In this issue, Hwang and Ahn1 describe their success in treating amblyopia in patients we would normally regard as too old. This paper covers many recent publications that are changing our long-held notions about amblyopia.

Most current ophthalmologists' education in this area followed the neurophysiological experiments of the 1960s and 1970s begun by Hubel and Weisel, and for which they shared 50% of the Nobel Prize for medicine in 1981. These experiments in cats and monkeys suggested that the neural basis for amblyopia was a massive reduction in binocular neurons and a shift in the ocular dominance of neural activity towards the unaffected eye. Following the period of cell division and development of the visual cortex, thought to end at about 7–8 years in humans, the visual cortex became fixed and little further improvement could be made.

It is now clear, however, that some amblyopes can improve later in life. The best evidence for this comes from the work of Rahi et al,2 who studied adults in the UK who had lost their good eye. In many, the vision in the amblyopic eye improved, but the chance of this happening was related to the degree of amblyopia:

It seems that not all amblyopes are the same, and/or many cases of amblyopia are not human versions of the cats and monkeys of the neurophysiological experiments, with a finite critical period. For example, Goodyear et al,3 using functional MR imaging, found that in six early-onset amblyopes there was indeed a shift in the ocular dominance columns of the visual cortex, but in two late-onset amblyopes this was not true—the ocular dominance columns were of the same size. Davis et al4 found that there were qualitative differences in visual evoked potentials and contrast sensitivity in the early- and late-onset amblyopes, indicating differing neural mechanisms in these two groups. Levartovsky et al5 found that following treatment of amblyopia up to the age of 9 years, between 51 and 75% of patients deteriorated again.

To his eternal credit, Duke-Elder6 foresaw this difference. Although he devoted barely a couple of pages of the gigantic System of Ophthalmology, his comments are pithy. He distinguished between ‘amblyopia of arrest’ in which normal vision was never allowed to develop in the amblyopic eye, and ‘amblyopia of extinction’ in which normal vision had developed but was then inhibited. The latter came on after the age of about 4 years and was easier to treat. It seems likely that ‘amblyopia of arrest’ is indeed the human version of the animal experiments, with a shift in the neural populations producing amblyopia that is difficult to treat—due for example to a unilateral congenital cataract. ‘Amblyopia of extinction’ is probably the condition that fills orthoptists' clinics—caused by anisometropia, squint, or the combination of the two.

This recent research suggests three things: firstly that we should be more aggressive in treating amblyopia in younger children (3 years and under). Secondly, it clearly is worth while treating older patients, despite the difficulties they have with compliance. Thirdly, amblyopia treatment should not cease at 7 years. Probably it should continue throughout life, but in order to do this, we must find a treatment that is acceptable to older children, teenagers, and adults.