 |
| Jennifer Aniston |
G. Yes, there was excitement when the "Jennifer Aniston" neuron was discovered accidentally during a treatment for epilepsy. The initial paper on it, "Invariant visual representation by single neurons in the human brain" was by Quiroga, et al. in Nature (2005). There are youTube videos as well as many articles on it. There is a Scientific American article "Building Blocks of Memory" by Quiroga and his collaborators.
The name is from the "grandmother cell" story of neuroscientist Jerry Letvin in 1969 @ MIT, and from William James in 1890. Letvin's "grandmother cell" story is about a fictional neurosurgeon with a patient who wanted to forget his impossible mother. The neurosurgeon opened up the patient's brain and simply removed several thousand neurons related to the concept of his mother, and all memories of her disappeared. The patient was elated, however, he now wanted to find the cells for his "grandmother".
 |
| Rodrigo Quian Quiroga Univ of Leicester |
When epileptic seizures cannot be controlled with medication, an alternative is surgery to remove the area where the seizures are initiated, called the epileptic focus. This requires determining accurately which small region of neurons is the problem. This can be determined with brain imaging techniques such as EEG, MRI and SPECT.
The "experienced meditator" research work at Yale in which i am a subject and collaborator is exploring the use of the high-resolution EEG technology developed for epileptic focus location. Could it watch, as is done in an fMRI, to see if the posterior cingulate cortex (PCC), a core center of the default mode network (DMN), the source of the problematic self-referential narrative, was operating or not, in "real time"? This work is covered in blogposts "What is the default mode network?...", and "Folk who meditate decrease mind wandering".
 |
| IONS EEG "Cap" "Presentiment" study |
When EEG cannot accurately determine the epileptic focus location, then microelectrodes thinner than a human hair are precisely positioned inside the skull. These continuously measure, over several days, the activity of individual neurons. As it is not painful, patients are able to do simple tasks like looking at pictures and having their brains' responses observed which is how the "Jennifer Aniston neuron" was found.
The "Jennifer Aniston neuron" fired however she was dressed, in profile, sitting, standing, walking, if you wrote or said her name - it responded. It did not respond to any other celebrity, landmark, politician, relative, etc. However, there were also "Oprah Winfrey" and "Luke Skywalker" neurons which were similarly dedicated and selective. These individual neurons were selective to a broader "sense" of Jennifer, Oprah, or Luke; more focused on a "concept" than just one image.
 |
| Yoda and Luke Skywalker w/o light sabres |
However, it got more complex when the "Luke Skywalker" neuron responded to a picture of Yoda. The "Jennifer Aniston" neuron responded to pictures of her co-star Lisa Kudrow. So this "concept" extended to higher level processing and accessing of complex relationships.
 |
| Different Visual Processing Areas |
There are several levels of visual processing of images. The primary visual cortex (V1) is where input is processed like a pixel in a digital image. More sophisticated processing is done in "high level, secondary" visual cortex areas (V2, V3, V4, V5) where one neuron can discriminate between a face and a building no matter how it is lighted.
 |
| Medial temporal lobe and hippocampus (where hippos go to school) |
These "high level" visual cortex areas send their information to the medial temporal lobe. The medial temporal lobe, particularly the hippocampus, is where the memory functions become more sophisticated; this is where the Jennifer Aniston neuron was found.
There are about a billion neurons in the medial temporal lobe. What are the chances that researchers would have found the 1 out of 1,000,000,000 neurons that was Jennifer's? There must be a larger "Jennifer" network, but how many neurons might it have?
Smart folk @ CalTech working under Cristof Koch calculated that a "concept" could be represented by 20,000 to 100,000 neurons, any of which might respond to Jennifer's pic. BTW, we remember about 10,000 concepts, so the math works.
 |
| Cristof Koch CalTech |
This network is described by Quiroga, et al. in "Sparse but Not "Grandmother-Cell" Coding in the Medial Temporal Lobe" (2008). This "sparse" network for "concept" cells is designed for rapid associations, not precise recall. It was evolved for "Is this person/animal dangerous, friendly or a potential mate?" not for "In which Star War's episode did Yoda sit beside the swamp and instruct Luke and what was the result?".
This more complex task requires deliberation and a much broader data base retrieval so a "distributed" neural network is used. A typical task would be "musical sight reading and keyboard performance". These distributed neural networks do, however, produce more errors, as they gather input from wide ranging brain regions, particularly if they involve complex memories with sounds, feelings, sights, etc. These are discussed in the blogposts "The Influence of Misinformation", "Traumatic Memories Feel True But Are Always Changing" and "How...our poor memories change the law..."
So, if we have a well-defined default mode network,
 |
| Default Mode Network Andrews-Hannah |
 |
| Default Mode Network Andrews-Hannah |
Well, as you can see, these centers are deep in the brain and "all over the place". Also, the core centers all have additional functions beyond their complex and highly-innerconnected role in the DMN. we really just don't know, yet, precisely what is doing what and if any of it can be separated out cleanly.
Also, the critical region of interest (ROI) w/in the PCC, for example, is very small, probably only a few voxels or cubic centimeters. As discussed in the blogpost "Does non-dual meditation change our "interior" and "exterior"?", there is even a great deal of variability in the location and configuration of brain centers between subjects. Any mistakes could be catastrophic and irreversible and the liability considerations staggering.
So, for the foreseeable future, we're just going to have to persist in our "Who am I?"s, and "let go, let go"s, and allow the brain to do the surgical work on "itself" precisely deactivating its regions of interest and reconfiguring the neural networks, sparse or distributed. It does do it, all by itself, if we just show it what needs to be done.
No comments:
Post a Comment