the works of artists suffering from dementia, we can gain new perspectives on the neural basis of the perception and production of art. We review the literature in an attempt to highlight common patterns in the changes inflicted on artistic output by three types of dementia: Frontotemporal Dementia (FTD), Alzheimer's Dementia (AD) and Dementia with Lewy Bodies (DLB). We compared the brain areas thought to be involved in artistic production, with areas affected in each dementia type, and the respective neuropsychological findings. In doing so we attempted to identify the neural basis for each observed change in artistic ability. A keyword search on OVID SP PsychINFO and Medline databases, and a reference search, produced 45 articles that were used in the final review. A total of 14 were used as the final cases to be reviewed (AD = 5, FTD = 7, DLB = 2). The results showed that in AD, artists tended towards simplification of colours, loss of perspective and distortion of the image. DLB patients showed more pronounced visuospatial impairments than AD. FTD patients by contrast showed preserved visuo-spatial constructional abilities and an obsessive drive towards artistic production.

A perspective view of Leonardo Da Vinci's Last Supper

George Sperling

Leonardo Da Vinci's mural, The Last Supper, was painted from 1495 to 1498 on a wall of the refectory of Santa Maria delle Grazie, a church and Dominican convent. It was painted in one-vanishing-point perspective. It apparently has not previously been observed that this painting was designed to be viewed from a particular point, approximately in the middle of the refectory, where the church officials would be dining. Viewed from this point, the perspective lines of the structures depicted in the painting blend with the perspective of the room so that the painting presents to the eye the same image as a live re-enactment of the scene. In other words, to a stationary observer at this point, the painting should appear to be a real three-dimensional scene. A photograph of a real re-enactment by actors and a photograph of the painting would be indiscriminable. Leonardo's sophisticated use of perspective is in start contrast to the large fresco by Giovanni Donato da Montofano painted on the opposite wall, also in 1495. which has the perspective defects typical of medieval art.


Nicholas Wade

Icons are eye-cons: they provide distillations of objects or ideas into simple pictorial shapes (see pdf). They create the impression of representing that which cannot be presented. Eye-cons are tricks of vision so that what is seen does not necessarily correspond to what is presented. They are visual allusions rather than visual illusions, although they can display illusory effects. Iconography can refer to representations of people, and it has been applied to visual artists and scientists: their portraits are often reproduced in histories of art and science. Until the nineteenth century, artists were mostly represented in pigment (paintings) and scientists on paper (engravings). After the birth of photography, both have been captured by the camera and more recently manipulated by computer. Eye-conographs are 'perceptual portraits' of artists and scientists; they combine facial features with the styles and phenomena with which the artists and scientists are associated. The survey commences with eye-conographs of Richard Gregory and Tom Troscianko; it then focuses on the artists and scientists who have produced eye-cons, that is, pictures which are intended to fool the eye.

Climate, illumination statistics, and the style of painting

Isamu Motoyoshi

There is a remarkable difference in the style of classical paintings between Europe and East-Asia. This diversity may originate from the light environment specific to each climatic zone (Motoyoshi, 2011, Journal of Vision, 11(11): 1188). It is well known that Monsoon Asia is far more rainy and cloudy than Mediterranean. Analysis of large weather database indicates that increasing cloud ratio in the sky decreases contrast and skewness of the 'light field' (r > 0.84). Importantly, these illumination statistics critically constraint visual appearance of natural objects (c.f., Motoyoshi & Matoba, 2012, Vision Research, 53, 30-39). Under light fields of high contrast and skew (Mediterranean), objects tend to have variegated shading, sharp highlights, and cast shadows. Under light fields of low contrast and skew (Monsoon), on the other hand, objects have shadings only in the deepest concavities, and no highlights or cast shadow. Such characteristics are consistent with the styles of paintings in East-Asia and Europe. The consistency was validated in terms of simple image statistics as coded in early visual cortex. These results support the notion humans prefer artworks that represent the average, or canonical, property of light environments in their province.