Visual function development and Perceptual Learning
Prof Uri Polat
Since the pioneering studies by Hubel and Wiesel, and their suggestion for a critical period for visual development, Amblyopia (the lazy eye syndrome) is typically treated at very early age by patching the strong eye, to allow the formation of stronger connections for the input arriving from the weaker eye. It was generally accepted that Amblyopia is untreatable in adulthood.
Our team member, Dr. Uri Polat and colleagues were the first to demonstrate that vision can be markedly improved in adults suffering from amblyopia using a novel perceptual learning technique (Polat et al., Proc Natl Acad Sci U S A. 101:6692-7, 2004). This is done by presenting an oriented Gabor stimulus at low contrast (target) together with co-aligned flanking stimuli of the same orientation and spatial frequency at different distances (see below). The participant's task is to report the target stimulus orientation, and the target contrast is adjusted to maximize training at near-threshold conditions. Following training for 4-6 months, participants showed a major improvement in their visual acuity (78% compared to baseline, see learning dynamics in the Figure below). The contrast sensitivity was also remarkably improved. The treatment produced a significant improvement in sensitivity in all spatial frequencies, such that the CSF of amblyopic patients was within the normal range (shaded region). Similar improvement in visual acuity and contrast sensitivity was also found when applying the training on amblyopic children age 6-9 years old. These results suggest that corrective training using the simple paradigm described above, may well improve visual capabilities in our patient population.
Figure 1: Left: The Gabor stimuli used in the training sessions. The spatial frequency, target contrast, and distance between target and flankers was manipulated to obtain the best learning results. Middle: Dynamics of CSF improvement. Grey zone indicates the normal range. The gains are sustained a year later. Right: Learning curves of adult amblyopic patients showing improved visual acuity. Black line shows patient without treatment.
Polat's group is also studying crowding and binocular vision. Visual crowding is the inability to recognize objects in clutter and sets a fundamental limit for letter and object recognition throughout most of the visual field. Thus, a letter easily recognized on its own, becomes unrecognizable when surrounded by other letters. To appreciate this maintain fixation on the red spot below and try to recognize the middle letter in the triplet presented on the right side. This is much more difficult than when the letter is presented on its own as in the left side. This perceptual difficulty arises from crowding.
Crowding is classically considered a hallmark of the peripheral vision, and is generally absent in the fovea in typically-developing adults. Foveal crowding effects are pronounced in early childhood but are typically overcome with further development. Similarly, lateral effects, induced by nearby flanking elements, which can either enhance or mitigate target detection (see Figure 1, left) depending on their distance from the target are modified during development. The graph below, from Polat's group recent paper "Development of global visual processing: From the retina to the perceptive fields" shows this clearly:
Polat and colleagues studied what other processes might contribute to perceptual improvement in various visual functions (crowding and contrast sensitivity) by concurrently measuring foveal thickness (using optical computer tomography; OCT) in typically-developing participants 3 to 17 years old. Normal retinal maturation involves displacement of the retinal layers in three developmental steps; peripheral migration or displacement of the inner retinal layers that form the foveal pit, central-ward migration of cones and their elongation, and diminution in the thickness of the cones, which increases the density of foveolar cell packing. These processes reflect the maturation of the retina and accordingly affects visual acuity and contrast sensitivity. Indeed, in the above study foveal thickness and macular volume for children below 6 years were significantly lower than for the older group. Moreover, significant correlation was found between age, increased foveal thickness, better contrast sensitivity and lesser foveal crowding. The figure below shows the relationship between foveal thickness and contrast sensitivity in the studied group of children.
Figure 2: A graph depicting the pattern of target-flanker interactions as a function of their separation (see example on the right). Values of collinear modulation above zero indicate enhancement of target detection at low contrast levels. Negative values correspond to target suppression. The open circles depict data of the younger cohort (3-6 years) while the filled circles are the data of the older children (11-17 years). Note that with age, the curve is shifted to the right, such that target enhancement is seen across a wide range of target-flanker separations. This is presumed to result from changes in the patters of connectivity of neurons in primary visual cortex.
Figure 3: Scatterplot depicting the contrast sensitivity of the studied children (higher value is greater sensitivity) as a function of their foveal thickness (measured using OCT). The open circles depict data of the younger cohort (3-6 years) while the filled circles are the data of the older children (11-17 years). The two groups differed in both measures, with the older children having, on average, better contrast sensitivity and larger foveal thickness. In this group, better contrast sensitivity was associated with greater foveal thickness. In the preschoolers, behavioral performance is likely to be affected by other factors (such as motivation and attention) obscuring the correlation with the foveal thickness measure.
The results above indicate that physiological factors at the level of the eye (i.e. foveal thickness) affect performance on various visual tasks such as contrast sensitivity, visual acuity, and crowding. However, vision does not end at the eye, and much of the refinement of function is achieved through further processing that occur in the brain regions devoted to visual analysis. These are compromised in the lazy eye syndrome, but can be substantially improved with designated perceptual learning techniques. Crucially, these techniques are also applicable in adults (see Figure 1). Till now we discussed a case of monocular amblyopia in which vision in the "lazy" eye was dramatically inferior to that of the fellow eye. One may convincingly argue that in such cases, the poorer vision is mainly due to competitive processes between the two eyes, and this has little bearing to our case with the late-treated cataract patients in which both eyes were affected, rendering the children practically blind.
However, Polat and colleagues also studied an adult person (LG; 20y) that had binocular amblyopia and suffered from face and object agnosia (in which case, both eyes have poor vision; similar to the pre-operative condition of our late-treated children). Prior to training, LG’s visual acuity and crowding effects, were extremely poor, similar to those of ~6 year old children. Intensive visual training for 9 months, based on the protocol described above, led to dramatic improvement in visual acuity, crowding, binocular stereopsis and contour integration reaching near-age-level performance, and showing long-term (over 4 years) persistence. Performance on the crowding task is shown on the left. In marked comparison, LG’s object and face perception skills did not improve (see figure on right), indicating that the protocol used is inefficient for inducing learning in these domains, or that object and face perception are more prone to critical period for their proper acquisition. For more details See "Training-induced recovery of low-level vision followed by midlevel perceptual improvements in developmental object and face agnosia" in the publications section.
Figure 4 left: (A) the stimuli used to assess visual crowding, by comparing performance in the crowded vs. uncrowded case. The central stimulus was always presented in the fovea. Each eye was tested separately by occluding the other eye with a patch (B) The level of crowding experienced by LG as a function of the length of training. Lower values indicate lesser crowding. Note the dramatic improvement in LG's performance which is largely sustained well after the end of the training program. (C) LG is still relatively poor at recognizing visual objects (such as the cat depicted on the left, in full or behind bars). Numbers indicate % correct responses.