Over the past few years, there has been growing interest is something called "resting state functional MRI", a technique for seeing what your brain is doing when you aren't doing much of anything at all. It turns out that brains at rest are pretty restless, consuming far more energy than they do when doing. More interesting, "resting" activity is not random, but highly coherent, consistent, and predictable. The discovery of the brain's characteristic resting behavior led some years ago to the postulation of a "default network" for the brain--a set of regions that consistently cooperate to do . . . well, what, exactly, we don't know. But surely it must be something interesting. Your brain would hardly waste all of that energy dancing to the beat of its inner drummer if there weren't some reason for it, right?
Our ignorance regarding the function of all that fluctuation isn't for lack of trying. The discovery of the brain's default network has led to hundreds of studies relating the default network to the brain's anatomical structure, as well as to mood disorders like depression, developmental problems like autism, and degenerative diseases like Alzheimer's. It has even been suggested that resting state activity holds the key (cue deep voice and echo effect) to understanding consciousness itself. Now, when neuroscientists start brandishing the C word, there are two predictable reactions: increased public interest and attention; and increased scientific scrutiny and criticism. Both have happened here, generating a cadre of enthusiastic adherents, and an equally committed group of critics, who question whether we should continue wasting our energy figuring out why the brain appears to be wasting its energy. Or, as one prominent neuroscientist put it to me recently, "It's just such a fad. I kind of hate it."
Hyperbolic allusions to cracking the mystery of consciousness excepted, I don't think anybody should hate it. But to see why we should care about the brain's intrinsic activity requires us to think about brain function in a new and unfamiliar way.
The brain is, in essence, a collection of oscillators: billions of biological yo-yos going endlessly up and down. The electrical charge in individual neurons rises and falls; as do the local concentrations of neurotransmitters like dopamine and serotonin, the amount of oxygen in the local blood supply, and the brain's overall electrical field (creating the brain waves you can see with an EEG). Each of these oscillations has a different beat for different circumstances--like the fast "beta" waves seen in the alert working brain, and the much slower "delta" waves of dreamless slumber. These rhythms interact in various ways, from the shared swaying of a slow dance, to the counterbalanced ups and downs of two kids on a seesaw, all the way to the complex, syncopated interplay of a jazz band. And like all pendulums, each of the brain's various oscillators also has its preferred swing, a way it will move if left to its own devices. The combination of all these individual, periodic, preferred oscillations is the brain's resting state.
Which brings us to the first reason to care about intrinsic activity: what we typically think of as brain function--seeing, thinking, deciding, acting--is actually a disturbance, an alteration of the brain's natural harmonies. When we think about brain function, then, it is not enough to ask "why this activity?", we must also ask "why this change?" Why did this (perceptual, behavioral, electrical, chemical) input cause this exact deviation? If we want to understand the brain's particular sensitivities, we need to know not just that some part of it reacts to some stimuli or task, but how much it reacts, and what else reacts (and interacts) with it. Resting state fMRI is one method for ascertaining the background against which such changes must be measured.
Similarly, just as in music no note has meaning in isolation, so too the brain's local activity can only be fully understood when set in its context. Consider in this regard hippocampal place cells--those neurons famous for firing whenever an animal is at some particular spot in the world. The apparent one-to-one mapping between hippocampal cells and environmental locations has led their function to be understood in terms of simple location representation, as if each cell is designed for saying: "you are here." But these cells have been in the news lately because of the discovery that they are also sequentially activated prior to novel spatial exploration, which indicates their function is not so simple, and underlines the point above that to understand the brain's activity now one must also consider what its activity was and will be. [Dragoi, G. & Tonegawa, S. (2011). Preplay of future place cell sequences by hippocampal cellular assemblies Nature, 469: 7330 (397-401).]
Moreover--and more to the current point--these cells fire not just when an animal is at a given location, but just before, and just after, too. Interestingly, the differences between concurrent firing (the "you are here" signal), prospective firing (signaling in advance of being at a location), and retrospective firing (signaling after the animal has left a location) is marked not by any difference in the neuron's activity itself, but rather by its relationship to the background theta-band (~6-10 Hz) oscillation of the whole Hippocampus. In its retrospective role the cell fires earlier, and in its prospective role later, in the theta cycle than it does when the animal is actually at the location in question. [Buckner, R.L. (2010). The role of the Hippocampus in prediction and imagination. Annual Review of Psychology, 61:27-48.]
Place cell B fires at different times relative to the background theta wave as the rat moves from locations A thru C
Figure: Place cell B fires at different times relative to the background theta wave as the rat moves from locations A thru C. Reprinted from Buckner (2010) with permission of the author.
In other words, what that cell's activity means--what it is actually signaling--depends on how that activity relates to the ongoing background oscillations.
Thus, understanding the resting state is important because it draws our attention to the ever present context within which function needs to be measured and meaning assessed. In the dynamic brain, local activity is always a change from what was happening before, and occurs in relation to all the other things happening now.
It is hard to think about the brain this way, especially given the continuing influence of the computer metaphor for the brain, and of the focus on localization (vision here, language over there, motor control in this bit) that we inherited from an earlier age of neuroscientific investigation. In a computer what matters is the nature of the local processing, what's happening in this chip at this time. Background processes are irrelevant to the process of interest, which can therefore be safely isolated and studied as such. But the brain is not that sort of machine: the background is not irrelevant, and relationships between oscillations do a lot of the functional work.
This leaves us with a lot of thinking still to be done. What does it mean for brain function to be defined not just by the intrinsic characteristics of the supporting neural activity, but also by the deviation that it represents from some default? Can the change from one dynamic equilibrium to another be itself a functional event? How is it possible for a machine to do work not with oscillations themselves, but with relationships between them? We are a long way from answering these questions, but we've made some promising beginnings. One theoretical development of particular interest is called liquid state computing, an attempt to understand how information processing can be carried out by variously coupled oscillators that respond to inputs the way a pond does to a stone. I'll devote some future post to explaining how brain function might arise from such interacting ripples in the neural aether. For now, however the important point is this: measuring, characterizing and reflecting on the resting dynamics of the brain are important early steps toward understanding the functional dynamics of the brain, in terms much more appropriate to the biological reality than electrons whizzing around in isolated silicon processors.
(Music-brain picture credit: ScriptPhD.com)