Nature Trifecta

A big day for systems neuroscience in Nature yesterday: three papers! Each paper investigates a different question about synaptic organization in the cortex. Not one paper created a new word ending in '-omics,' an auspicious sign.

I superficially describe the main results from each paper below, with some figures.

First, Brown and Hestrin bring us Intracortical circuits of pyramidal neurons reflect their long-range axonal targets. After fluorescent labeling of corticocortical (CC), corticostriatal (CS), and corticotectal (CT) pyramidal cells in cortex, they sliced the mouse brain and patched onto as many as four of the cells in layer five of V1 to measure the probability of cells synapsing onto cells of the same type (and later in the paper, different types).

They found distinct patterns of connectivity for the different cell types (see Figure). For instance, while 20 percent of the CS cells were connected monosynaptically, CT pyramidal cells only hooked up with one another about five percent of the time. They further showed this wasn't merely because some cells are more promiscuous than others (though they didn't show this for CS neurons).

Next up was a paper from Murayama and others in Larkum's group titled Dendritic encoding of sensory stimuli controlled by deep cortical interneurons. They loaded layer V cells in rat somatosensory cortex with calcium indicator, and then imaged layer 1-3 calcium activity during hindlimb stimulation. The supragranular signals represent activity solely in the apical dendrites from the loaded Layer 5 pyramidal cells. Via various pharmacological manipulations (often involve injecting more boluses into layer V), as well as in-vitro patch clamps, they support the claim that a particular type of inhibitory interneuron in layer V suppresses dendritic calcium levels. Then, using triple patch clamp (two neighboring layer V pyramidal cells, and one of their dendrites), they showed that stimulating one of the cells produced dendritic inhibition in the other cell via a disynaptic connection.

Of the three papers, this would be the best one to present in a journal club because it is fairly complicated and hard to understand on a quick once-through. A journal club audience would appreciate you doing the work for them. Frankly, I still haven't thought through the logic of all their experimental manipulations.

Third, from Petreanu and others in Svoboda's group is The subcellular organization of neocortical excitatory connections. It was only a matter of time before the channel rhodopsins spawned acronyms. In this quite elegant paper they described their application of sCRACM [subcellular ChR2-assisted circuit mapping] to determine the spatial organization of axodendritic synapses onto neurons in somatosensory cortex of mice. They did this in slices in which particular areas expressed channel rhodopsin ChR2. For instance, they expressed ChR2 in the VPM nucleus of the thalamus, which carries information to the whisker barrels in S1. Then they could stimulate the axons of the VPM neurons with a laser to find its postsynaptic targets.

They patched onto a cell in S1, and would then laser-stimulate the channelrhodopsin-expressing axons in the vicinity of the patched neuron. When the HH channels were blocked, a laser pulse on a ChR2-expressing axon would still generate PSPs in the postsynaptic cell. The spatial resolution of the mapping was approximately 60 um², so they were able to map the distribution of synapses with decent precision. The figure accompanying this paragraph is an activation map of a layer 5B pyramidal cell in which the ChRs were expressed in the VPM nucleus of the thalamus. Most of the VPM->S1 synapses occur in Layers IV and VB, though there is some activity in supragranular layers.

The second and third papers use very cool methods to achieve fairly unsurprising results. The first paper used more common methods, but the results were a bit more interesting (i.e., less predictable). All in all, a good week for synapses.

Happy Birthday, Darwin

Why not read something cool about evolutionary biology in honor of one of our greatest naturalists?

Some good online reading:
*Darwin: an online exhibit from the American Museum of Natural History.
*15 Evolution Gems: beautiful piece put together by Nature discussing 15 papers published in their journals that provide striking examples of evolutionary thinking. Unfortunately no figures.
*Evolution 101: A nice introductory course put together at Berkeley.

Good books:
*Frogs, flies, and dandelions: a book that focuses on what we know, and don't know, about speciation.
*How and why species multiply: a great book on speciation using Darwin's finches as a case study. Written by two monsters in the field.
*Evolution of Nervous Systems: an amazing four-volume set edited by Kaas. Too expensive to buy, but you should be able to find chapters online through your library. Check this link to see if you have access from your IP address.
*Endless forms most beautiful: a wonderful introduction to the hot new science of evo-devo (evolutionary developmental biology).

Consciousness (4): Levelling with Mr B

Mr B has hypothesized that the brain is necessary and sufficient for conscious experience in humans (and probably other animals). The brain, unfortunately for Mr B, is an incredibly complex object. It consists of multiple interdependent processes that operate across different spatial and temporal scales.

Spatial scales in the brain
The left-hand side of the figure below lists neuronal processes that operate at different spatial scales (spatial scale increases as you go down the figure). Associated with each level of organization are experimental techniques typically used to access the phenomena at that level. A few of these techniques are listed to the right of each level.


Measurements at a lower-level are often used to help illuminate what is going on in higher levels. For instance, there are many studies that correlate single cell responses with behavior. A detailed anatomical characterization of individual cells slowly builds up a picture of the distribution and abundance of neuronal cell types in the entire brain.

While the techniques used at higher levels typically don't reveal the details of the lower-level processes (e.g., fMRI does not tell you what is happening in an individual neuron), the data from higher levels do provide useful clues about the functional roles of the lower level phenomena. For instance, behavioral studies can suggest how individual motor neurons help to govern behavior. This is important because, as we discussed before, part of Mr B's job as a biologist is to discover the function of the mechanisms he is studying. Just as you won't understand the biological role of sperm by focusing narrowly on how a sperm locomotes, it is not possible to understand the function of an individual neuron without studying its role in the neural network in which it is embedded, and ultimately the role of this network in behavior.

Temporal scales
As spatial scales increase, the relevant temporal scales also tend to increase. This is because the higher-level processes emerge from the interaction of many events at lower levels. For instance, a single action potential lasts about a millisecond, so network dynamics take place on longer time scales (network dynamics require presynaptic action potential propagation, neurotransmitter release, and postsynaptic responses often in a large number of neurons).

Note this positive correlation between spatial scales and temporal scales is not a hard and fast rule. There are lower-level molecular processes that can take longer than minutes to unfold, for instance.

At what level is consciousness?
At what level(s) of organization should Mr B begin his investigation of consciousness? This is not something that can be decisively answered a priori. He needs to dive in and do some experiments to discover the spatiotemporal organization of processes in the conscious brain. We will visit many of the techniques and levels of organization in the above chart as we follow Mr B.

However, Mr B knows enough neuroscience to form tentative hypotheses about the levels of organization required for consciousness. For instance, it is quite unlikely that a single neuron is sufficient for consciousness. This would be a brittle way to build an important process into the brain. This means that we should be looking at the level of the neural network or higher for the neural signatures of consciousness.

On the other hand, we know that the entire brain is not necessary for consciousness. People lose bits of their brain all the time (e.g., car accidents and strokes) without losing consciousness. Further, it is clear that behavior isn't constitutive of conscious experience. We can be paralyzed by curare but still conscious, and when we dream our motor system is effectively shut down but we still have experiences.

Hence, somewhere between small neural networks and areas/nuclei in the levels chart are the most likely candidates to find processes essential to consciousness. This suggests the time-scales of the processes should be relatively long (i.e., longer than the millisecond scale at any rate).

Minimal basis for consciousness
Mr B can use the above discussion to put a finer point on his working hypothesis from the previous post (namely, the brain is necessary and sufficient for consciousness in humans). Because the entire brain isn't necessary and sufficient, there must be some subset of the brain that is. It could be a certain set of cortical areas and subcortical nuclei. It could be the entire cortex, or perhaps just the brain stem. We just don't know right now. But what Mr B is after is this minimal subset of neuronal processes that is necessary and sufficient for consciousness.

Behavior and consciousness
Perhaps paradoxically, while Mr B doesn't think behavior is constitutive of consciousness, the most direct experimental indicators of consciousness he has are behavioral. That is, the best way to find out if someone is aware of something is to ask them if they see (or feel, or hear) it. There is presently no foolproof neuronal measure of conscious experience. While Mr B does think that such a measure should exist, it is something we will have to discover by doing the science.

The good news and the bad news for Mr B
That consciousness seems to be a higher-level phenomenon makes it both easier and harder to study. Mostly harder. The good news is that most of the techniques targeted at lower levels can be used to study consciousness, such as single cell recordings. And of course, those techniques developed to record larger-scale activity, such as fMRI, may also be revealing. This is useful because such techniques are noninvasive and can readily be applied to humans.

That consciousness is a relatively high-level phenomenon has an obvious down side: it is likely incredibly complicated. We don't even understand how the motor cortex controls limb movement, a relatively simple phenomenon. It will take a herculean effort over many decades to develop the empirical infrastructure required to establish a consensus about how the brain is conscious.

In the next post, we'll follow Mr B as he starts on this daunting series of investigations, starting with binocular rivalry.

Consciousness (3): Mr B's first look at consciousness

Now we'll look at how your garden-variety biologist (Mr B) approaches the phenomenon of consciousness. For now, we'll have him treat it as he would any other biological phenomenon.

Recall that Mr B takes a naturalistic, empirical approach to things. His first order of business is to determine what variables are correlated with conscious states (just as he did with action potential generation). We'll focus mostly on conscious perception of external events (e.g., seeing a sunset), so as to avoid the complexity of things like consciousness of one's thoughts (e.g., the experience of thinking about a chess move).

Mr B does not focus narrowly on experiments that tell us about the neural basis of consciousness, but also on experiments that reveal important details of the structure of consciousness itself and its relationship to external stimuli (i.e., psychophysics). The more empirical constraints, the better. It is possible to learn a great deal about respiration without knowing anything about the respiratory system: you can learn how the inputs (composition of air breathed) and outputs (exhaled air) relate to one another, and to other variables such as the blood pressure and breathing rate of the organism. Hence, learning about a biological mechanism doesn't mean focusing in on that mechanism exclusively: much can be learned by studying its products, how it is perturbed by inputs, etc..

The brain is necessary for conscious experience
At the grossest and most obvious level, Mr B notes that the only organ necessary for consciousness is the brain. Contrary to the Greeks' heart-based theory of mind, he knows people have literally lived without hearts, perfectly conscious, for months (article here). You can lose kidneys, arms, your stomach, etc, and while you may not be healthy or happy, you will still be conscious. Conversely, if you inactivate a brain with an anesthetic, the loss of consciousness will be quite dramatic.

The brain is sufficient for conscious experience
Take a powerful hallucinogen and entire new experiences are evoked endogenously. Something similar seems to happen while dreaming: a world is experienced that is largely independent of present sensory inputs. Amputees often feel that the removed limb is still present, moving around, making gestures, in the well-known 'phantom limb' phenomenon (this has been shown to not be due to irritation of the nerves at the end of the severed limb).

In all such cases, we experience a world that is not actually there. So the brain in effect constructs the experience. Some might like to say that the brain builds a 'representation' or 'simulation' or 'virtual reality model' of the world, and this is what we experience. Mr B may slip into such (often metaphorical) language, but for now he just means that experience is a neural construction, which is a more neutral way to put things (though note by saying it is a 'construction' he doesn't mean to imply it is a "mere construct" with no validity).

Note this hypothesis already generalizes beyond the data: Mr B is assuming that perceptual experience during normal waking periods is generated by similar mechanisms to those used during sleep, hallucinations, and phantom limbs. Mr B realizes this could be a mistake, but as a provisional hypothesis, it seems reasonable, especially given the existence of illusions generated even in healthy brains (we will have more to say about illusions later).

Implicit in the hypothesis that experience is a neural construct is the claim that neural processes of a certain sort (to be determined) are not just necessary, but sufficient for experience. Given his general biological approach, it seems a conclusion almost forced upon Mr B.

In the next post, we'll continue to follow Mr B in his quest to understand consciousness. He'll see just how complicated a problem he has taken on.

Consciousness (2): Introducing the garden-variety biologist

This is the second of my ongoing discussion of biological approaches to consciousness and creationists' recent attacks on such approaches. In this post I sketch a portrait of a fictional character, a garden-variety biologist we'll call 'Mr B.'

Let's assume Mr B doesn't understand how neurons fire action potentials. In the rest of this post we'll examine his general approach to the problem. In a future post we'll consider how he approaches the problem of consciousness.

He believes that neuronal excitability is likely complex, but that it will ultimately be explained in terms of individually innocuous mechanisms, a complicated orchestra of proteins, lipids, carbohydrates, and other ingredients standardly found in cells. The mechanisms should all conform to physical principles, even if many of them cannot strictly be derived from the laws of physics. For instance, if there are untethered chemicals in a neuron, he expects their diffusion to follow the rules laid out in physical chemistry.

Mr B takes an empirical approach to his subject matter. He is likely to sit down at the lab bench with an example of what he is studying (a model system), and poke and prod at it to see how it behaves.

For instance, to get a bead on how neurons are activated, he may prepare a single neuron in a dish and treat it with various chemicals (e.g., sodium, potassium, neurotoxins), expose it to different temperatures, different light and oxygen levels, etc while measuring the voltage across its membrane. Such experiments will reveal how the behavior of the neuron depends on different variables in the preparation.

The experiments, guided by his best guess at how neurons work, will help him form new ideas or refine his old ideas. For instance, when he removes sodium from a neuron's bath, he finds that the neuron stops firing action potentials. This suggests to him that the action potential is caused by an influx of sodium into the neuron.

Mr B will usually write out equations to summarize what he has observed. However, he doesn't just want to describe his observations. He will attmpt to come up with new experiments to test his ideas about the action potential (e.g., if his sodium-based theory is true, then increasing the concentration of sodium in the neuron's bath should result in a larger action potential). The desire to turn his ideas into predictions often involves translating his words into mathematics so the concepts can be more clearly expressed, make his assumptions explicit, and provide a basis for precise predictions.

So far, Mr B is not much different than a physicist or chemist. All take an empirical approach to their subject matter, prefer mathematical to word-based models, and value empirical tests of their theories.

I've been painting Mr B as a bit myopic, focusing exclusively on how this little mechanism works. This leaves out his broader uniquely biological perspective. By focusing in on mechanisms, Mr B might be able to explain how a sperm locomotes, but that will tell him nothing about its function, about its role in the biological system in which it is embedded. This biofunctional orientation is what tends to distinguish Mr B from his colleagues in physics and chemistry.

Focusing in our our example, Mr B wants to know why neurons fire action potentials. What do action potentials contribute to the nervous system's higher-order goal of controlling behavior? Are action potentials involved in signalling from neuron to neuron? Could he be studying an epiphenomenon? It could be that the mechanisms he found in the dish are not even used in vivo. For instance, is there enough extracellular sodium in the nervous system for his sodium-based theory to work? Such questions will haunt Mr B and suggest new experiments.

Some might be tempted to insist that another facet of Mr B's approach is that he takes an evolutionary perspective on what he is studying. This is certainly possible, but not essential. Mr B realizes that brains are organs that evolved to help organisms navigate the world. But this doesn't necessarily help him understand how individual neurons work, or even their functional role in an intact animal. Evolution will indirectly color his perspective on the system he is studying, and certainly he has no patience with creationists who would say that the mechanism of action potential generation could not have evolved without divine intervention. Mr B realizes he doesn't even understand the mechanisms involved yet, and that is an important prerequisite to constructing a phylogenetic history.

Before taking leave of Mr B, we should note that he believes his ignorance of neural excitability is a relatively boring psychological fact about himself, not a deep fact with profound metaphysical implications (this is a point Patricia Churchland likes to make about consciousness, but right now we're leaving aside consciousness). He knows, as he approaches the problem of neuronal excitability, that he might be like the biologists in 1900 trying to understand the mechanisms of inheritance, that it might be a long time before he succeeds. A novel conceptual and empirical infrastructure might be required before the problem can be solved, or even posed in a way that yields results. His ignorance spurs his curiosity and creativity, it doesn't make him think there is something fundamentally wrong with biology. He stubbornly resists creativity sinks such as claims that neuronal excitability is forever beyond our understanding, or that supernatural beings are required to explain the strange animal electricity observed in nervous systems.

In the next post Mr B will examine the neural basis of consciousness at an abstract level, considering what types of processes in the brain are most likely to be conscious. He will also see how daunting his task is.

Creationists take aim at neuroscience (1): defining their target

A recent opinion piece in New Scientist, Creationists declare war over the brain, discusses the natural alignment between antievolutionists and those that think the human mind (in particular consciousness) is forever outside the explanatory reach of neuroscience. The topic of consciousness tends to bring out the nutballs, and creationism ties people's knickers in knots, so the article has received a good deal of attention from the internet commentariat.

Since I've thought about this topic way too much, I thought I'd throw my crap into the ring too. I'll discuss the arguments of the neodualists indirectly at first, dividing my discussion of consciousness into multiple posts. Because 'consciousness' is a dirty word in some neuroscience quarters, in this post I'll clear the air by clarifying what I mean by the term.

What is consciousness?
What are you experiencing right now? For instance, are you aware of hunger pangs in your gut, words on a screen, the deep red hues of a freshly picked rose? 'Consciousness' is just another word for this ability to perceive or be aware of the world. Indeed, for those who want to avoid the C-word, 'awareness' is a perfectly good synonym.

The canonical instances of conscious awareness are moments when we are awake, alert, and attending to something interesting such as a sunset. However, even while dreaming we are conscious of something, perhaps a sort of neuronal simulation of the world.

Should scientists bother with consciousness?
Over beers many neuroscientists are dismissive when consciousness comes up. They treat it as a "philosophical" problem, a waste of time for real scientists. I find this attitude strange. New data fuel conceptual progress in science, so it seems an empirical approach is the best way to make headway on something that is clearly a real and important phenomenon. Avoiding the topic leaves it in the hands of the philosophers, a fate just a little better than death.

I suppose one could argue that there is no way to study consciousness experimentally because it is inherently subjective or something. This argument doesn't work, though, as there already exist fairly straightforward experimental probes of consciousness. For example, binocular rivalry. If you show a different image to each eye (see example rivalrous stimulus below), you don't see a fusion of the two images. Rather, you perceive the images one at a time (a dog then a cat, not a dog-cat). Neuroscientists can compare the bits of the brain that track the eye-locked stimuli (which stay the same) with those that oscillate with the visual percept. This has provided a useful roadmap that tells us which parts of the brain are locked to the stimulus, and which shift with the object of conscious awareness.


The dismissive types are typically either unfamiliar with such experimental paradigms, or they tend to be skeptical of all research with a psychological component. For the former, Koch's book The Quest for Consciousness gives a nice summary of many experiments. For those skeptical of all cognitive neuroscience, there isn't much to be done (frankly, I am sympathetic to general skepticism toward cognitive neuroscience, which is a very speculative discipline right now). Hence, my take-home argument is that consciousness is just as legitimate (or illegitimate) a research topic as more mainstream psychological phenomena like attention and memory.

I should add one caveat. I have been writing as if all uses of the term 'consciousness' refer to the same thing. This may be false. Perhaps there are separate mechanisms for different sensory modalities. Or even within a modality: for instance, there could be different mechanisms for awareness of things in the center versus the periphery of our visual field. Maybe the mechanisms that underlie dreaming have little overlap with waking awareness. It could be that 'consciousness' is a mongrel term like 'memory,' and it will splinter as the science progresses.

My next post will begin describing what a biological approach to consciousness would look like.

Model systems in systems neuroscience?

Gilles Laurent makes an excellent point in 23 Problems in Systems Neuroscience:

Integrative neuroscience is an odd biological science. Whereas most biologists would now agree that living organisms share a common evolutionary heritage and that, as a consequence, much can be learned about complex systems by studying simpelr ones, systems neuroscientists seem generally quite resistant to this empirical approach when it is applied to brain function. Of course, no one now disputes the similarities between squid and macaque action potentials or between chemical synaptic mechanisms in flies and rats. In fact, much of what we know about the molecular biology of transmitter release comes form work carried out n yeast, which obviously possesses neither neurons nor brain. When it comes to computation, integrative principles, or "cognitive" issues such as perception, however, most neuroscientists act as if King cortex appeared one bright morning out of nowhere, leaving in the mud a zoo of robotic critters, prisoners of their flawed designs and obviously incapable of perception, feeling, pain, sleep, or emotions, to name but a few of their deficiencies. How nineteenth century!

I do not dispute that large, complex systems such as mammalian cerebral cortices have their own idiosyncrasies, both worthy of intensive study and critical for mental life. [...] Yet considering our obsession with things cortical, can we say that we have, in over forty years, figured out how the visual cortex shapes the responses of simple and complex cells? Do we really understand the cerebellum? Do we even know what a memory is? Do we understand the simplest forms and mechanisms of pattern recognition, classification, or generalization? I believe that our hurried drive to tackle these immensely complicated problems using the most complex neuronal systems that evolution produced--have you ever looked down a microscope at a small section of Golgi-stained cerebral cortex?--makes little sense.

Take that, you vertebrocentrists!

In defense of people who want to study the more complicated systems, Ed Callaway once said to me that there are two "model" approaches you can take. One, take a relatively simple system (e.g., the leech) and study the hell out of that. Another option is to take a relatively simple part of a complex system and study that(e.g., the retina, a single cortical column). Both approaches have their place.