Thinking Neurons?

[John Freestone]

I wonder if anyone can help enlighten me about this:
http://cbcl.mit.edu/news/files/caltech-6-05.html
...which suggests that individual neurons in the brain have been shown to 'think', or more accurately, respond to particular images by firing.

The reasoning seems quite wrong from what I've read, so I assume I must have misunderstood it or it's not very well reported. In particular, I'm thinking this: since these neuron firings were observed in living human brains responding to the particular stimulus, isn't it a logical possibility that their firing represented part of a pattern of brain activity involving many more neurons, as per more conventional theory, in which neurons act rather like switches?

In order to eliminate this possibility, wouldn't it be necessary to simultaneously measure the firing of massive numbers of neurons, or in some way sit a single neuron in front of a picture?

 

[drkitten]

I wonder if anyone can help enlighten me about this:
http://cbcl.mit.edu/news/files/caltech-6-05.html
...which suggests that individual neurons in the brain have been shown to 'think', or more accurately, respond to particular images by firing.

Oh, cool! They found the "Grandmother cell"!

The reasoning seems quite wrong from what I've read, so I assume I must have misunderstood it or it's not very well reported. In particular, I'm thinking this: since these neuron firings were observed in living human brains responding to the particular stimulus, isn't it a logical possibility that their firing represented part of a pattern of brain activity involving many more neurons, as per more conventional theory, in which neurons act rather like switches?

Yes, but switches based on what?

If you look at how the visual cortex processes data, it's basically as a series of layers. The lowest layer simply responds to, for example "light on top, dark on bottom." The next layer up responds to combinations -- for example "horizontal black band in middle of vision" or "black dot moving from left to right." The next layer up responds to combinations of combinations, while the next layer from that responds to combinations of combinations of combinations, while the next one,.... well, you get the idea.

The question is how far (and how specific) these combinations get. Is there, for example, a specific neuron in your head that responds ONLY to those particular combinations of low-level features that define your grandmother, or a cat, or Halle Berry?

The answer appears to be "yes."

 

[Dancing David]

Well certainly
http://cbcl.mit.edu/news/files/caltech-6-05.html

is a pop science article, we can look for the one in Nature but I don't see how they reached their conclusion.

 

[John Freestone]

Well certainly
http://cbcl.mit.edu/news/files/caltech-6-05.html

is a pop science article, we can look for the one in Nature but I don't see how they reached their conclusion.

Ah good, it's not just me then! First thing I caught was actually a snippet of a radio broadcast, and some guy played what sounded like a slowed-down recording of the firing of a neuron, and waxing lyrical about how 'we' used to think that a neuron just fired, like a digital thing, but this sounds more like birdsong or whale music (or some such banal tripe), and this supposedly supported a new theory that individual brain cells 'thought'.

I guess if you slowed down a recording of a transistor switching, it would possibly make some weird noises. I assumed there must be more rigorous science behind this, but maybe not. I'll have to do some more searching. Thanks.

 

[lupus_in_fabula]

This brings up more some interesting questions test wise. So, when a single neuron responds to the sting of letters: J-E-N-N-I-F-E-R-A-N-N-I-S-T-O-N, which can be recorded, then the following test setup comes to mind:

J-E-N | … [record]

J-E-N-N-I-F-E-R | … [record]

J-E-N-N-I-F-E-R | A… [record]

J-E-N-N-I-F-E-R | L… [record]

J-E-N-N-I-F-E-R | A-N… [record]

J-E-N-N-I-F-E-R | L-O… [record]

J-E-N-N-I-F-E-R | A-N-N… [record]

J-E-N-N-I-F-E-R | L-O-P… [record]

Wouldn’t it be plausible to suspect that the neuron firing becomes more specialized when more letters after J-e-n-n-i-f-e-r are being shown since it becomes increasingly clearer (i.e. narrowing down other possibilities) for what the string of letters represent?

 

[John Freestone]

The only reports I found were journalistic interpretations, so there are a great many gaps and questions that could be asked. I haven't had time to search for more information yet. I don't know if the 'string of letters' representing someone's name would have been presented at once in the visual field or serially. A single neuron responding to 'JENNIFER ANNISTON"

 

[John Freestone]

The only reports I found were journalistic interpretations, so there are a great many gaps and questions that could be asked. I haven't had time to search for more information yet. Did you take it that the 'string of letters' was presented serially? A single neuron responding to 'JENNIFER ANNISTON" flashed on a screen is a little more plausible than it waiting and thinking "Oooh, I think I know who this might be!", but either way I just can't understand how the response of a single neuron can be analysed in practice in the living human brain, when it is linked to so many others in complicated reciprocal patterns of influence and physical connection, nor how this could lead anyone to believe it was the site of recognition of a particular name, nor conclude that brain cells think. Sorry, I'm repeating myself. Maybe the bumping will attract someone who knows what all this is about and I can be lazy and not search for original scientific papers - I probably won't understand them anyway!

 

[quarky]

What about studying single celled organisms, and their response to stimuli?