The brain performs a massive number of complex computations, all day long. In a game of dodgeball, the visual cortex might signal that something is approaching your face — fast! — and promptly alert the motor cortex to duck! from an oncoming sphere. Or, maybe, when you detect the smell of your grandmother’s perfume, signals from the olfactory cortex combined with memories stored in your hippocampus might transport you back in time to enjoyable childhood visits. Each area of the brain is like a high-tech command center: taking in information, transforming it with computation and then rerouting it to new cerebral locations.
For decades, scientists have been trying to understand the content of those signals. Adding to the complexity is the fact that researchers are asking not only about the content of the information sent, but also how and where computations happen and how they effect thought and behavior. Tracking the paths and transformations of messages through multiple brain networks may be the key to understanding how the brain processes information so flexibly.
Enter the Simons Collaboration on the Global Brain (SCGB), a network of 76 scientists dedicated to doing just that: understanding how internal brain processes occur and also how they then impact the transformation of sensory information into actions. Now in its seventh year, SCGB is unveiling important lessons about memory, decision-making and the kinds of theoretical frameworks scientists will need to peel away the layers of complexity and reveal the inner workings of the brain in unprecedented detail.
For decades, SCGB investigator and University of California, San Francisco researcher Loren Frank has been interested in the neural circuits active in memory formation. As an animal moves through space, neurons in the hippocampus fire to record the animal’s location and to update other areas of the brain, effectively forming a spatial memory. These sequential firing patterns are reactivated during ‘sharp-wave ripples,’ events thought to act as time-compressed versions of spatial memories that are often replayed during stillness and sleep.
In 2020, Frank’s lab published a study in Neuron investigating how different regions of the hippocampus broadcast this spatial memory to the nucleus accumbens, an area involved in reward. Before this study, the team knew that both the dorsal and ventral regions of the hippocampus exhibited sharp-wave ripples, but not whether these regions communicated with different or overlapping networks in the accumbens. By recording simultaneously from the hippocampus and the nucleus accumbens, they showed that dorsal and ventral ripples activated mostly separate sets of neurons and had opposing effects when they spoke to the same neurons.
Lead author Mari Sosa, who wrote the paper as a graduate student and now works as a postdoctoral fellow with SCGB investigator Lisa Giocomo at Stanford University, thinks these opposing effects may help the brain tease apart different parts of a memory, such as specific locations or the emotional and social context of an experience. “In order to disassociate specific pieces of information, you might want to have neural circuits dedicated to separately storing and retrieving those different pieces of information,” Sosa says. When you’re recalling your walk to the coffee shop, you can remember which corner it’s on, or the sense of relief after the first sip of your latte, or both. By integrating different sets of inputs, it’s likely that the accumbens neurons are able to tease apart these memories.
The actual messages being sent between the hippocampus and the nucleus accumbens, however, are still a mystery. “We don’t know what the content of these replay messages are just yet, but one possibility is that the dorsal and ventral hippocampus are routing different types of information during memory storage and retrieval,” Sosa says. It’s possible that these different messages are what’s reflected in the different firing patterns in the accumbens. Sosa’s work adds to an increasing pile of evidence that most information processing happens in neurons distributed across brain areas. “You might have groups of neurons within a brain area that are specialized to process a certain type of information, but that’s probably not the only thing they’re doing,” Sosa says. “They could also be modified to do something a little bit different depending on what inputs they receive from other areas.”
Flexibility is indeed a crucial feature of how our brains work — it’s how we remember, learn and ultimately change our behavior. It’s possible that having different routes for information can help the brain more flexibly compute information based on context, experience, mood and more.
Still, the mechanisms of this flexibility, and their timescales, are a bit of a mystery. “Given that almost everything in the brain is connected to everything else, how do brain regions connect and disconnect?” SCGB and Howard Hughes Medical Institute investigator Karel Svoboda asks. “What are the mechanisms, and how do they enable flexible computation?” To address these questions, Svoboda is working on expanding our capacity to image multiple brain areas simultaneously.
Working with fellow SCGB investigator Liam Paninski, a researcher at Columbia University, Svoboda is developing novel ways to image large swaths of brain activity down to the resolution of synapses. Last year, they helped develop a microscope that can simultaneously image more than 9,000 inhibitory neurons across four different cortical areas, along with callosal projection neurons spanning two hemispheres. The hope is that this technology will shed light on the precise connections that enable behaviors.
Other researchers are turning to high-yield electrophysiology devices such as Neuropixels recording probes to determine how brain areas work together to route information. Working as part of the International Brain Laboratory (IBL), a 21-lab collaboration funded in part by SCGB, SCGB investigator and University of Washington researcher Nick Steinmetz is using Neuropixels to understand how rodents make decisions. Steinmetz is primarily interested in how an animal’s internal state influences its brain activity and ultimately its behavior; researchers at the IBL are determined to figure out what is different in the brain when an animal is paying close attention versus when it’s disengaged. He trains mice to make visual decisions and has noticed that an animal’s engagement with a task can change how quickly and how well it makes a decision. “The same stimulus on the retina fails to generate a behavior,” Steinmetz says. “Somewhere, the activity is different.”
Steinmetz and his colleagues believe that changes in the structure of communication between brain regions may be what distinguishes different attentional states. “It’s actually by modulating information flow, and via high-dimensional communication patterns, that the behavioral effects of engagement are brought about,” Steinmetz says.
But activity happening synchronously through multiple brain regions can be difficult to parse. Most computational theories of brain activity describe the flow of information from one area to another, with many transformations occurring along the way. New theories and methods are needed to tackle the logic of multiple brain regions speaking to many other regions.
This is where SCGB investigators like Byron Yu come in. Yu, a neuroscientist at Carnegie Mellon University, is working with fellow SCGB investigators Adam Kohn of Albert Einstein University and Christian Machens of the Champalimaud Foundation to develop new statistical methods to parse the relationships between brain areas that enable visual perception. In the past two years, their team published two papers showing that different areas in the visual cortex could communicate through specific channels, or subspaces. Their idea is that this works like a lock and key: Signals that match the channel are sent to the next brain area; mismatched signals are not. Identifying this neural mode of signaling relied on new mathematical approaches, including a few that can tease apart feedforward and feedback signals, a crucial distinction in understanding how information is routed.
Frameworks like these underscore the importance of understanding the brain as an interconnected network. “We have a great temptation to draw box-and-arrows diagrams and think that they tell us how the system works,” Steinmetz says. “It’s pretty clear it’s going to be more complicated than that.”