Should the Machine have Non-REM Sleep?

Posted: June 30, 2011 at 3:50 pm

One of the points of discussion with the philosophers was whether the dreaming machine will have analogues of all the characteristics of human sleep. Some of these don’t make sense to include, as in the alteration of self-consciousness (as the machine has no consciousness), but one in particular, the stages of sleep ranging from REM sleep to Slow-Wave-Sleep, could be relevant.

While dreams occur in both REM and non-REM sleep (albeit with divergent characteristics), the former is more correlated with what we normally think of as dreaming, and the latter correlates with the recuperative aspects of sleep. If we accept the link between NREM and recuperative sleep, there is little sense in it being included in the system, as the machine has no biological (mechanical/electrical) need for recuperation.

From the standpoint of an EEG, SWS is characterized by high amplitude, low frequency signals. This corresponds to a high degree of synchrony in neuron firing (many neurons fire nearly simultaneously). On the other hand, REM sleep is characterized by low frequency and high amplitude signals that indicate neuron asynchrony.

In the current conception of the system, REM sleep has a functional role: to maintain links between the “neurons” that represent visual prototypes. In humans, REM sleep does not occur first, rather sleep is initiated with a increasing depth of sleep stages from waking to SWS. After this point, activity oscillates between stages, moving between periods of REM and SWS. It may not be consistent to have the dreaming machine fall asleep and directly enter a REM stage.

Lets say that during REM sleep the asynchrony is due to particular neurons being activated, as they are during waking. During SWS large regions of neurons are all activated at the same time. The function of REM sleep, for the machine, is to strengthen existing neural links, which would likely involve the activation of particular neurons. If REM sleep strengthens connections, then what process “prunes” weak connections?

Perhaps the initial broad and asynchronous activation of the brain during the first bought of SWS has the effect of inhibiting the whole network and weakening all the connections equally. This would remove all weak links, and leave only strong links. The subsequent REM stage could be the activation of particular regions of the brain, bringing the remaining connections back up to a higher level.

There are issues with this idea, and it has not been considered in relation to empirical evidence. Asynchronous activation of the brain would not be required to strengthen the remaining links, this could be easily accomplished through SWS-like synchronous activation. Additionally, its not clear how the overall synchronous activation would lead to the weakening of neuron links throughout the network.