Neural Resonance

Similar events, to ideas, to resonance

A brain first notices an event, a piece of data, keeps for a certain time. Another piece of data comes along, another event, a different ‘thing’ but something is fishy. Something is similar about these events, a connection is made.

The events are now stored in the mind, something analogous to a node with many complicated branches stemming out radially, with sub branches and sub nodes. The two ‘things’ are connected at some branches, many branches left unconnected.

Another thing happens, again there is a feeling of similarity, and is loaded up. There is something about this new ‘thing’ that is similar to the last two things. Again, in our analogous space, a big, complicated and many branched thing (its complexity and number of arms relating to how much about the thing is still remembered) is connected to the other two.

While there are about as many connections (similarities) between 3 and 1 (and 3 and 2) as there were between 1 and 2 originally, the amount of branches connecting all 3 (rings of connected sub nodes that do not include the core node) representing facets all three things have in common, is now lower.

After repeated added events/data/things, this process continues. More things, more connections (many more if we aren’t pruning any branches as we go along), but the number of common connections between all things is reducing to a mathematical limit ie. what connects all these things together.

If we are to try to select for these core connections, we may suggest a model where a branch’s survival over time is proportional, or at least has a positive relationship, to the number of connections it has. We can imagine a branch something like a muscle, where use leads to growth, or like a footpath through a meadow, that requires regular travelers to maintain. Like the footpath, lower connecting routes lead to fewer travelers, fewer footsteps, and eventual overgrowth if neglected for long enough. It also seems apt that a branch can effectively ‘prove its worth’ if used enough for long enough, crystallizing, like a meadow path eventually stomped until it is no longer suitable for grass.

This mechanism should prune the connections in our group over time so that it eventually becomes little more than the common branches between all of its former things, with the extraneous branches removed over time. This starts to resemble less a bundle of things, and more something in its own right. A mechanism, or common idea, that linked its predecessors.

Until now we have considered an analog space, where processed events, memories, and experiences, and resulting ideas, can be moved around with ease, and positioned next to each other. In a brain made of cells, it seems that we need a new model. How can a new ‘thing’ be found to be connected to another already in the brain? It can’t connect directly to every other idea in the brain simultaneously — there isn’t the space in mere three dimensions. When we feel something we are seeing is similar to, or can be explained by, something we’ve seen before — something specific in a vast number of previously processed data — how does this work?

Imagine a room filled with thousands of glasses of water of all shapes and sizes, each filled with water to varying heights. An opera singer stands in front of them all, and sings a note. When she stops, she hears a faint ringing, like an echo but not quite. She hears her note played back to her by a couple of the glasses. Each glass has a resonant frequency, decided by its shape, thickness, and amount of water. All glasses will vibrate in response to sound, however if the sound is at its resonant frequency, the vibration will continue to add to itself in the glass. A glass will resonate best at exactly that frequency, and will show some lesser response to frequencies near it.

Now imagine instead of sound traveling through the air, it travels via a cup and string. The singer sings into a cup with strings connected to all other glasses. Now imagine that carrying frequencies frequently improves a string’s ability to carry more (strengthening the connection). Glasses whose frequency is lower than our singer’s register never resonate, or return frequencies back along the string. Their string withers, and eventually falls apart.

What if a pruned nest of connections, like our idea, exhibited a resonance to a frequency — or more likely, a complex waveform. What if a new stimulus or piece of data could be connected with this idea efficiently, by sending its waveform out to all ‘ideas’ and receiving a response from the few that it resonated with, with an amplitude proportional to how close the waveform was.

Neurons tend to carry signals unidirectionally, unlike our cup and string. This would allow an idea to have different ‘incoming’ and ‘outgoing’ resonant waveforms, and for idea one to trigger idea two, which then triggers idea three, without idea one directly triggering idea three. Otherwise the complexity of connections and pathways would be very limited (a certain waveform/frequency would resonate, and then slowly diminish, with no pathway to another new frequency).

What happens over time? There is no cost function as such, or back propagation. Structures form according to how well they resonate with incoming data, and later with incoming data as well as each other.

Idea: incorporate Myers–Briggs into the system — sensory vs. intuitive.

Sensory: hoovers up and keeps details at the expense of forming large connecting structures. Furry, spiky clusters.

Intuitive: readily discards details in service of forming large, lean connecting structures. Long-limbed networks, structure at a bird’s-eye view, bare looking up close.