RISE-ing to the Occasion
How do we know the difference between sarcasm and sincerity, or understand that in many social situations people don’t always say what they mean?
The subtleties of speech and hearing have eluded neuroscientists for more than a century, but now two professors at Columbia have come up with a novel approach to try to understand the intricacies of verbal communication.
Nima Mesgarani, assistant professor of electrical engineering, and Sameer Sheth, assistant professor of neurological surgery, are tackling the age-old problem by observing the brain’s electrical patterns when an individual listens to complex conversation. They want to observe how multiple regions of the brain interact with and support one another to produce comprehension.
The novelty of their approach lies in the fact that they are placing electrodes deep inside the brain; previous research placed sensors on top of the cortical surface on top of the brain. The technique will allow them to identify low-level sensory processing, such as when the brain distinguishes a murmur from a yelp, as well as high-level cognitive processing, such as our ability to recognize and understand contextual information, long-term danger and ambiguity.
“Our study will enable the creation of more accurate models of the brain mechanisms involved in speech communication and will advance our understanding of how these processes become impaired,” said Mesgarani. “At the same time these models and algorithms could enable machines to more closely perform such human tasks as speech and language comprehension.”
The research could also lead to significant advances in health care, particularly the diagnosis and treatment of patients with speech-related illnesses such as dyslexia, language-learning delay, autism and cerebral palsy, as well as cognitive impairment resulting from brain trauma and injury.
The National Institute of Deafness and Other Communication Disorders (NIDCD) reports that approximately 8 million U.S. residents have some form of language impairment, while an additional 5 percent of American children have speech disorders without a known cause.
The professors received a University grant of $160,000 over two years through a highly competitive program called Research Initiatives in Science & Engineering (RISE), run through the Office of the Executive Vice President for Research. In addition to the Mesgarani and Sheth project, the University granted seed funding to five other winning teams out of 60 applicants for its 2015 competition.
“RISE encourages PIs [principal investigators] to take risks and to push the boundaries of their conventional disciplines, always with the goal of making fundamental new discoveries,” explains Dr. G. Michael Purdy, executive vice president for research.
The research involves working with patients who are undergoing invasive surgery to place electrodes deep in their brains. Such interventions are done solely for medical reasons, such as for identifying the source of seizures, and the scientists are using this opportunity for other neuro-scientific research. The research protocols are all approved by the Institutional Review Board (IRB) at Columbia University.
By programming computers to evaluate the data gathered by the brain’s scattered electrical impulses, Mesgarani and Sheth aim to shed light on such neural mysteries as how we can overhear someone mutter our names in a crowded room, or attend to a particular person in a cocktail party while suppressing the other voices.
“An accurate neurophysiological model that adequately explains the neural transformation involved in speech communication will have an overarching impact in many areas, including cognitive and systems neuroscience, neurolinguistics and speech technologies,” said Mesgarani. “Furthermore, a detailed model of the mechanisms involved in speech communication will advance our understanding of how these processes break down in people suffering from various speech and communication disorders and may motivate new therapeutic measures.”
Perhaps Mesgarani and Sheth’s novel methods in the exploration of this uncharted territory of human brain electrophysiology may even help us learn someday how to craft that elusive perfect conversation.
What happens when a cell is starved?
Much of a cell’s energy is devoted to making proteins, one of the fundamental building blocks of life. Yet little is known about the mechanisms underlying a cell’s ability to turn on and off protein production during times of nutrient limitation. Discovering the underlying regulatory mechanism, and how it works in times of energy restriction, will answer important biological questions in cellular protein generation and cellular metabolism. Applications for these discoveries could lead to a better understanding of human nutrition, or therapies involving cellular metabolism, which may be useful in diseases ranging from obesity to cancer.
In 2013, Dr. Jonathan Dworkin, Associate Professor of Microbiology & Immunology, and Dr. Ruben Gonzalez, Professor of Chemistry, collaborated to study this elusive problem, and were awarded $160,000 in seed funding through the Office of the Executive Vice President for Research’s Research Initiatives in Science & Engineering (RISE) competition.
Their RISE-funded research identified two regulatory factors that help govern protein synthesis in response to limitations or abundance of nutrients that they are now investigating as part of an R01 [GM114213-01, “Regulation of Protein Synthesis by Ser/Thr Phosphorylation”]. One of these regulatory factors allows the cell to chemically modify and inactivate the protein synthesis machinery when cellular energy intake is low, essentially putting the machinery into a sleep-like, or dormant, state. Once the cell has adequate energy intake, the other regulatory factor then reverses the chemical modification, allowing the cell to rapidly activate the dormant protein synthesis machinery and initiate the process of building new proteins. A detailed molecular understanding of the process through which the protein synthesis machinery can be switched from “active” to “dormant”, and vice versa, has the potential to unlock new avenues of basic and biomedical research.
RISE grants target particularly imaginative research proposals that are not yet ready to obtain funding from traditional sources, but those that, if successful, would significantly advance scientific knowledge, understanding, and interdisciplinary discovery.
With questions, please contact the RISE program staff at email@example.com.
Do you trust your weatherman? Neither do most Americans, which is a noteworthy problem given the frequency and intensity of extreme weather events such as tornados, hurricanes, thunder and lightning storms, floods, hailstorms, and heat waves. The National Weather Service can now anticipate extreme weather two to three days before impact, which, while a significant achievement, is not yet adequate for environmental scientists and engineers, or for the people and communities affected. Dr. Michael Tippett, Lecturer in SEAS’ Department of Applied Physics & Applied Mathematics, is challenging the paradigm by developing novel methods to forecast extreme weather variations one full month in advance of impact.
Predicting the seemingly unpredictable is an elusive task, but Dr. Tippett and his co-PIs—Dr. Suzana J. Camargo, Lamont Research Professor; Dr. Adam Sobel, Professor in SEAS’ Department of Applied Physics & Applied Mathematics and A&S’ Department of Earth & Environmental Sciences; and Dr. John Allen, Postdoctoral Research Scientist in the International Research Institute for Climate and Society—believe there is method in the dark sky’s madness. The Office of Research Initiatives’ Research Initiatives in Science & Engineering (RISE) program provided $160,000 of seed funding over two years, allowing researchers to lay the groundwork for seasonal predictions of US tornado activity. The team’s subsequent Tornado Index gathers and analyzes data on the correlation between fluctuating atmospheric temperature, humidity, pressure, and extreme weather events. It has been presented at conferences and workshops, with three publications already accepted into the American Meteorological Society’s Journal of Climate.
In May 2014, with two years’ worth of RISE-funded data supporting the theory, the Tippett team’s innovative prediction model was recommended for a $300,000 grant from the National Oceanic & Atmospheric Administration’s (NOAA) Climate Program Office. Dr. Tippett recognizes this success as emerging out of his participation in the RISE program, which provided him and his colleagues with the seed funding essential to craft a quality, impactful, innovative, and ultimately successful government grant proposal. We ask the Columbia University community to join us in congratulating the Tippett Group on their pioneering grant.
RISE grants target particularly imaginative research proposals that are not yet ready to obtain funding from traditional sources, but those that, if successful, would significantly advance scientific knowledge, understanding, and interdisciplinary discovery. Please visit the RISE website in August 2014 for information regarding the 2014-2015 competition. With questions, please contact Ms. Victoria Hamilton, Director of the Office of Research Initiatives, at firstname.lastname@example.org.
According to statistics compiled from the National Institute on Deafness and Communication Disorders, approximately 15% of American adults (37.5 million) aged 18 and over report some trouble hearing, and approximately 13% of Americans aged 12 and over has hearing loss in both ears. The primary method for alleviating severe hearing loss is a cochlear implant, which replaces the mechanosensing function of the ear by using behind-the-ear microphones that pick up and deliver sound signals to electrodes implanted within the ear. Successful as cochlear implants are, there is still room for their improvement, both in the sound-sensing and stimulation roles of cochlear implantation.
Dr. Ioannis (John) Kymissis, Associate Professor of Electrical Engineering, and Dr. Elizabeth (Lisa) Olson, Associate Professor of Otolaryngology/Head & Neck Surgery and Biomedical Engineering, collaborated to discover innovative avenues to improve upon the current cochlear implant. Dr. Kymissis – an expert in MEMS-based polymer piezoelectric sensors and actuators – and Dr. Olson – a leader in cochlear mechanics and intracochlear pressure sensing – were awarded $160,000 of seed funding from the Office of the Executive Vice President of Research’s 2012 Research Initiatives in Science and Engineering competition for their proposal to reinvent the cochlear implant.
Through their RISE-funded research, Drs. Kymissis and Olson have made significant progress in the mode of sound sensing, by developing an intracochlear microphone that is incorporated within the implant. Current cochlear implants use an external behind-the-ear microphone, like that in conventional hearing aids. External microphones pose difficulties for hearing when there are multiple sound sources and for certain activities (e.g. swimming). Early attempts at internal microphones were implanted under the skin behind the outer ear. Though cosmetically more appealing than conventional microphones, these picked up too much body noise. The researchers instead investigated whether a cochlear implant with an internal microphone constructed from polyvinylidene fluoride (PVDF), a polymer piezoelectric, would allow for better and higher fidelity sound detection. With RISE funding, Drs. Kymissis and Olson tested an intracochlear sound sensor first in the laboratory, then on rodent cochleae, and finally on human cadaver cochleae, in collaboration with Dr. Hideko Nakajima of the Massachusetts Eye and Ear Infirmary.
Elizabeth Olson commented on their most recent tests, in which the sound signal detected with their intracochlear PVDF sensor, implanted in a human cochlea rivaled that of a conventional hearing aid microphone: “The students sang ‘happy birthday’ to the temporal bone inside the sound booth and we could hear the song quite clearly, coming from the PVDF sensor signal, delivered to a speaker outside the booth.” John Kymissis noted “It’s exciting to see systems that can directly measure sound through the same path that is normally used for hearing – access to the inner and middle ear allows us to build new sensors to take advantage of that pathway.”
As a long-term goal, the piezoelectric device could also be used to directly stimulate the auditory nerve. This innovative use of piezoelectric polymer would allow the implant to transduce the intracochlear pressure signal of the inner ear directly into an electrical charge to stimulate the ear’s auditory neurons. This future cochlear implant will provide accurate localized readings of pressure and local characteristic frequency, which will allow for objective mapping of the electrode-to-stimulus frequency and most effective stimulation of the auditory neurons.
Much progress has already been achieved from the research conducted through this RISE grant. In addition, this project has had continued funding from Advanced Bionics, a leading cochlear implant company, and an NIH application is underway. The success of the piezoelectric cochlear implant will not only mean functional benefits, but also cosmetic benefits for those suffering from hearing loss. By furthering developing cochlear implants, Dr. Kymissis and Dr. Olson are defining a future in which science ultimately harmonizes with the human body instead of imposing itself on it.
Breathe in; breathe out. This critical bodily function goes largely unrecognized. That is, until your breathing stops. People with Chronic Obstructive Pulmonary Disease (COPD) must carefully determine whether the simplest of tasks – standing at the stove, walking up the Low Library stairs – is within reach. COPD’s end-result – respiratory failure – ranks as the third-largest cause of death in the United States, especially rampant within minority communities. Conventional medical treatments target the lung airways, but this methodology has achieved only limited success with reducing patient mortality rates.
Emlyn Hughes, Professor of Physics, and R. Graham Barr, Florence Irving Associate Professor of Medicine and Associate Professor of Epidemiology, are battling this disease by teaching an old gas new tricks. Their Multi-Ethnic Study of Atherosclerosis (MESA) project has substantiated a most creative hypothesis: COPD is caused by smoking-related damage to the heart’s pulmonary vessels that carry blood to the lungs. Their innovative idea, stimulated by $160,000 of seed funding from the Office of the Executive Vice President for Research’s Research Initiatives in Science & Engineering (RISE) program, allowed Drs. Hughes and Barr to utilize MRI techniques to observe that COPD patients show early signs of both a decrease in pulmonary blood flow and reduced volumes in the heart’s right ventricle. Both conditions, combined as what they term cor pulmonale parvus, are now trustworthy predictors of COPD mortality. In other words, while traditional treatments have targeted the lungs, Hughes & Barr demonstrate that proper treatment should instead target the pathway connecting the heart to the lungs.
And then, this is how Hughes & Barr took our breath away. Polarized helium-3, an element abundant in our environment due to Cold War-era nuclear weapons testing, contains a nucleus with the perfect spin to align with the MRI’s magnetic field, thereby producing a crystal-clear image of the lung; this allowed Drs. Hughes and Barr to progressively visualize the pulmonary blood flow decrease and identify times and targets for clinical intervention. Utilizing this rare gas for biomedical research brings us significantly closer to detecting and treating COPD without exposing sick patients to harmful radiation, thereby potentially eliminating one of the gravest threats to global human health.
By its completion, theirs will be one of the most extensive biomedical research projects employing 3He as a biomedical tool. Gathering the preliminary data to support their highly innovative hypothesis would not have been possible without the funding they received from the RISE program, which supplied the requisite seed funding to execute a cutting-edge and effective clinical experiment. We ask the Columbia community to join us in congratulating Dr. Hughes and Dr. Barr on their exceptionally cutting-edge research: A testament to what can be achieved by marrying Columbia University’s considerable strengths in the physical and biomedical sciences.
Babies communicate by crying, but how can parents discern a standard narration from a battle cry of agony? Colic – an observed set of symptoms of baby (and parent) distress, rather than a formal disease – is characterized by crying for over three hours per day, for three or more days per week, for over three weeks’ time, and affects between 2% and 5% of infants. Colic has a troublingly strong correlation with Shaken Baby Syndrome, which accounts for between 240 and 400 deaths per year in the United States; it is additionally responsible for thousands of hospital visits and hundreds of thousands of emotionally-drained parents. Nevertheless, the causes, diagnoses, and treatments of infantile colic are still unknown: We are no closer to understanding or curing this ailment than we were before the advent of medicine. This raises a terrible hue and cry.
Dr. Ansaf Salleb-Aouissi, Associate Research Scientist within the Center for Computational Learning Systems, is leading a team that applies machine learning techniques to better understand the causes of colic, by identifying similarities across medical records of babies brought to the hospital due to crying. Insight into better diagnosis, treatment, and even prevention of colic may be found within the hundreds of thousands of doctors’ notes created from hospital visits.
Funded by the Executive Vice President for Research’s Research Initiatives in Science & Engineering (RISE) program, the Salleb-Aouissi team received $160,000 of seed funding from 2011-2013, and analyzed pediatric notes for 1,240 babies brought to New York-Presbyterian Hospital due to excessive crying. Despite inconsistency across all note formats, along with the informality of many of the records, the Salleb-Aouissi lab generated a number of fascinating initial findings: 63% of colicky babies were male (though only 51% of babies studied were male); constipation was noted four times as often in non-colicky babies than in colicky babies; excessive crying was noted over 1000% more often in colicky babies than in non-colicky babies.
Following these initial observations, the Salleb-Aouissi team received an NSF Early Concept Grant for Exploratory Research grant for $175,457 entitled, “EAGER: Collaborative Research: Advanced Machine Learning for Prediction of Preterm Birth,” to continue developing computational models that can accommodate hundreds of thousands more pediatric notes. They also aspire to use social media and parental blogs to gain more insights about infant colic and bring it to the public sphere. The activities of the Salleb-Aouissi team perfectly demonstrate the value of solving looming biomedical problems with methods in the physical sciences.
Please join the Office of the Executive Vice President for Research in congratulating Dr. Salleb-Aouissi on her team’s research accomplishments – we are a far cry from being helpless in the face of colic.
Should I get my morning coffee from Joe’s or Starbucks? Black or with milk and sugar? These are just two of the infinite number of decisions that we make each day, the outcomes of which depend on how we cognitively organize the information we receive from our environment. For example, I may prioritize a coffee shop closer to me. Or I may prioritize taste over health benefits and order a Venti Frappuccino with extra Splenda. But why?
We intuitively expect some variation in people’s preferences for coffee over, say, tea, the difficulty or cost of obtaining a cup, one’s need for coffee at a particular time, and how people perceive their environment. For instance, if I happen to see someone walking past with a cup of coffee, I may decide to get one myself whereas if I didn’t notice that person, I may not decide to. Furthermore, how I obtain the coffee depends on how well I attend to the environment, e.g. whether or not I pay sufficient attention to appropriate signs like “Coffee Shop.” However, this variation in how we seek information from specific sources at specific times – “information sampling behavior” – is a key empirical observation neglected in conventional decision theory relevant to neuroscience, psychology, business, and economics.
In a highly novel undertaking to account for variation in decision-making behavior, Michael Woodford, John Bates Clark Professor of Political Economy in the Department of Economics, and Jacqueline Gottlieb, Professor in the Department of Neuroscience and the Mortimer B. Zuckerman Mind Brain Behavior Institute, have been collaborating to close this gap between the empirical and the theoretical: between how we think people choose and how they actually make complex choices. When asked why an interdisciplinary team is necessary for this project, the team explained: “In economics we consciously consider the tradeoffs of certain decisions. In neuroscience, the brain selectively – and unconsciously – filters information, which clues us into the cognitive mechanisms that construct certain perceptions of choice options. Taken together, neuroscience provides the mechanisms through which we make decisions that are relevant to economics. However, the fundamental disconnect between the empirical nature of neuroscience and theoretical nature of economics necessitates an interdisciplinary team to collaborate on the problem of decision-making.”
One of the problems traditionally hindering this unification is a lack of means to collect a sufficient amount of data on information sampling behavior. Once access to multivariate data is gained, it becomes an additional problem to determine how to analyze it. “Access to more information has created problems that were easy to ignore in the past. For example, people are now crossing the street with smartphones in hand and have to make split-second decisions on whether they should sample information from their phones or look at the street light. This is why we see so many people being hit by cars while staring at their phones – they’ve sampled information faultily. Because people now have access to so much information all at once, it is more important than ever to understand human decision-making processes.”
In light of heightened accessibility of information – taking advantage of the “Big Data” revolution – and recent developments in software engineering, Drs. Woodford and Gottlieb have proposed downloadable mobile applications to gather data from tens of thousands of participants, including difficult-to-reach or geographically-distant populations in Brazil or India. Data from this highly-innovative sampling method far exceeds the depth and breadth of data from traditional collection methods such as laboratory testing. “Our mobile application initiative is inspired by The Great Brain Experiment conducted at University College London, wherein they managed to garner 30,000 pieces of data in only one day’s time. So, this method of using mobile applications to collect massive amounts of data looks very promising.”
With access to such a rich data set, this research may further uncover how demographic factors (e.g. age, native language, gender) correlate with certain behaviors. The team may be able to answer novel questions ranging from “what factors determine whether I prioritize coffee flavor or caffeine level” to “what factors determine whether I choose to get seasonal flu vaccinations” or “what factors determine who I will vote for in 2020?” Many industries invest substantial resources in marketing science and anticipating consumer behavior, but they have historically been hindered by the substantial gap in our predictive theories versus observed experience. This is even more challenging because rational influences do not always govern people’s choices. Nonetheless, this project tackles both of these challenges through a novel means to collect massive amounts of empirical data. Ultimately, this project aims to intuit a random person’s preferential choices based on their basic demographic and sociographic qualities.
With $160,000 in seed funding from the Office of the Executive Vice President for Research’s Research Initiatives in Science & Engineering (RISE) program, the Woodford-Gottlieb team is currently designing new experimental designs for the mobile application. The seed funding has provided the opportunity for an economist and a neuroscientist to “integrate their disciplinary perspectives to construct a more focused snapshot of human behavior. Without funding from the RISE program, we would not have the means to create and execute our innovative experimental designs and acquire our data. The RISE program is an invaluable resource that grants the opportunity for unconventional programs to take off.” And Woodford and Gottlieb’s unconventional project goes far beyond its inspiration: Rather than focusing well-established behavioral measures to conventional tasks, this project is developing completely new experiments and avenues for exploration that could ultimately lay the technical foundation for future data-intensive behavioral studies. Developing these novel theories may further position the team towards successfully securing external funding through the National Science Foundation or the National Institutes of Health – a key consideration for the RISE grant and larger program.
By constructing a new experimental gaming paradigm, Woodford and Gottlieb’s research will be one of the first to use big data-gathering mobile applications to tackle never-before-researched questions of information sampling behavior. Please join the Office of the Executive Vice President for Research in congratulating Drs. Gottlieb and Woodford on their high-risk, high-reward research endeavors. In shedding light on the realities of human decision-making, the Woodford-Gottlieb team is paving the path for us to finally wake up and smell the coffee (your choice of Joe’s or Starbucks, but soon, we may be able to accurately guess which one).
RISE grants target particularly imaginative research proposals that are not yet ready to obtain funding from traditional sources, but those that, if successful, would significantly advance scientific knowledge, understanding, and interdisciplinary discovery. With questions, please contact the RISE program staff at email@example.com.
—Written by Delia Wu, Senior Undergraduate Fellow