Who’s in Charge?

Sameer Sheth, M.D., Ph.D.
Assistant Professor in Neurosurgery
Columbia University Medical Center
New York Presbyterian Hospital
Dana Grantee 2013-2016

We make decisions thousands of times a day. How does our brain do this in such a rapid way? What happens when decision-making goes wrong, as it does in psychiatric disorders such as obsessive-compulsive disorder, schizophrenia, and depression? These disorders of the circuitry underlying decision-making and emotional control affect a significant chunk of the world population. Sameer Sheth’s laboratory studies the neurophysiology of prefrontal circuitry by recording brain activity directly from neurons in people undergoing neurosurgical procedures for clinical purposes. His team has discovered a set of neurons that act as sentinels, detecting situations that require attention and alerting the rest of the brain.

What attracted you to study how the brain makes decisions?

These kinds of higher cognitive functions that humans participate in routinely are what differentiate us most noticeably from our closest evolutionary neighbors, the great apes. While animals also make decisions, there is no argument that these traits are most highly developed in humans. We face decisions all the time; they characterize our daily lives. The way our brains do it is a very complex process that we’re only beginning to understand.

Secondly is that we can study this very human brain function with the types of recordings I have access to as a neurosurgeon. Many of the procedures I do as a matter of course in my surgical practice involve placing electrodes into the brain to record cell activity for various diagnostic and therapeutic reasons. These procedures provide access to many parts of the brain, including prefrontal areas involved in decision-making. To have that access and to be able to study these questions in this way is extremely exciting.

Has this access to cortical regions in living humans changed the study of these complex cognitive functions?

Tremendously. It presents the opportunity to study decision-making in a very in-depth way, at the level of individual neurons and populations of neurons in distinct regions of the prefrontal cortex. We’ve been doing recordings in animals for decades, but in the last several years we have seen an increase in the number and types of clinical procedures in humans. This has presented opportunities that have greatly influenced this field.

A case in point is intracranial recordings in epilepsy patients. Traditionally, this has been done via craniotomy, in which neural activity is recorded from electrodes on the surface of the brain. But a lot of the cortex is buried below the surface, and these areas are difficult to access with surface-recording techniques.

A technique called stereo EEG—pioneered in Europe and now taking off in the US based on its usefulness in clinical decision-making —allows us to go deeper. Using minimally invasive percutaneous methods, we insert depth electrodes that penetrate the substance of the brain (intraparenchymal electrodes). That’s just one example of how techniques that are clinically useful open new possibilities for recording from areas that were not easily accessible before.

How does this work relate to therapies such as Deep Brain Stimulation?

Deep Brain Stimulation is an example of the burgeoning field of neuromodulation. In the last 10 years, there has been tremendous growth in using therapeutic neuromodulation to treat psychiatric disorders by targeting neural networks that have not been explored as much, especially circuits involved in decision-making circuitry and emotional processing. Neuromodulation for OCD, depression, and many other psychiatric disorders has come into the forefront in the past several years, fueled by promising data regarding efficacy.

Is the rush to “neuromodulate” psychiatric disorders putting the cart before the horse, given our still-limited knowledge of the circuitry underlying these conditions?

It’s true that if you try to neuromodulate for depression, for example, it requires you to understand depression circuitry to effectively access it, modulate it, and improve symptoms. These things go hand in hand: the research has to come along with the clinical work. It is a coordinated, parallel effort.

What in your view is the most important piece of your findings from the Dana-funded research?

The result I’m most excited about—and am about to submit for publication—focuses on the relationship between single neurons and larger populations of neurons in decision-making and cognitive control.

It relates to a theory that’s been around for many years and has many proponents, sometimes called “conflict theory,” which postulates that some parts of the brain are specialized to detect conflict, notably the cingulate in the medial prefrontal cortex. Conflict, in this context, occurs when information received is contradictory or incoherent in some way. If everything is congruent in the environment, it may be easier to make a decision, but if competing information is presented, you need to take into account all the streams of information. The latter scenario triggers a signal to other parts of the prefrontal cortex that essentially says: “Something is going on here, and you need to slow down your process and gather the relevant information to make the right decision.”

What we are finding is that there is a relatively small subpopulation of neurons, on the order of 10 percent of those we record from at a given time, that is clearly sensitive to these competing streams of information. Their firing rate changes. They fire more when there is conflicting information than when information is not conflicting.

So, you’ve discovered a subset of cells that kind of rev up when they detect conflicting information? What is the result of their increased firing?

What’s interesting is that there is a bigger population of cells, sometimes nearly half of the cells we record, that at first don’t appear to be sensitive to conflicting information. If you look at their firing rates alone, it seems like they don’t do anything. But when you look at the timing of their firing in the context of background activity in the cell population—what is called local field potentials, or oscillations of neural activity—it becomes clear how coordinated their firing is with these background oscillations. It suggests these neurons pay attention to conflict: They coordinate their firing with the background activity in a way that we think may help transmit that information to other parts of the brain.

A single neuron firing is like a voice in the forest: It can get lost easily. But if the voice can somehow be amplified across a large spatial scale, say by being echoed by someone 100 yards away, who passes that signal on and recruits others to do the same, the signal is less likely to be lost. That mechanism is what we find in this small population of cells. They recruit a much bigger population of cells to convey the information to other cells and other areas of the brain.

Why is this finding significant?

It speaks to the mechanism. There has been a lot of conflicting information from human fMRI studies vs. neural recordings in animals, and there is disagreement in the literature about whether individual cells are sensitive to these kinds of signals, such as conflict or difficulty. This work helps answer that question by identifying cells that are definitely sensitive to this type of information.

Still, not that many cells seem to have this sensitivity, so we asked how they coordinate with other cell populations. Our findings provide a mechanistic explanation of how a small population of cells can have their voices heard, so to speak, to convey the necessary information to other brain regions to coordinate decision-making.

Is this then proof of the conflict theory?

I would broaden the terminology and say these cells are sensitive to overall difficulty, or to situations that demand attention or control. It may be conflict, defined as some sort of incongruence, or it may be that there is a low level of information, like a hazy picture that only becomes clear as you get more information. It could also be danger, or something particularly rewarding or opportune. These cells sense the situation and send a signal to pay attention. It’s as if they’re calling for reinforcement, or requesting allocation of more cognitive effort to the task at hand.

Is there a name for these cells?

Good question. We could call them the Dana cells. [laughs] We just call them cells that are sensitive for demand or control. We don’t have a cool name for them yet.

How has the Dana grant advanced your research program?

This was the first research grant I got when I started my lab here almost five years ago. It has been instrumental in starting this work, hiring a post-doc, and in discovering the first pieces of data. The initial research has now snowballed into NIH funding both at an individual level, with an RO1 grant that is essentially a continuation of this work, and in collaboration with a couple other groups to examine these issues in concert with related questions. In addition, the big center that we’re starting at Columbia, the Cognitive Science and Neuromodulation Program, was seeded by the Dana grant.