The function of sleep

One of the good things of working at RIKEN is that the Brain Science Institute (BSI) is just across the road, and they have an appealing lecture series. Today there was a presentation by Giulio Tononi about why animals sleep. In other words: sleep must have a biological function, and the question what that function is, still remains unsolved. He quickly discards older ideas that for instance it is to save energy that cannot be spent usefully during the night, and tells that it must have to do with handling the experiences obtained during the day.

During sleep your brain activity differs from the waking state, and with current technologies such as  functional MRI this can be pinpointed in ever greater detail. One observation is that, during the non-REM part of sleep, neurons in parts of the brain will cease firing completely for a short period of time (milliseconds), and this happens about 1000 times per night.

The nervous system and especially the brain consists of neurons that are connected to each other, the connections are called synapses. If the synaptic link is strong, the chance that the next neuron will fire if the previous one does, is greater. The strength of the synaptic link can vary. It is quite uncontested that by using a particular neuron (for instance by hearing a sound or performing a manual action), the synapses involved are strengthened (potentiated in the jargon).

There is cost associated with having strong synaptic links, for instance they consume more energy and space. Therefore, if some synapses are strengthened by new experiences, that must be balanced by weakening others, which is called renormalization. Tononi’s hypothesis is the following (if I understood correctly): during being awake, the synapses that you have used just now (on that day) are the most excited ones. If you were to renormalize there and then, the experiences of that day would be represented too strongly, so that you will forget other memories too soon. Instead, during sleep you are disconnected from the environment, and you can ‘sample’ all your experiences, past and present, to renormalize your synapses more effectively (whether this sampling is dreaming cannot be determined). Here the being disconnected, hence sleeping, is essential.

Obviously this is not my area of expertise, and I cannot judge how controversial this hypothesis is. It makes sense. In any case it’s very interesting, and the progress on this topic is impressive.