The foundations of development and deprivation in the visual system

Abstract

The pioneering work of Torsten Wiesel and David Hubel on the development and deprivation of the visual system will be summarised, together with some comments on their influence, and some personal reminiscences by the author.

Shortly after David Hubel and Torsten Wiesel published their magnum opus on the physiology and function of the cat visual cortex (Hubel & Wiesel, 1962), they came out with six shorter papers on the development and effects of deprivation on the visual pathways. Their work on the visual cortex was a model for how the cerebral cortex is organized, putting function and connections together to show how it all works. Their work on development and deprivation pulled together a much wider array of subjects – philosophy, clinical problems, and effects of experience on the connections of the nervous system as well as function and organization.

Four principles came out of their work. First, they showed that there is some organization to the visual cortex at birth (Hubel & Wiesel, 1963). This may not be as inborn as they first supposed, because they did not know that there is synchronization of activity within afferent pathways before birth, determining some aspects of the organization, in addition to whatever genetic influences there might be (see Hubel & Wiesel, 2005, p. 402). However, there is no doubt that the cortex has some organization at birth, and certainly by the time of opening of the eyes. Therefore, it is not totally a tabula rasa which is organized by the visual environment as the animal develops.

Second, they showed that the connections degenerate as a result of visual deprivation, such as monocular deprivation, binocular deprivation, and squint, also known as strabismus. Monocular deprivation reduces the percentage of cells that can be driven by the deprived eye (Wiesel & Hubel, 1963), binocular deprivation reduces the specificity of the receptive fields of single cells in the cortex for orientation and direction of movement (Wiesel & Hubel, 1965), and squint reduces the percentage of cells that receive binocular input (Hubel & Wiesel, 1965). Moreover, the connections also develop as a result of visual input: although there is some organization at eye opening, the connections become stronger and the receptive fields more tightly organized with age (Hubel & Wiesel, 1963). These two points put together show that both nature and nurture are at work in development – a question debated by philosophers over many centuries, often using the visual system as an example.

Third, they showed that competition between afferent inputs is responsible for some of the synaptic changes. Thus the results of binocular deprivation is not the sum of the results of two monocular deprivations. If it was, very few cells would be driven by either eye. However, many cells can still be driven by both eyes (the percentage has changed over the years, depending on the investigator and techniques used), although their receptive fields are significantly less specific than normal (Wiesel & Hubel 1965).

Fourth, they showed that there is a critical period for the effects of monocular deprivation, lasting in the cat from eye opening to about three months of age (Wiesel & Hubel, 1963). Monocular deprivation in the adult does not lead to similar synaptic changes in the visual cortex.

These four points were elaborated by them in further work on the cat, and some mammoth papers on the macaque (Hubel & Wiesel, 2005), but essentially the major points were all brought out in the initial six papers. The work has since led to a plethora of contributions from other authors, on the synaptic reorganizations involved in monocular and binocular deprivation, the effect of deprivation on other parts of the visual system, development of a number of visual properties besides binocularity, critical periods in a variety of other animals with eyes in the front of the head, techniques to avoid amblyopia in humans, or to treat it if it is not avoided, and the biochemical mechanisms that allow plasticity in young animals, and bring the critical period to a close around puberty (see Daw, 2006). The last two will be treated in more detail below.

Some of the concepts, such as the importance of both nature and nurture, the existence of critical periods, and the role of competition in development, have had an influence well beyond the visual system. For example, the idea of competition as a mechanism determining synaptic connections in development has been elaborated in numerous other systems, such as the neuromuscular junction and the superior cervical ganglion (see Purves & Lichtmann, 1985). The existence of critical periods for the learning of language and violin playing, inter alia, was known before Hubel and Wiesel came along, but the investigation of critical periods for the loss of the ability to vocalize some sounds came later (Werker et al. 1981), as did the effects of violin playing on the size of the motor cortex (Watanabe et al. 2006). The comparative influences of nature and nurture on development is an abiding question in development of behaviour and the nervous system, but the conclusion of Hubel and Wiesel that they both play a role probably applies to nearly all systems.

Some personal reminiscences

My own contribution to this field started much later than my two-year sojourn in their laboratory as a postdoctoral fellow. I first applied to work with them as a graduate student, but was told that graduate students were not accepted, following the example of Francis Crick, who is said to have told one applicant ‘I rarely take graduate students: Americans, never’. After four years working on the goldfish retina, during which I discovered double opponent cells, I was acceptable, and started a collaboration with Alan Pearlman in Harvard Neurobiology.

In those days, one had to choose a topic that did not conflict with what Torsten and David were doing. Indeed, it was impossible to find out what they were doing. The door to their laboratory would be open during experiments, and one could wander in and look at the set-up. One would be met politely, but all activity would cease until one left again, after some small talk or comments on a few minor topics. Daily discussion in the lunch room would concentrate not on what David and Torsten were doing, but on what others were doing, and whether or not they were turkeys.

Alan and I proposed that we look for double opponent cells in the macaque lateral geniculate nucleus (LGN). These existed in the goldfish retina (Daw, 1967), and in the macaque visual cortex (Hubel & Wiesel, 1968), but had not been seen in the macaque LGN (Wiesel & Hubel, 1966). We were encouraged in this venture by a letter from Russell DeValois, who claimed to have seen some in his studies of the macaque LGN, mentioned in a brief sentence in an obscure publication. We should not have been so stupid as to think that Torsten and David would have missed anything, but we were encouraged by them to proceed. However, after failure in 13 macaques (Daw, 1972), Torsten quietly suggested that we find another topic. Today, we would have desisted, or been told to desist, much sooner.

So, we went to work on the cat LGN. There, we obtained evidence that Granit's red modulator could be explained as a subtraction of the rod curve from the measured spectral sensitivities of the cells recorded (Daw & Pearlman, 1969). We postulated that maybe cats could see colour by using rods in conjunction with a single class of cones, as humans can do (McCann, 1972). This hypothesis was encouraged by the disdain at Harvard Neurobiology for Granit's dominator/modulator theory (Hubel & Wiesel, 2005, p. 193), which had just been featured in his Nobel prize lecture (Granit, 1968). However, it became clear from behavioural experiments that cats could see colour at levels at which the rods are saturated (Daw & Pearlman, 1970), so some cones apart from the green-absorbing cones must exist. We finally found a double-opponent colour-coded cell at the end of a penetration into the cat LGN, with input from green-absorbing, and blue-absorbing cones. David and Torsten, perhaps goaded by Eric Kandel's remarks on their own work (Hubel & Wiesel, 2005, p. 404–405), told us that we had to find some more. At the end of several more months of work, we had found four in a population of 434, all in layer C (Pearlman & Daw, 1970, 1971). This was perhaps the first evidence that the properties of cells in layer C of the cat LGN are different from those in the A layers.

My ventures into the effects of deprivation in the cat visual cortex started unintentionally. Harry Wyatt and I wanted to look into the synaptic mechanisms of visual deprivation. We argued that this would be much easier if we could find a form of deprivation that affected the retina, rather than the visual cortex, because the synaptic mechanisms in the retina were, and still are, much more clearly understood than those in the visual cortex. Torsten and David had found that monocular deprivation was an effective form of deprivation in the cortex, because it is a competitive stimulus for the synapses onto the cells there. We hypothesized that directional deprivation might be a competitive stimulus in the rabbit retina, where direction-selective cells are a large percentage of the ganglion cells recorded. We therefore started rearing rabbits in a drum with vertical stripes on it, with the drum continually rotating around the rabbit in one direction, and recorded ganglion cells to see if the percentage of cells responding to the rearing direction increased. The result was negative (Daw & Wyatt, 1974).

We then had to determine if the negative result was due to the apparatus or the site and species of recording. We therefore tried the same experiment in the cat, recording from the cortex rather than the retina, with a positive result (Daw & Wyatt, 1976). It seemed that the rabbit retina was hard wired soon after eye opening, whereas the cat cortex had some plasticity.

Around this time, two other groups also showed that a form of directional deprivation can affect the cat visual cortex (Tretter et al. 1975; Cynader et al. 1975). We therefore decided to extend our study, and compare the critical period for directional deprivation with the critical period for monocular deprivation. Surprisingly, they were different: the critical period for directional deprivation ended earlier (Daw & Wyatt, 1976). Shortly afterwards David and Torsten published a result showing the same general point in macaques – the M system becomes set in place before the P system, judging by the ocular dominance columns seen in layer IV of striate cortex (LeVay et al. 1980).

These personal reminiscences serve to emphasize a point made by Torsten and David in their overview of their work (Hubel & Wiesel, 2005): how progress is made in science is not reflected in how either papers or grant proposals are written. None of my major discoveries – the finding of double opponent cells, the point that cells in the C layers of the cat LGN have different properties from those in the A layers, or the discovery that different visual properties have different critical periods – ever found its way into a grant proposal. Sometimes they were mentioned in the Progress Report and the Background and Significance, but never the Proposal, and none of them would have been funded by today's Study Sections without preliminary evidence that the discovery had already been made. What appears in grant proposals are the minor developments that elaborate on a result after the major discovery has been found. Indeed, the only major discovery that I made because I was looking for it was the discovery of double opponent cells, which were suggested by the work that I had been doing with Edwin Land on colour vision. The others were all made in the course of looking for something else.

Effect of the work of Hubel and Wiesel on clinical practice

David and Torsten were both trained as MDs, and spent their initial time together in the Willmer Ophthalmology Institute at Johns Hopkins Hospital. Thus they were well aware that strabismus leads to amblyopia in children, if not corrected before the age of 7 or 8. Indeed, their clinical experience was one of the reasons for their study of deprivation in the visual system, and presumably led them to look for a critical period for monocular deprivation. Ophthalmologists knew that strabismus needed to be treated early. Nevertheless, the discovery of anatomical and physiological changes underlying the effects of visual deprivation led to renewed research into the critical periods in humans, and increased efforts to intervene early (see Mitchell & MacKinnon, 2002).

One of the outstanding questions in clinical practice is: why can the effects of visual deprivation on visual acuity (amblyopia) be treated up until the age of 7 or 8, or even later in many cases of anisometropic amblyopia, but the effects on stereoscopic vision (with a few notable exceptions such as Stereo Sue; Sacks, 2006) are irreversible after the age of 2? This might be related to different critical periods for these different functions. However, the critical period for the effects of deprivation on stereopsis in animals has never been measured, primarily because the experiments are so difficult. Torsten and David decided not to tackle it, and nobody else has had the fortitude. Thus this important clinical question still does not have a basic science underpinning.

Patching the non-deprived eye as a treatment for amblyopia was suggested by Buffon in 1743, long before surgery for strabismus came along, but fell into relative disuse after that. David and Torsten's work provided some rationale for patching, but also suggested an unfortunate complication – that while the acuity in the deprived eye may improve, the acuity in the non-deprived eye may get worse. Experiments by others since then have shown that patching the non-deprived eye has to be accompanied by some binocular exposure to retain binocular vision, as well as acuity in the non-deprived eye. In work with cats, patching for 50–70% of the day works best (Mitchell, 1991), and experience with patients follows that (see Mitchell & MacKinnon, 2002; Mitchell & Sengpiel, 2009).

Torsten and David's original paper, comparing the effects of monocular and binocular deprivation, agreed with clinical experience that the effect of monocular cataract on the acuity in the deprived eye is worse, and harder to reverse, than the effect of binocular cataract (Maurer et al. 1993). It should also have been clear that binocular cataracts would affect other properties of vision, such as motion sensitivity, that are not as severely affected by monocular cataract. However, this point was not established in patients until quite recently (Ellemberg et al. 2002).

Directions since their original work

Much of the work on visual deprivation since David and Torsten's original six papers has consisted of elaborating on the points that they made. Inter alia, they pointed out that the behaviour of their cats could not be accounted for solely by the changes that they saw in primary visual cortex, and that there were also some effects on visuomotor capabilities. Since that time, work has described deficits in motion sensitivity, global form perception, positional uncertainty, and face perception in human patients (Simmers et al. 2006; see Hess & Daw, 2009), and some global motion deficits in cats (Mitchell & Sengpiel, 2009). Several of these properties depend on extrastriate cortex, and extensive deficits are seen in extrastriate cortex in amblyopes using fMRI (Barnes et al. 2001).

Perhaps the most exciting area to develop since Torsten and David's time has been the description of the molecular properties that define the critical period. Plasticity can be abolished by a long list of treatments, some representing substances that are more abundant during the critical period and diminish afterwards. The process is driven by electrical activity coming from the retinas, which activates glutamate receptors, second messengers, various kinases and phosphatases, and transcription factors, producing the proteins necessary for the decay of some synapses and the growth of others. Much of this work has been started using LTP and LTD in the hippocampus, before being transferred to the visual cortex. However, the visual cortex has one distinct advantage over the hippocampus – it has a critical period – and the timing of the start and end of this critical period can be manipulated by rearing animals in the dark (Mower, 1991).

Several factors may influence the end of the critical period, and help to explain why young animals are susceptible to visual deprivation, whereas adult animals are not. One is a reduction in activity of some of the factors involved in the plasticity pathway. Another is the relative balance of excitatory and inhibitory influences in the cortex (see Morishita & Hensch, 2008). There are also events that occur around the end of the critical period, which help to solidify the connections in place. These include (1) condensation of chondroitin sulfate proteoglycans around the soma and dendrites of the cell and (2) myelination. Treatments to prevent either of these from happening can extend the critical period (Berardi et al. 2004; McGee et al. 2005). So can environmental enrichment (Sale et al. 2007), and treatment with the antidepressant fluoxetine (Maya Vetancourt et al. 2008). Perhaps continued research along these lines will lead to a treatment for amblyopia.

Conclusion

The six papers of Torsten and David in 1963 and 1965 opened up a whole new field of research. They initiated the study of plasticity and of development in the visual cortex, and prompted studies of these topics in all other areas of the cortex. They revealed general principles of development and plasticity that have been applied in numerous other areas of the nervous system. They gave a scientific underpinning to the facts known about amblyopia, and prompted changes in treatment for this problem, which is a risk for 2–4% of the population. They provided a scientific underpinning for observations such as that it is easier to learn a language at an early age, by establishing the existence of a critical period for monocular deprivation. They also illuminated the philosophical argument about nature vs. nurture, with the answer (often true in biology as opposed to physics), that both are true. This work would have deserved a Nobel Prize, even if they had not also worked out the general principles of organization of the cerebral cortex.