May 15, 2013
My graduate student, John Sheppard (pictured here next to his rig), published a paper last week in the Journal of Vision. In the paper, John reported something interesting about the way that rats (and humans) combine multisensory information. He examined behavior during situations where one modality is more reliable than the other. You might have experienced this kind of situation yourself: when you are speaking to someone in a crowded room, the auditory information is lousy, but the visual information is pretty good. John found that in such situations, the lousy auditory information doesn’t get thrown away, but it influences the decision to a much lesser degree compared to when it is more reliable. John was able to study this phenomenon in the lab by putting two sensory stimuli in “conflict” with one another: the auditory cue indicated one thing while the visual cue indicated something else.
John’s experiment built on established methods in multisensory integration that have shown that multisensory estimates tend to slide towards the more reliable unisensory inputs (like in the sliding distributions shown here). His approach was novel in two ways: first, he did it not only in humans but also in rats. The rodent work paves the way for a bunch of amazing electrophysiology experiments which he is now doing. Second, he used a stimulus that affords insight into how subjects use the time period in which the auditory and visual stimuli are presented. Were subjects procrastinating? Or perhaps making snap judgements? Nope: John found that mostly subjects used time pretty well. A few humans were exceptions: their tendency to make snap judgments underscored the importance of his analysis. The rats were more consistent: they made use of almost all the information available to them. A worrisome thought if you find yourself in a battle of wits with a rodent hiding in your kitchen!
This week, my lab members and I attended a mini symposium at our own Cold Spring Harbor Lab. Organized by colleagues Tony Zador and Adam Kepecs, the focus of the day was understanding how the cortex and striatum interact to guide behavior. The speakers were Rui Costa, Linda Wilbrecht, Bernardo Sabatini, Anatol Kreitzer, and Tony Zador.
Linda Wilbrecht presented data aiming to understand the “currency” that the brain uses to make decisions. The talk included data from her recent Nature Neuroscience paper about behavioral effects of stimulating each of two classes of striatal dopamine neurons (D1-expressing and D2-expressing neurons). The first finding is that the two classes exerted opposite effects: stimulating D1 neurons caused more choices in one direction, while stimulating D2 neurons caused more choices in the opposite direction. The really interesting part, though, is that she found the effect of stimulation interacted with the animals’ reward history: for example, when stimulating neurons that caused more leftwards choices, she found the effect was enhanced if the animals had just experienced a few failed trials on the left side. WIthout stimulation, they would rarely go left after so many failures. But with the stimulation, they overcame this reluctance, and went left anyway. Understanding how recent reward history and current sensory information interact might one day give us insight into why the cycle of addiction is so hard to break.
Bernardo Sabatini described his experiments that aim to understand role of the direct versus indirect pathways. He stimulates each pathway optogenetically and looks at the effects on both behavior and firing rates of neurons in the motor cortex. He finds that the two pathways are not equal in the degree to which they affect firing rates and behavior. To visualize firing rates of motor cortex neurons, which are notoriously dynamic and complex, he reduced the dimensionality of the data using PCA. This analysis makes it possible to tell whether stimulation simply modifies an ongoing trajectory in high-dimensional space, or whether the stimulation drives the network to a novel state. The ability to perturb motor cortex and watch the resulting neural activity affords insight not just into striatal projections, but into developing motor commands as well.
April 15, 2013
With the growing elderly population in the United States and around the world, questions about age-related cognitive decline are on everyone’s mind. Despite massive research into possible causes for age-related decline, numerous mysteries remain. A recent article in Nature Neuroscience, led by researcher Josh Dubnau at Cold Spring Harbor, suggests a new mechanism that may drive cognitive decline.
Josh’s lab uses Drosophila melanogaster (above) as a model system because of the feasibility of genetic manipulation in that species. The focus of their current paper is transposable elements: DNA sequences that can change where they are in the genome, potentially creating mutations or changing the size of the genome altogether. Readers who are not genetic experts might still remember Barbara McClintock’s famous corn that exposed the existence of transposable elements.
This new study identified transposable elements in drosophila that were active during normal aging. That might have been just a coincidence, except they found that a mutant with extra transposon expression had even more age-dependent cognitive decline. It was as if the mutants lost their cognitive sharpness more quickly than their wild type siblings. The mechanism by which this abnormal transposon activation affects cognition remains a mystery and the lab has yet to show that the abnormal transposons cause the cognitive decline. Nevertheless, this research suggests a new avenue through which we might understand, and ultimately prevent, the cognitive decline that so often accompanies aging. Perhaps the work might also provide insight into the variability in this process: might adults who suffer more rapid cognitive decline have more abnormal transposon activity? If so, what might be the cause of this? Answering such questions could cause a dramatic improvement in the lives of a large population.
ga(‘create’, ‘UA-41061457-1′, ‘wordpress.com’);
April 5, 2013
Last week I was fortunate in having the opportunity to give some talks at the Riken Brain Institute outside Tokyo, Japan. I learned about fantastic ongoing work including during my visit: Tomomi Shimogori describe insights her lab has made about the driving force behind topographic organization, Hokto Kazama explained how his lab will use imaging to decode information in the fly brain about incoming olfactory signals and Keiji Tanaka described how the brain changes as learners transition from amateur to expert status
I also spent a lot of time with Justin Gardner discussing recent insights from his lab about perceptual processing humans. Recently, they have been delving into the neural basis of an established behavioral phenomenon: when subjects estimate the speed of low contrast gratings, they tend to mis-judge them a bit, rating them to be slower than they truly are. This has been interpreted by Eero Simoncelli’s lab as evidence for a “slow speed prior”: the idea is that we interpret sensory information by combining sensory evidence with prior beliefs. When the contrast is low, our perception is fuzzy, so we tend to go with the prior, rather than the sensory evidence. The Gardner lab is in a strong position to understand how this works for two reasons: first, they regularly image the brains of well-trained humans using fMRI; second, they have developed methods for decoding the signals they record, allowing them to understand how the brain changes with changes in the stimulus, such as contrast and speed. Using this approach, they find a clear signature of the slow speed prior, most noteably in the same early visual areas that represent the sensory evidence for speed such as V1 and MT. My lab is likewise interested in how priors are combined with sensory information. In the picture, you can see Justin and me arguing about whether insights from speed estimation priors apply to multisensory integration priors.
This past week, some lab members and I attended the Cosyne meeting in Salt Lake City. We heard emerging work about cue combination of spatiotemporal frequencies from Alan Stocker’s lab as well as a new perspective on LIP neurons from the labs of Alex Huk and Jonathan Pillow.
On the last day, we heard a fascinating talk by Deborah Gordon, an ecologist from Stanford, who studies ants. Dr Gordon has uncovered the signals that ants use when they make decisions about whether to go and forage for food. This decision is critical: if ants forage when conditions are poor, they dehydrate themselves and end up with little to show for their efforts. But by paying close attention to olfactory cues from co-foragers as they return, ants can make an informed choice about whether to look for food. The above image shows trajectories of ants in the lab: each line shows the path of a single ant as she wanders around, encounters colony-mates and adjusts her path accordingly. By automating the process of ant tracking, Dr. Gordon can simultaneously monitor the position of many animals and make broad conclusions about what they do.
Dr. Gordon drew an interesting parallel between ant behavior and neural transmission. She likened the pool of at-the-ready foraging ants to the readily releasable pool of synaptic vesicles that sit at the synapse. Just as the neuron integrates calcium over time, the ants integrate olfactory signals from other ants. And just as the neuron will spike when enough calcium is accumulated, the ant colony will “spike” out ants when enough signal from returning foragers is accumulated.
Many of us are familiar with seeing similar evolutionary strategies played out in different species, but this is something more. Here, it seems that there is a similar mechanism operating at very different levels: the level of the whole ant colony and the level of an individual, tiny synapse within one organ within one individual animal.
December 17, 2012
This post is to highlight the accomplishments of the students and postdocs in my lab in the past 12 months. We’ve made a ton of progress thanks to their hard work, talent and dedication. David Raposo (back row, right) published his first paper about rat and human behavior and made major inroads in understanding the neural circuits driving the behavior. John Sheppard (back row, left) has a paper in submission about cue weighting, and was awarded two graduate fellowships based on his proposed thesis work. Amanda Brown (front row, second from left) became an expert in animal training and is known around the lab as the “mouse whisperer” for her ability to generate stable, consistent behavior from these animals. Matt Kaufman (front row, left) joined the lab and has already got us thinking about the dimensionality of neural responses during decision formation versus during movement. Onyekachi Odemene (front row, second from right) joined on as our newest Watson School student and is already training mice effectively and laying the groundwork for a new direction in the lab. Mike Ryan (back row, center) joined us as a part-time technician on his way to graduate school and is already so good at making drives that we are plotting to keep him here. All told, it is a great pleasure working with these folks and I am excited to see what discoveries they will make in 2013.
December 12, 2012
I visited Baylor college of medicine in Houston last week. I heard about ongoing work in a number of labs, including Dora Angelaki and David Dickman. I also met with Andreas Tolias whose work aims to understand how responses in primary visual cortex lead to perception. We discussed his new optical imaging approach of steering the two-photon excitation using AODs (acousto optic deflectors). With AOD, he can rapidly image the responses of large numbers of cortical neurons. He and his students this method to examine population activity and correlations of neurons in primary visual cortex. Previous attempts to examine these issues relied on much smaller populations of neurons, so Andrea’s approach has the potential to shed new light on age old questions about population coding.
The picture below shows two of Andreas’s students, James Cotton (left) and George Denfield (right).
Two approaches for understanding schizophrenia gene DISC1 are reported at the Cold Spring Harbor in-house symposium
November 21, 2012
This week my lab attended the annual Cold Spring Harbor in-house symposium. We weren’t presenting this year, but heard about work from many labs here on campus. Some particularly interesting work focused on schizophrenia. Our interest in this topic stems from the fact that schizophrenics are known to have some abnormalities in how they use sensory information to guide behavior.
A major challenge in understanding the genetic basis of mental illness is that for may diseases, multiple genes are clearly involved. In the case of schizophrenia and bipolar disorder, for example, the genetic basis is clear, but studies of large populations show that two patients who suffer from the same disorder may have very different underlying mutations. The discovery back in 1990 of a large Scottish family with a high instance of schizophrenia and shared disruption of a gene called DISC1 affords a rare opportunity to understand the connection between genes and disease. But much remains unknown! At the symposium this year, two labs took very different approaches to tackle this problem.
Dick McCombie talked about his lab’s use of a genetic approach to further understand the mechanism by which mutations in DISC1 lead to bipolar disorder or schizophrenia. They seek to identify which proteins interact with DISC1 in the hopes of better understanding how its disruption might lead to disease. This could be important in understanding why some family members are susceptible to the mental disorders while others are not: perhaps there are also mutations in the proteins with which DISC1 interacts in patients, leading to a greater mutational burden that results in disease.
Bo Li‘s lab presented a poster that took a different approach to understanding how DISC1 mutations lead to disease. The poster was presented by Watson School student Kristen Delevich who probed the effect on neural circuits of DISC1 mutations. She measured electrophysiological responses in the medial prefrontal cortex and compared their magnitude and frequency in wild type mice and those with DISC1 mutations. Further, she took advantage of technology developed by Josh Huang’s lab that allowed her to disrupt DISC1 only in specified classes of interneurons. The attached image demonstrates the technique: she used a DISC1 hairpin and restricted it to the desired class of neurons. This targeted approach could reveal whether DISC1 mutations preferentially disrupt inhibitory neurons, an appealing idea since a disruption in the balance of excitation and inhibition has been speculated to have devastating consequences for cognition.
Taken together, these two studies begin to close the gap between basic science and the clinic: Using genetic tools that reveal candidate sites of disruption alongside electrophysiological tools that probe mechanism is a powerful approach. Given the devastating nature of schizophrenia and bipolar disorders, and their frequency in the population, bridging this gap between the lab and the clinic is critical.
This past summer, I have been the co-director of the Undergraduate Research Program (URP) at Cold Spring Harbor. I’ve had the pleasure of interacting with 26 bright, enthusiastic students from the all over: we had students from Cambridge, Brown, Berkeley, the University of Barcelona, and many other places. Each student is placed in a lab and works on their own project for 10 weeks.
Although many students this year worked on innovative and exciting projects, I chose to highlight the work of Zach Collins, an undergraduate at George Washington University. Zach researched the expression patterns of inhibitory interneurons in collaboration with Partha Mitra’s lab and Josh Huang’s lab. Zach is particularly interested in whether the expression pattern of subtypes of inhibitory neurons is unusual in Autism. Although the possibility of disruption in the balance of excitation and inhibition has been previously suggested as an underlying cause for Autism, the precise differences in inhibitory circuits have not yet been thoroughly examined. Zach’s goal was to find an automated way to make 3-d reconstructions of mouse brains that show expression of sub classes of inhibitory neurons. By comparing the brains of a mouse model of autism with control mice, Zach and his colleagues will be able to identify where the differences in inhibitory neurons are most pronounced. The image below is a heat map of expression for Gad2, an enzyme marker of inhibitory GABAergic neurons. The section is from Pavel Osten’s lab; Zach worked to develop an automated algorithm to detect the labeled cells and quantify them across sections. Below that is am image of Zach (on the right) relaxing with fellow URPs Emily Glassberg and Ed Twomey (taken by Constance Brukin).
August 6, 2012
Last week we got together at Cold Spring Harbor with folks from NYU and MIT who share an interest in parietal cortex, but approach it from different points of view. Our lab, with a focus on neural circuits and decision-making, shared recent developments in rodent behavior and electrophysiology. Bijan Pesaran’s lab, from NYU, described their recent work exploring connections between sub-regions of parietal cortex in primate, and their emerging work on identifying neurons in the parietal cortex that project to frontal areas. Finally, Kathy Rockland, an anatomist from MIT, answered our many questions about injections, expression levels, and anatomical landmarks, and then later described her work connecting parietal cortex and subregions of the hippocampus.
At the moment, it is clear that may aspects of posterior parietal cortex are conserved from mouse to rat to monkey to human: in all species, the region gets inputs from similar thalamic nuclei (posterior nucleus in rodents and pulvinar + posterior nucleus in primates), and has feedforward connections to frontal areas and the superior colliculus. One thing that remains unclear is the degree to which species differ in the sub-regions that comprise posterior parietal cortex. In monkey, subdivisions are clear, although there is considerable overlap in the selectivity of neurons in each area. In rodent, the existence of subdivisions remains a mystery. Are they there waiting to be discovered? Or are the neurons in the most medial and lateral portions of the area functionally equivalent? We hope to be able to address these questions in the coming months.
The image below parietal cortex neurons in the rat (From Tomioka & Rockland) as well as a (somewhat incomplete) group photo taken from the cafe at Cold Spring Harbor Lab.