Every so often I’m overcome to speculate wildly beyond the limits of my own understanding. This is one of those occasions. Just saying…
A while ago @DrDav tweeted me an image she’d taken of a New Scientist article. The item concerned some experiments done on a guy whilst he was undergoing an operation to help deal with his epilepsy. (This is fairly normal practice.) Electrodes were inserted into particular regions of his brain and, presumably, a small current was applied to stimulate those regions. In one location, the stimulation resulted in the man hallucinating that he was in his family’s pizzeria.
A much older New Scientist article (which I’ve lost the reference for) related an experiment on mice where the formation of proteins in certain neurons was correlated with the mouse learning a new route through the maze.
Alongside these there is the discussion of “the Halle Berry neuron”..!
The metaphorical drift of these news stories is that the memory is stored in the neuron maybe even, somehow, in the protein… I’m really struggling with this. More particularly, I’m struggling to imagine how a protein can encode the richness of a memory.
Take any memory.. this one for instance: @DrDav and our first child barely a day old.
The memory of this moment is overwhelmingly emotional but contains other elements too, primarily visual, and links to other memories. (The phrase “couveuse afdeling” springs to mind, a long corridor and a new-born baby screaming as he’s connected to an intravenous drip.) Each of these memories is multi- (or supra-) modal.
Now the question remains, how do our brains deal with these supra-modal memories. Option 1 – the memory is stored in one neuron but that neuron has links to the different parts of the brain that mediate the different aspects of experience that are needed to deal with that memory. Option 2 – the memory is distributed across the brain in the relevant regions; in each of those regions there are neurons involved in this memory and those neurons are linked together in a network. Now, the material I’ve come across tends to suggest the latter. (The idea that “neurons that fire together wire together” and that coordination of neurons around a particular activity is managed by synchronous firing of neurons both support this.)
Option 2 offers up an alternative to the ‘memory in a neuron’ idea above. That is that the neuron should be seen more like a junction box. That is, the neuron is actually linking together with other neurons across the brain and that the memory is (somehow) distributed across the network. When one neuron has been stimulated, the network is stimulated. The protein formation observed in the mice may then relate to the formation or strengthening of a particular network.
The other problem with the memory in a neuron option is that you end up needing a vast number of neurons to deal with the vast number of concepts and memories that we deal with. In contrast, and to speculate even further, the junction box / network option requires fewer neurons because a single neuron might conceivably be involved in a wide variety of concepts. In this case those newly formed proteins might be less the memory than a part of the switching mechanism in the junction box. Initially, this train of thought led me to wonder whether the network structures might map onto our conscious taxonomies of concepts. Then, thinking about the ‘scaffolded mind hypothesis’ I noted in the previous blog, I began instead to wonder whether the networks might be built ground-up as it were: We start with raw experience and simple associations and the networks grow, or are strengthened, as our experience (and our ability to reflect on it) grows.
This still doesn’t explain the emergence of the subjective quality of memories from electro-chemical interactions but it changes the question about the role of the neuron. But how to test the speculation? Another while ago I came across diffusion tensor imaging (It is a magnetic resonance imaging technique that detects the flow of molecules, including water, in the brain. The information can be used to map connections between regions in the brain.) I’ve not really had time to look into it in detail and find out the resolution that it works at but it would be interesting to see whether you could use this approach to look for the networks associated with particular concepts. Anyway, that’s quite enough speculating for one day.
(P.S. I’ve decided that I don’t want to do research – I want my own research institute! One dedicated to coordinating the multifarious aspects of studying object handling. There’s too much for my poor, lone brain to manage.)