I've been thinking about this and I've come up with some ideas.
Rtaher than seperate brains and muscular systems, as we have now, its easier to conceive of one common system from which they both evolved, which they selected from to specialise in aspects, and secondly, to conceive of this as an organisational system and motive system in individual stem-like cells.
The early eukaryotic animal formed as a collective that needed to decide where to go.
In individual cells, it is demonstrated they can sense their environment (non-chemically) and adapt in numerous ways, as well as choose where to go. They have various well known motive systems which must be the molecular motors and structures co-oped by myocytes to make muscle millions of generations later.
At the same time, individual parts of cells need to be coordinated or enlse all the cell can do is sprawl and tear itself apart. Coordinating means in practice turning off all local responses which are not coordinated to the particular desired movement.
What I believe is that the nervous system started to evolve in the very first animals in which the group has to coordinate local members to for example wriggle or swim in a particular direction. This proto-brain is demonstrated in certain slime moulds.
This haracteristic I believe originates as a cell coordination system to allow local parts of cells i.e. individual beating hairs, to work together to the cells motile advantage. These motor systems can be coopted by the cell to allow groups to wriggle together, but the cell already needs an organisation system to gather environmental data, select a response, and coordinate local cell activities towards this goal, as a free living creature / stem cell / stage, and as an eukarytotic ancestor.
So here's asimple explanation of what I think allowed for such simple organisation and behavioral development in eukaryotic colonies of cells;
1 - the cell sensing system is distributed in the cytoplasm and membrane, and is developed early in evolution of prokaryotes as a natural benefit.
2- the early cell has to evolve a way that any cell structure that grows, needs to be coordinated automatically. The logical solution is to design cell bodies that intrinsically distribute environmental data to other parts of the cell - thereby automatically taking environmental data and efficiently and quickly distributing it to all functioning parts of the cytoplasm and motor system. This system is an inbuilt property of the various cell 'regions' involved in movement.
3 - this natural system could evolve to include a kind of amplification, as computer networks do over larger distances
4 - the system distributes information, but it has an effectively spare bandwidth - its distributing all the information that could be needed to all nodes from wherever - it doesn't even realise what its distributing
5 - local nodes only have to 'pick up' relevant information to which they are tuned
6 - organisation of sensor nodes can pick up universally distributed information, and based on confluence of signals, some can thus competitively switch off other cell nodes, and using this, activate coordinated responses in different regions of the cell body (made of compartmentalised functions and a distributed motor and metabolic system).
Now, this system I believe occurs because, as part of cell organisation, it has had to slowly evolve number 1. This has led to a 'high bandwidth' communications 'pipe' of some description, that grows with the cell, and provides all nodes that wish to tune in around this network, with all the information they could want. I believe this is an electro-magnetic system, so this is a bit like sending envronmental data (chemical-light interactions from background sources perhaps) through something like fibre optics, that are only analagous but not literally present, the system being more like a 'smart jelly' made of repeating structures that under buiologically genertated circumstances conveys certain em information bands around the cell, much more than it could ever initially require.
This allows all nodes in the network to 'tune in' and it allows interaction with this high-bandwidth system to facilitate group control over parts of the network, once an organisational and input-heirachy, has created the dominant response pattern. There be local arrangements of nodes and logic orientated structural memory, but it effectively extends continuously across each cell, and between them. Movement of ions in tubes physically connecting nerves may be a vital component of the system.
The key properties of this system are what allows brains to be built so easily. This occurs from groups of these cells simply extending themselves into each other by tubes that create a 'networked cytoplasm'. This is observed in brain cells and is thought to be an underestimated (thanks to theoretical fashions) important feature of neural networks. The early network easily forms in earlier cell structures. This cell structure retains it for forming a moving colony. Coordinating taht movement would be a naturally self organising property of the group, if each cell has a 'distributed sensing system' so that all nodes can get information from all parts of the network because the bandwidth of that cuytoplasm is conducive to do so. From taht point, a networked cytoplasm should share repeating, scalable organisational rules that take this sense and allow individual nodes, in the organisational heirachy described, to communicate back through modifications to that distributed signals, to all the nodes in the network, which in turn collapse out a dominant pattern (strongest pattern) which becomes the group response. This strongest pattern would naturaally be improved by natural selection.
Thus, a brain starts to form as a natural property of any cell network with group motion. The muscular aspect of the neuro-motor system, from which all brains evolved, would be through the innaate cell motor system being overexpressed in some cells that are less fit (by location) to make communicative networks, and thereby evolve towards specialised motive prowess, whilst neurons select the other way. Initially, the motive system would be part of the cell community, as would the 'brain', all present at the same time and tending to varying degress of expression.
Because of this high bandwidth distribution of information, naturally arising in the cell membrane and cytoplasm, and because of selective pressures to use it, a number of cells can link up beyond there initial size and organisation limits. The cytoplasm is seen as a self organising entity only maintained by ribosomes and the cell nucleus.
With this communications / distribution system (of unknown exact form), selection of still greater bandwidths may be a straightforward structural matter in each cell, if needed, and this allows billions and billions to be hooked together effectively, all making group calculations in a massively parallel way, using local organisational systems, leading to big brains and all their amazing experiences. The communication system 'leaks' out data flowing arounf to the local parts of each cell, which are selected for it so that that part of the cell can make what ever cell-based response is necessary. As the family of nerve cells has evolved, it has become able to divert this local response into some kind of local memory once related to organisational matters concerning local cell processes relevant to environmental sensing. This is part of the original cell organisational system, cooped again, and which facilitates use of this distributed data. I suggest the high 'bandwidth' needed to make such a simple series of steps evolve into something like what we have, is accomplished by virtue that what is being sent around this network, actively and passively, is only accessible by computational and output parts of the node (that is, the parts where energy leaks into molecular structures and thereby cell interactions and events, or subsequent interactions to that field (thereby, can act as a logic system that influences the responses of other nodes). The elements that effect the ability of nodes to tune in to relevant data, would be based on exact energy levels and of course, the quantum mechanical properties of active structures in those molecules to allow for energy transfer to occur in standard packets. The quantum aspect is that because this aspect of the bandwidth is not accessed by any thing else, each node consumes little energy and so the field effect cannot easily diminish, thus distance and I believe bandwidth (rather by which I mean distribution), as it were, is high. If energy cannot leak except to 'resonnant structures' it cannot decay - and such fields extend much further than they should, explaining potentially for example how evanescent fields go much further than magnetism should normally allow - and evanesent fields and selective access to the energy locked within, is room temperature large distance quantum field effect close to commercialistion, and a case study in how space finds ways to convey energy differently if it is unable to dissipate down normal entropic paths. It is I think the kind of way to talk about communicational systems, able to interact at nodes distributed in 3 dimensions, at small and larger scales, that could operate at much lower magnetic fields than could normally allow for such transmission (i.e. as in life), and which culd result in electron energy level changes which also could influence theoretically the form and activity of molecules.
The system may naturally arise, and result from natural metabolic and escpecially motor functions in the cell. Groups of cells would likely coordinate firing of local metabolic paths as part of developing group motion and such behavior seems to be present across comlonies of single celled organisms. This motor aspect may also encourage such signal distribution in a proto-animal, allowing better group coordination through sensing and timing.
This would already have to be present inside the cell to be extebded in a network hypothesis, but it ties potentially motion along with the sensing / motor pre-nerve capacirty of the network system. The metabolic process i.e. large scale movement of electrons or protons, leads towards generation of fields that out to be structurally entrained in organelles and other cell structures. Such a signal distribution capacity in the cytoplasm, perhaps through hydrogen(water)-protein interactions, would then allow other systems to automatically pick up any energy distributed in it, as they will be constructed of identical units. The wider organisation of these nodes, constructed by the cell as structures connected to the nodes (the nodes may be highly distributed in of themselves) and through node-signal translation to electron energy levels, naturally through their increasing complexity, in turn determine what particular nodes may be interacted with across the common field and under what circumstances - and this is what allows for higher level organisation of many nodes. When they do 'communicate' they automatically transfer energy to local enzymes and proteins or RNA structures in corresponding nodes, allowing for decisions whether to strengthen the connection (resonnance) or change it.
A simpler analogy may be to conceive of light, say IR frequencies to which water is naturally translucent, beaming about the cell as it would. The cytoplasm doesn't block it, because nothing within it can interact with that frequency. But certain 'nodes' are present witha special dye that absorbs that frequeny very well. According to standard physics, they can only absorb that light frequency if the molecule absorbs the energy in some way. Thus it is locally distributed into biomolecular proccesses i.e. in activity of an enzyme. But at the same time, to determine a means for each distributed node to communicate back to the group and an order to energe, it would seem necessary for each node to be able, in the molecular decay of some system, to luminescence in that or similar frequency that can also be sensed by the group. To ensure bandwidth, a 'quantum' system is proposed simple because by tuning out of other groups, those who are on particular frequencies, potentially unique, are not able to lose energy to anything not appropriately dyed. But quantum mechanics also allows for energy to find wells to flow into, and thus, the light that was generated, finds its targets through distributed field effects. The homologous biological structure and synchronised decay would allow for both ends to be so tuned, if such a field effect could survive the heat of the cell. There are various ways several scientists have proposed that that could happen.