These are one of the cells that are most atrophied in the schizophrenic brain. Other types are lost in elliptic tissue. While an acceptable first abstraction, they are still very complicated. You have a couple hundred billion of these constantly computing at the millisecond time scale. How are you going to deal with the really difficult read out of the brain over some reasonable time scale? This highlights one of the key success stories of the 20th century. From ecology. Antidepressants, antielleptics. These bathe the entire brain in these drugs. This addresses the circuits that matter and the circuits that don't matter to some pathology. Perturbing neural activity matters.

Here's my neuromodulation theory slide. Here are three neurons that are connected. You can see activity propagating about. Neurons are usually negatively charged with respect to the world. You bring the voltage up when they are active. When one neuron sends information across this gap, you have chemicals being transmitted. It causes polarization of the next neuron. You have a channel and a membrane. THere are postiively charged ions outside. When a neurotransmitter arrives, you can get those channels to open. The charge flows in and you get a spike. Of course, the transmitter usually goes away and it goes back to its resting state. Can we program this pattern of activity by very specific cell activity? Can we generate computation, behavior, or one day a subjective experience? We can try to put that information into the down stream cells. One of the ideas that we have been working heavily on is whether or not we could use light as a sculpturing tool for neuronal activity. Can we spawn an activity with a pulse of light? These are two photon images. Deep images in the brain. These kinds of microscopes have been used in basic science, but endoscopists have been using these. What if we used light to turn cells on and off? What if we beam lights to those cells?

Nagel et al. in 2003 found a blue light gated ion channel. It's a molecule found in a green alga used to drive its flagella around. So it can work in normal tissue. When you light it up with blue light, it lets them charge. And more importantly, we can target specific genetic types. And we can do that by taking the gene for this genetically encoded protein, putting it into a virus, and popping it back into the cell. We can see it lighting up on the cells. We can see the cells on the border here. If we modify it, we can have it modify specific spike trains. Not like the ones in your auditory cortex right now. There are little "thoughts" under neath the thought train. These are being used to test the specificity of generating a specific behavior. You can put this into an animal and it will respond as if it had been touched.

I want to highlight a new clinical area where some people have lossed their photoreceptors in their retina. Several groups including ours have been pursuing the opsis. Sackpath and Bil Housworth at Florida. Here is a mouse. The blindness mimics that of a human. We delivered a gene to a mouse. We were able to sensitize them to light. Take a camera, an electrode array, and finally you can digitize the that data. Or if you can just go straight to the retina then that might be better. This is a mouse we put in a water maze. This is a blind mouse. It's supposed to go to the platform underneath the map. It goes down an alley that is incorrect. If you do one administration of the virus and the gene for the light-sensitive protein. You put the mouse in and then you can see it's now doing much better. It can avoid obstacles and go to the platform. Can there be a clinical use here to treat forms of blindness? The acting CEO of Eos Neuroscience. Is this model safe? We'll talk about viruses second. The molecule comes from algae. Is this okay to put in the brain?

Hans et al., 2009 Neuron 62(2):191-198.

We wanted to look for entabodies as serum. We wanted to see if there were reactions against these molecules. So far we haven't seen any pathology. Adeno-associated viruses have had >600 people in a trial without a single adverse problem. Blue-lighted elicited spikes in non-human primate brain. Activating neurons with light. Over a period of many months, does the signal run out? We see high fidelity signals throughout.

It would be nice- both from a neuroscientific aspect- but also there are some cases where we wish to be able to silence neurons. Halorhodpsins - light activated chloride pumps from archaebacteria. You shine light and the spikes are deleted. The amplitude, as you see, is not very powerful. The amplitude of the hyperpolarization is much less than the spike. We started exploring genomic diversity. We screened molecules from all over the world that could yield higher currents. This is an example of a digital off-switch where we can shut down 99.9% of the spikes of the neuron. The neurons look great. They are not suffering. We've also started to discover that you could shut the spectrum on these. Here are two populations of cells. One is shut down by blue but not by red. We can delete projections from one region and those from another. Independently we can perturb them to understand when they are needed. From a hardware standpoint, one of the problems with visible light is that it doesn't go very deep into tissue. Jake in the lab wanted to figure out if there's a way we could do implantable high fiber optic arrays. This might be high-throughput screening for regions of the brain. Bernstein et al., in preparation. And, recently we've been able to fabricate these things. This thing is 8 or 9 mm. It can be under independent control to 2 dozen sites in the brain. Maybe even hundreds or thousands of lights to control. There's some other tech that you can combine with this. Chan et al., figured out how to do fluidic injection of these viruses. The brain is 3D and complicated. We need to sensitize precise circuits. We need to do a behavioral and scientific standpoint. One of her triple injection arrays. This is a paper that is going to come out soon. You can see 3 different points in the motor cortex that can be individually controlled. Invasive implants are used in an electric form for neurology and psychiatry. More than 100,000 people have implanted electrode arrays. 30,000 or more people with parkinsons or more with deep brain stimulators. There's a surgeon doing tuourrete's syndrome implants in adolescents. Can we use our optical strategy?

Here's an example that Jake Bernstein along with Emily Coug who wrote the software. This is also going at the lab at MIT. We labeled different prefrontal regions with the different molecules. We implant the fiber arrays to hit these regions. Here's an example of a result where we do pavlovian fear conditioning (a tone with a shock). One of the most popular neuroscience method over the last 100 years. A neutral queue becomes associated with a neutral state. The controls, even though they are being exposed to a neutral stimulis, and hope that it becomes neutralo again, is not getting better. With the interface treating cognition, emotion and movement.

Towards noninvasive means. TMS has been around for 25 years. Last fall, it was approved by the FDA for stimulating the left dorsolateral prefrontal cortex. Can we improve this by sculpting fields and targeting deeper structures? There was this idea of developing "brain coprocessors". Devices that can read out and deliver information. We can develop hardware that comfronts the 3D structure of the brain. How do we generate hypotheses about the brain? Can we start to build intelligence? Not just open loop, but ways of augmenting these emotional and cognitive circuits with real intelligence. I started teaching a class called Principles of Neuroengineering. There's a lab class too, Applications of Neuroengineering. In collaboration we are also doing Neurotechnology Ventures in a high-risk high-drama space like neuroengineering.