Chapter I
Soups
Neuroanatomy had shown that neurons are separate cells, and electrophysiology that they signal with electrical pulses. How did a pulse cross the gap? In 1904 Thomas Elliott, a Cambridge student, noticed that adrenaline mimics the effects of certain nerves and suggested that the nerves might release it. Few followed up.
The decisive experiment, by his own account, came to Otto Loewi in a dream on the night before Easter Sunday 1921. He woke, scribbled a note, and in the morning could not read it. The next night the dream returned, and he went straight to his laboratory in Graz. He stimulated the vagus nerve of a frog's heart, which slowed it, then moved the fluid bathing that heart to a second heart. The second heart slowed as well. The nerve had released a substance. It proved to be acetylcholine, and in 1936 Henry Dale in London showed that it also carries the signal from nerve to skeletal muscle. Loewi and Dale shared the Nobel prize that year. Two years later Loewi, who was Jewish, was arrested after the Nazi annexation of Austria and released only after handing over his prize money.
Chapter II
And Sparks
The fastest synapses act within a millisecond, and many physiologists doubted that chemistry could be so quick. The ablest of them, John Eccles, argued through the 1940s that synapses in the brain and spinal cord are electrical. In Dunedin the philosopher Karl Popper encouraged him to state his theory so that an experiment could refute it. In 1951 Eccles and his colleagues pushed fine glass electrodes inside motor neurons in the spinal cord of cats. Inhibitory signals made the inside more negative, which his electrical theory could not produce. Eccles announced that he had been wrong and became one of the chief investigators of chemical synapses. He shared the 1963 Nobel prize with Hodgkin and Huxley.
Sparks were not wholly extinguished. In 1957 Edwin Furshpan and David Potter found an electrical synapse in the crayfish, where current passes directly through channels joining two cells. Such synapses exist in mammals too, but chemical synapses are the rule.
Chapter III
Packets
At University College London, Bernard Katz, a refugee from Leipzig, studied the junction between nerve and muscle in frogs. In 1952 he and Paul Fatt saw that even when the nerve was silent, the muscle showed small random blips of about half a millivolt. With José del Castillo he lowered the calcium in the bathing fluid so that each nerve impulse released only a little transmitter. The responses then jumped in steps the size of a blip. Transmitter was released in fixed packets, and at the same time electron microscopists were seeing nerve endings crammed with tiny vesicles, each about 50 nanometres across, which proved to be the packets.
Acetylcholine was soon joined by others. Arvid Carlsson showed in 1957 and 1958 that dopamine is a transmitter in its own right, concentrated in the basal ganglia, and Oleh Hornykiewicz found it depleted in Parkinson's disease. Glutamate turned out to be the main excitatory transmitter of the brain and GABA the main inhibitory one. In 1970 Jean-Pierre Changeux used a snake toxin to extract the acetylcholine receptor, among the first receptors to be isolated. It was a protein with a channel through its middle, and the chemistry of biochemistry had reached the synapse.
Chapter IV
A Closer Look: Counting Packets
If transmitter comes in packets, the response to each impulse should be a whole number of units, and the number should vary by chance. Katz's model was that a nerve ending holds many packets, each released with a small probability. The number released per impulse then follows a Poisson distribution with mean , the quantal content:
This makes a prediction that can be checked two ways. Take an illustrative low-calcium experiment with 400 impulses. Spontaneous blips average 0.40 mV, and responses to impulses average 0.92 mV. Dividing gives
The Poisson formula then predicts how often an impulse releases nothing at all, , or about 40 failures in 400. Counting the failures directly gives a second, independent estimate, . Del Castillo and Katz found that the two estimates agreed in their experiments. The whole distribution can be predicted too:
| Packets released | Expected impulses (of 400) | Response size |
|---|---|---|
| 0 | 40.1 | 0 mV |
| 1 | 92.2 | 0.4 mV |
| 2 | 106.1 | 0.8 mV |
| 3 | 81.3 | 1.2 mV |
| 4 | 46.8 | 1.6 mV |
| 5 | 21.5 | 2.0 mV |
| 6 | 8.2 | 2.4 mV |
| 7 or more | 3.7 | 2.8 mV and up |
A histogram of measured responses showed peaks at multiples of 0.4 mV, in roughly these proportions. In normal calcium the same junction releases a hundred packets or more per impulse, and the chance of a failure is about , which is effectively zero. That is why a nerve impulse reliably makes a muscle contract, while many synapses in the brain, which release only a few packets, fail often and by design.
Chapter V
Synapses That Learn
A synapse is not a fixed wire. Eric Kandel chose the sea slug Aplysia because its neurons are few and large, and showed in 1970 that when the slug learns to ignore a harmless touch, the synapse from sensory to motor neuron weakens. Later he found that lasting memories need new proteins and changes in which genes are active. In 1973 Tim Bliss and Terje Lømo found that a brief burst of stimulation strengthens synapses in the rabbit hippocampus for hours, and later work showed it can last for weeks. This long-term potentiation became the leading model of how memories are laid down. Carlsson, Paul Greengard and Kandel shared the 2000 Nobel prize.
The rule that a synapse strengthens when the cells on both sides are active together had been guessed in 1949 by Donald Hebb, and it became a founding idea of computational neuroscience. Where a particular memory resides, and how it lasts a lifetime when every molecule of a synapse is replaced within weeks, is still not known.