Sleep and learning
Wherever there is consciousness, there is also unconsciousness. Of the latter, there are two types, the exceptional type (eg due to anaesthesia or disease) and the routine kind that we call 'sleep'. TDE/GOLEM Theory claims that the evidence of its teleology [1] is unequivocal and overwhelming- the purpose of sleep is to consolidate temporary, short-term memories into persistent long-term ones.
STM encoded by autolatching neurons
As we move through the situations that make up our lives each day, we maintain the short-term memories that characterise the 'spectrum' of each situation. These STM's consist physically of situation-semantic neurons in a 'latched' state. The purpose of the latching is so we can maintain items at 'front of memory' without needing to constantly attend to them. After all, attention is one of the most expensive of all our mental resources, and should be conserved where possible.
unlatching frees STM for next situation
When we move to new situations, eg when we leave home in the morning and go to our workplace, we must 'unlatch' the STM neurons, in order to be able to 'latch' others, the ones that characterize the 'next' situation. This unlatching of each neuron's excitatory state requires the action of inhibitory (endorphinergic) neurons, which are drawn from a strictly limited pool located in the Ascending branch of the Reticular Activation System (ARAS). This pool contains inhibitory neurons [2] which use Serotonin (ST a.k.a. 5HT) as their primary neurotransmitter.
As more and more endorphinergic (EP) neurons are needed to unlatch situational neurons, more and more of the serotonergic neurons in the 'pool' become 'exposed', causing increased inhibition of the subject's ability to generate states of arousal such as the 'spotlight of attention'. Consequently, the subject finds it increasingly difficult to maintain attentional focus specifically and task motivation generally, a homeostatic effect called 'sleep pressure' -see figures 7.1(b) and 7.1(c) below.
(a) (b) (c)
Figure 8.1
LTM consolidation
= unlatched ARAS features copied into DRAS sequences
The ARAS is responsible for unlatching STM neurons from all of the situations that the subject has encountered, using inhibitory links drawn from a fixed quantity of mental arousal. If this happens repeatedly each day, the brain will soon run out of inhibitory links [4]. Also, the STM semantic states corresponding to the interesting or relevant features of each separate situation will be forgotten. Therefore, brains of all animals, even drosophila (fruitfly) and C. Elegans (nematode) have evolved sleep, an off-line state in which-
(i) the inhibitory 'sensor' links 'loaned out' each day are 'paid back', thereby reducing sleep pressure and restoring an alert state upon waking.
(ii) before the sensor links are returned to the ST pool, the pattern of unlatched features that they recorded is copied by identical sequences of inhibitory links, typically dopamine (DA), which are part of the motor learning system. Whereas the total amount of sensory arousal available to the conscious mind is limited, the number of learned motor patterns is effectively unlimited. This is how long term memories (LTM) are formed.
If the ascending branch of the RAS (the ARAS, whose outputs go to level-2) in vertebrates is damaged, the subject cannot be wakened, hence the ARAS causes consciousness. If the descending branch of the RAS (the DRAS, whose outputs go to level-1) is damaged, the subject cannot sleep, hence the DRAS causes sleep.
(a) (b)
(c)
Figure 8.2
Sleep is not a simple process. Since there are three brain levels 1-3 at which adaptation changes occur in each individual human (the fourth level-0 is the instinctual, or 'unconditioned' level), the restoration and memorisation of unlatched STM feature links involves three separate phases. Furthermore, the brain levels are not functionally independent. The changes in higher order levels are built upon the adaptational changes in lower order ones. This is hardly a mystery, it reflects a truth about all types of construction, where foundational footings must be laid before laying the bricks which make the walls.
DRAS must also command the ARAS to maintain the 'standby' readiness of the software layers responsible for conscious arousal functions. In order to adjust effector subsystem (level-1) defaults, in response to accumulated daily errors, the hardware and software layers must be separately switchable. The hardware layer is switched off during REM sleep (a process called 'sleep paralysis') to allow hardware 'adjustments', while the (3 or 4, depending on who you believe) software layers are switched off in order, then switched back on in reverse order, during non-REM sleep episodes N1..N3..N1
Consider the uppermost part of figure 8.3 below. These are the well-known facts about the RAS (both branches), reproduced from the McGill University website. The facts clearly demonstrate the validity of the basic GOLEM ‘latched neuron‘ model. The DRAS maintains somatosensorimotor (level-1) subsystems- muscle tone and posture (level-1 efference) [6] while the ARAS maintains visuospatiotemporal (level-2) subsystems-alertness, wakefulness and learning (level-2 afference). The RAS can only service these subsystems if the pattern of connections to each neuron are functionally consistent with those described in section 6 (ie the ‘a’ thru ‘g’ labelled inputs). It is impossible in practice to empirically verify (ie ‘prove’) the veracity of multivariate systems (what would the null hypothesis be?). BUT if the goal is to propose a hypothetical multivariate system that has validity, ie is internally consistent (ie the ‘bits’ all fit together and do what they are supposed to do), then by Occam’s Razor, we choose the simplest (but no simpler) valid solution, ie GOLEM.
(a) (b) (c)
Figure 8.3
(a) (b) (c)
Figure 8.4
The lower diagram of figure 8.3 and figures 8.4(a),(b) & (c) are depictions of the entire human ‘governance’ system, which includes brain, mind and self. unfortunately, we don’t really have an ideal term to describe this ‘gestalt’ concept, all we can do is talk about its constituents-
-the brain, ‘wetware’ or ‘hardware’ when we discuss level-1 issues,
-talk about the ‘mind’ for level-2 matters,
- refer to ‘subject’ or ‘self’ at level-3.
Hence, diagrams become the best available method of unambiguous description. This doesn’t suit some scientists who believe text is best. I hope this discussion shows the limitations of words when talking about models of very complex systems.
1. the problem with this statement is not that sleep scientists disagree with it- far from it, most of them would put 'long term memory consolidation' as one of the main effects of sleep. No, the problem is that most of them don't believe in teleology! This position is as ridiculous as denying that the sole purpose of the heart is to pump blood. The issue they have with with ascribing 'purpose' to bottom-up processes like (presumably) evolution. For purpose, they say, you need a plan. Hello, what about our DNA? Isn't that a top-down plan for the body? Doesn't it contain a subordinate plan for building the heart? They confuse evolution, which is an optimisation algorithm, and is patently 'bottom-up', with genetics, which is a just as patently a 'top-down' construction plan. 'Purpose' and 'teleology' are just words, at the end of the day, which never have perfect denotational semantics. Purpose as a word was invented to describe the intended use for things people make, eg to stop someone using a rare and valuable bronze bust to drive a rusty nail into a lump of wood. "Its not a hammer, you fool, its art!"
2. GOLEM theory suggests that the brain regards states of arousal as a kind of pain to be passivated, hence the body 'treats' them with EP links, just as it does at injury sites, using EP as the remedial neurotransmitter. Even in the injury situation, the purpose of using temporary passivation is the same- to free up attention for other things, in the case of injury, freeing it up for the important task of helping the animal (or human) return to a safe place, one where rest and healing are possible.
3. the usual acronym used is 5HT (5-HydroxyTryptamine), but ST is used here for consistency with others like EP & DA.
4. Arousal is a strictly limited resource, and this is reflected in the neurotransmitter 'ledger', whose loans and returns of inhibitory links must be kept balanced by the Reticular Activation System (RAS)
5. Although not strictly part of this stage of the problem solving routine, we can perhaps speculate on what extra features if any are needed for autolatching. Perhaps C++ ion transfer? Mg++? The answer is not needed for GOLEM construction, just for my curiosity.
6. It controls gamma motoneurons- the ones that take up muscle 'slack' so stretch receptors respond immediately to loads.