THE ROLE OF PATTERNED ACTIVITY IN NEURAL DEVELOPMENT
Over the past few years a set of most impressive advances have forced an entire rethinking of the mechanisms responsible for the laying down of neural connections, both before and after birth. The discovery of orientation selectivity and orientation columns in newborn monkeys had seemed to indicate a strongly genetic basis for the detailed cortical wiring. Yet it seemed unlikely that the genome could contain enough information to account for all that specificity. After birth, to be sure, the visual cortex could fine-tune its neural connections in response to the exigencies of the environment, but it was hard to imagine how any such environmental molding could account for prenatal wiring.
Our outlook has been radically changed by a heightened appreciation of the part played in development by neural activity. The strabismus experiments had pointed to the importance of the relative timing of impulses arriving at the cortex, but the first direct evidence that firing was essential to the formation of neural connections came from experiments by Michael Stryker and William Harris at Harvard Medical School in 1986. They managed to eliminate impulse activity in the optic nerves by repeatedly injecting tetrodotoxin (a neurotoxin obtained from the putter fish) into the eyes of kittens, over a period beginning at age 2 weeks and extending to an age of 6 to 8 weeks. Then the injections were discontinued and activity was allowed to resume. By the eighth week after birth the ocular dominance columns in cats normally show clear signs of segregation, but microelectrode recordings indicated that the temporary impulse blockade had prevented the segregation completely. Clearly, then, the segregation must depend on impulses coming into the cortex.
In another set of experiments Stryker and S. L. Strickland asked if it was important for the activity of impulses in optic-nerve fibers to be synchronized, within each eye and between the two eyes. They again blocked optic-nerve impulses in the eyes with tetrodotoxin, but this time they electrically stimulated the optic nerves during the entire period of blockage. In one set of kittens they stimulated both optic nerves together and in this way artificially synchronized inputs both within each eye and between the eyes. In these animals the recordings again indicated that the formation of ocular dominance columns had been completely blocked. In a second set of kittens they stimulated the two nerves alternately, shocking one nerve for 8 seconds, then the other. In this case the columns showed an exaggerated segregation similar to that obtained with artificial strabismus. The normal partial segregation of ocular dominance columns may thus be regarded as the result of two competing processes: in the first, segregation is promoted when neural activity is syn- chronized in each eye but not correlated between the eyes; and in the second, binocular innervation of neurons and merging of the two sets of columns is promoted by the synchrony between corresponding retinal areas of the two eyes that results from normal binocular vision.
Caria Shatz and her colleagues have discovered that a similar process occurs prior to birth in the lateral geniculate body. In the embryo the two optic nerves grow into the geniculate and initially spread out to occupy all the layers. At this stage many cells receive inputs from both eyes. Subsequently segregation begins, and by a few weeks before birth the two sets of nerve terminals have become confined to their proper layers. The segregation seems to depend on spontaneous activity, which has been shown by Lamberto Maffei and L. Galli-Resta to be present in fetal optic nerves, even before the rods and cones have developed. Shatz and her colleagues showed that if this normal spontaneous firing is suppressed by injecting tetrodotoxin into the eyes of the embryo, the segregation into layers is prevented.
Further evidence for the potential importance of synchrony in development of neuronal connections comes from work at Stanford by Dennis Baylor, Rachel Wong, Marcus Meister, and Caria Shatz. They used a new technique for recording simultaneously from up to one hundred retinal ganglion cells.
They found that the spontaneous firing of ganglion cells in fetal retinas tends to begin at some focus in the retina and to spread across in a wave. Each wave lasts several seconds and successive waves occur at intervals of up to a minute or so, starting at random from different locations and proceeding in random directions. The result is a strong tendency for neighboring cells to fire in synchrony, but almost certainly no tendency for cells in corresponding local areas of the two retinas to fire in synchrony. They suggest that this local retinal synchrony, plus the lack of synchrony between the two eyes, may form the basis for the layering in the lateral geniculate, the ocular dominance columns in the cortex, and the detailed topographic maps in both structures.
Shatz has summed up the idea by the slogan "Cells that fire together wire together".
This concept could explain why, as discussed in Chapter 5, nerve cells with like response properties tend to be grouped together. There the topic was the aggregation of cells with like orientation selectivities into orientation columns, and I stressed the advantages of such aggregation for economy in lengths of connections. Now we see that it may be the synchronous firing of the cells that promotes the grouping. How, then, do we account for the fact that connections responsible for orientation selectivity are already present at birth, without benefit of prior retinal stimulation by contours? Can we avoid the conclusion that these connections, at least, must be genetically determined? It seems to me that one highly speculative possibility is offered by the waves of activity that criss-cross the fetal retina: if each wave spreads out from a focus, a different focus each time, perhaps the advancing fronts of the waves supply just the oriented stimuli that are required to achieve the appropriate synchrony—an oriented moving line. One could even imagine testing such an idea by stimulating the fetal retina to produce waves that always spread out from the same focus, to see if that would lead to all cortical cells having the same orientation selectivity.
All of this reinforces the idea that activity is important if competition is to take place. This is true for the competition between the two eyes that occurs in normal development, and the abnormal competition that occurs in deprivation and strabismus. Beyond that, it is not just the presence or absence of activity that counts, but rather the patterns of synchronization of the activity.
THE BROADER IMPLICATIONS OF DEPRIVATION RESULTS
The deprivation experiments described in this chapter have shown that it is possible to produce tangible physiological and structural changes in the nervous system by distorting an animal's experience. As already emphasized, none of the procedures did direct damage to the nervous system; instead the trauma was environmental, and in each case, the punishment has more or less fit the crime. Exclude form, and cells whose normal responses are to forms become unresponsive to them. Unbalance the eyes by cutting a muscle, and the connections that normally subserve binocular interactions become disrupted. Exclude movement, or movement in some particular direction, and the cells that would have responded to these movements no longer do so.
It hardly requires a leap of the imagination to ask whether a child deprived of social contacts—left in bed all day to gaze at an orphanage ceiling—or an animal raised in isolation, as in some of Harry Harlow's experiments, may not suffer analogous, equally palpable changes in some brain region concerned with relating to other animals of the same species. To be sure, no pathologist has yet seen such changes, but even in the visually deprived cortex, without very special methods such as axon-transport labels or deoxyglucose, no changes can be seen either. When some axons retract, others advance, and the structure viewed even with an electron microscope continues to look perfectly normal. Conceivably, then, many conditions previously categorized by psychiatrists as "functional" may turn out to involve organic defects. And perhaps treatments such as psychotherapy may be a means of gaining access to these higher brain regions, just as one tries in cases ofstrabismic amblyopia to gain access to the striate cortex by specific forms of binocular eye stimulation.
Our notions of the possible implications of this type of work thus go far beyond the visual system—into neurology and much of psychiatry. Freud could have been right in attributing psychoneuroses to abnormal experiences in childhood, and considering that his training was in neurology, my guess is that he would have been delighted at the idea that such childhood experiences might produce tangible histological or histochemical changes in the real, physical brain.