Tuesday, May 28, 2019
University Campus Center Auditorium
1 Campus Way, Amherst MA 01003
This conference brings together neuroscientists and engineers to find areas of overlap for collaboration. It includes new approaches for visualizing and recording from neurons, manipulating gene expression in neurons, and understanding brain function. Speakers will also talk about interfaces between the brain and the rest of the body and the mechanical forces on neurons. The conference is preceded by a morning workshop on Methods in Systems Neuroscience and Neurotechnology.
Registration is free and open to faculty, students, and staff from UMass and other universities,. It features two keynote speakers: Andrew Barto (UMass) and Steven McCarroll (Harvard) and five local speakers. There will be an evening poster session for anyone to present their neuroscience or engineering research regardless of topic.
Download Conference Program (pdf file)
Download Poster Abstracts (pdf file)
Abstract submission is closed. There will be on-site registration.
UMass Neurosciences Lifetime Achievement Awardee
Andrew Barto
Professor Emeritus
College of Computer and Information Sciences
University of Massachusetts Amherst
Can Machine Learning tell us anything about
the neurobiological basis for learning?
Artificial intelligence (AI) is much in the news these days, with most of the attention due to the success of machine learning (ML) algorithms. Some of the most impressive AI systems rely on learning by multi-layer, or “deep”, artificial neural networks (ANNs). Despite the fact that ANNs were inspired by networks of real neurons, their correspondence to real neural networks remains superficial, and the ML algorithms they use are motivated by mathematical theory rather than by neuroscience. After briefly explaining the most common ML algorithms, I argue that while some popular algorithms are unlikely to tell us much about the neurobiological basis of learning, other algorithms correspond to neural data in surprising ways that have been enormously useful in making sense of data. I conclude by describing directions suggested by these correspondences that remain to be explored.
Keynote Speaker
Steve McCarroll
Dorothy and Milton Flier Professor of Biomedical Science and Genetics
Department of Genetics
Harvard Medical School
From cell types to synapses to genetic perturbations –
How much of brain biology might technologies make into “big data” problems?
We work to develop technologies for monitoring diverse aspects of neurobiology in new, data-rich ways, and learning in this way about normal function and illness in the brain. I will share ongoing work on ways to measure and learn from: cell types and cell states; synaptic connectivity; and the effects of natural genetic variation. A common theme of these engineering efforts is the use of nucleic acids, sequencing and computational analysis to create new ways to “listen to” biology in a high-bandwidth manner. Our aspiration is that engineering innovations of this nature can turn broader aspects of biology into big-data problems, in which the rate of progress can be accelerated by using computer science, math and statistics together with biology. We hope that these new approaches will create new and useful ways to understand the biological basis of brain function and brain illness.
Keynote speaker sponsored by Models to Medicine, Institute of Applied Life Sciences.
Schedule
8:30 Workshop registration
9:00- 11:30 Morning Workshop on “Methods in Systems Neuroscience and Neurotechnology”
12:00 – 12:30 Conference Registration
12:30- 12:40 Welcome Remarks: Paul Katz (IONs), Deputy Chancellor Steve Goodwin, Dean Tricia Serio (CNS), Dean Laura Haas (CICS), Assoc Dean Erin Baker (Engineering)
12:40 – 1:20 UMass Neurosciences Lifetime Achievement awardee: Andy Barto (CCIS, UMass)
1:20 – 1:40 Award presentation by Chancellor Subbaswamy
1:40 – 2:00 Coffee Break
2:00 Introduction by Jean King (Dean of Arts & Sciences, WPI)
2:00 – 2:20 Erin Solovey (WPI)
2:20 – 2:40 Jun Yao (ECE, UMass)
2:40 – 3:00 ChangHui Pak (BMB, UMass)
3:00 – 3:20 Yubing Sun (Mech Eng, UMass)
3:30 – 3:50 Dirk Albrecht (WPI)
3:50 – 4:10 Coffee Break
4:10 – 5:10 Keynote Speaker: Steven McCarroll (Harvard Medical School)
5:10 – 7:30 Posters and refreshments – Poster Abstracts (pdf file)
Local speakers:
Erin Solovey
Assistant Professor
Computer Science
Worcester Polytechnic Institute
Improving Human-Computer Interaction with Real-time Brain Input
Most human-computer interaction techniques cannot fully capture the richness of the user’s thoughts and intentions when interacting with a computer system. In this talk, I will discuss ongoing research on the effective use of brain sensor data to expand the bandwidth between the human and computer. Using functional near-infrared spectroscopy (fNIRS), as well as other biosensors, we can detect signals within the brain and body that indicate various cognitive states. These devices provides data on brain activity while remaining portable and non-invasive, which opens new doors for human-computer interaction research. The real-time cognitive state information can be used as an implicit, supplemental input channel to provide the user with a richer and more supportive environment, particularly in challenging contexts such as education and training, healthcare and anything involving information overload, interruptions or multitasking.
Jun Yao
Assistant Professor
Electrical and Computer Engineering
UMass Amherst
Syringe-Injectable Mesh Electronics
with a Plug-and-Play Input/Output Interface for Neural Interface
Neural probes employing a ‘mesh’ form, which have tissue-like flexibility and macro-porosity, can yield mechanical match and an intimate interface to neural tissue. These same properties, however, make input/output (I/O) connection to measurement electronics challenging. Here I will discuss a strategy of integrating the mesh electrode with a plug-and-play I/O (input/output) interface to enable rapid and user-friendly probe technology for neural interface.
ChangHui Pak
Assistant Professor
Biochemistry and Molecular Biology
UMass Amherst
Combining genetic and cellular engineering in human pluripotent stem cells
to model mental disorders
With the advent of high throughput sequencing of patient genomes and induced pluripotent stem cells, we are now able to probe the function of genetic risk alleles for mental disorders in human cellular context. In this talk, I will provide an example where I have combined genetic and cellular engineering approaches to identify a key cellular mechanism underlying schizophrenia.
Yubing Sun
Assistant Professor
Mechanical and Industrial Engineering
University of Massachusetts Amherst
Engineered in vitro models for early-stage human neural development
We recently developed engineered in vitro models based on human pluripotent stem cells to recapitulate early stage ectoderm development. Using this system, we discovered that the initiation of cell fate patterning and regionalization in early ectoderm might depend on the mechanical status within the tissue, working synergistically with the reaction-diffusion of BMP and Wnt signaling molecules. Our research demonstrates that by leveraging microengineering tools and pluripotent stem cells, we can gain a substantial amount of knowledge of human development and perform high-throughput screening assays to speed drug discovery.
Dirk Albrecht
Associate Professor
Biomedical Engineering
Worcester Polytechnical Institute
Neurotechnologies to study regulation of neural excitability and behavior
Neural imaging technologies directly observe brain signals in living organisms, but current methods are limited in experimental throughput. For example, “whole-brain” imaging of every neuron in the nematode C. elegans is limited to ~20 minute recordings, one animal at a time. This is great for mapping overall activity and inferring connectivity within an individual at one moment in time, but it is insufficient to study how this activity changes over hours, shaped by experience, internal states, or external stimulation. Instead, our work focuses specifically on expanding imaging throughput along all other dimensions—more animals, more perturbations, more stimuli, and longer times—while examining a smaller number of relevant neurons. Using microfluidic systems, we analyze population behavior while monitoring neural activity for up to 24 hours, exploring activity changes during learning, sleep-wake transitions, and development. Robotic microfluidic systems deliver hundreds of stimuli to automatically measure response variability and identify mechanisms regulating neural excitability via reverse genetic screens. We also developed a functional screening technology to assess compounds that alter optogenetically- evoked neural activity in vivo, and identified several neuroactive drugs that act intracellularly to suppress or enhance neural excitability. Future studies aim to identify regulators of synaptic communication, and to monitor neural dynamics in models of human neuropsychiatric disorders.