Research Area: Computational Neuroscience

​machine learning, modeling, artificial intelligence, neural networks, memory, ​visual intelligence, audition, natural language processing, learning theory

Ila Fiete joins the McGovern Institute

by Sabbi Lall |

Ila Fiete, an associate professor in the Department of Brain and Cognitive Sciences at MIT recently joined the McGovern Institute as an associate investigator. Fiete is working to understand the circuits that underlie short-term memory, integration, and inference in the brain.

Think about the simple act of visiting a new town and getting to know its layout as you explore it. What places are reachable from others? Where are landmarks relative to each other? Where are you relative to these landmarks? How do you get from here to where you want to go next?

The process that occurs as your brain tries to transform the few routes you follow into a coherent map of the world is just one of myriad examples of hard computations that the brain is constantly performing. Fiete’s goal is to understand how the brain is able to carry out such computations, and she is developing and using multiple tools to this end. These approaches include pure theoretical approaches to examine neural codes, building numerical dynamical models of circuit operation, and techniques to extract information about the underlying circuit dynamics from neural data.

Spatial navigation is a particularly interesting nut to crack from a neural perspective: The mapping devices on your phone have access to global satellite data, previously constructed detailed maps of the town, various additional sensors, and excellent non-leaky memory. By contrast, the brain must build maps, plan routes, and determine goals all using noisy, local sensors, no externally provided maps, and with noisy, forgetful or leaky neurons. Fiete is particularly interested in elucidating how the brain deals with noisy and ambiguous cues from the world to arrive at robust estimates about the world that resolve ambiguity. She is also interested in how the networks that are important for memory and integration develop through plasticity, learning, and development in the brain.

Fiete earned a BS in mathematics and physics at the University of Michigan then obtained her PhD in 2004 at Harvard University in the Department of Physics. She held a postdoctoral appointment at the Kavli Institute for Theoretical Physics at the University of California, Santa Barbara from 2004 to 2006, while she was also a visiting member of the Center for Theoretical Biophysics at the University of California at San Diego. Fiete subsequently spent two years at Caltech as a Broad Fellow in brain circuitry, and in 2008 joined the faculty of the University of Texas at Austin. She is currently an HHMI faculty scholar.

Peering under the hood of fake-news detectors

by Rob Matheson |

New work from researchers at the McGovern Institute for Brain Research at MIT peers under the hood of an automated fake-news detection system, revealing how machine-learning models catch subtle but consistent differences in the language of factual and false stories. The research also underscores how fake-news detectors should undergo more rigorous testing to be effective for real-world applications.

Popularized as a concept in the United States during the 2016 presidential election, fake news is a form of propaganda created to mislead readers, in order to generate views on websites or steer public opinion.

Almost as quickly as the issue became mainstream, researchers began developing automated fake news detectors — so-called neural networks that “learn” from scores of data to recognize linguistic cues indicative of false articles. Given new articles to assess, these networks can, with fairly high accuracy, separate fact from fiction, in controlled settings.

One issue, however, is the “black box” problem — meaning there’s no telling what linguistic patterns the networks analyze during training. They’re also trained and tested on the same topics, which may limit their potential to generalize to new topics, a necessity for analyzing news across the internet.

In a paper presented at the Conference and Workshop on Neural Information Processing Systems, the researchers tackle both of those issues. They developed a deep-learning model that learns to detect language patterns of fake and real news. Part of their work “cracks open” the black box to find the words and phrases the model captures to make its predictions.

Additionally, they tested their model on a novel topic it didn’t see in training. This approach classifies individual articles based solely on language patterns, which more closely represents a real-world application for news readers. Traditional fake news detectors classify articles based on text combined with source information, such as a Wikipedia page or website.

“In our case, we wanted to understand what was the decision-process of the classifier based only on language, as this can provide insights on what is the language of fake news,” says co-author Xavier Boix, a postdoc in the lab of Eugene McDermott Professor Tomaso Poggio at the Center for Brains, Minds, and Machines (CBMM), a National Science Foundation-funded center housed within the McGovern Institute.

“A key issue with machine learning and artificial intelligence is that you get an answer and don’t know why you got that answer,” says graduate student and first author Nicole O’Brien ’17. “Showing these inner workings takes a first step toward understanding the reliability of deep-learning fake-news detectors.”

The model identifies sets of words that tend to appear more frequently in either real or fake news — some perhaps obvious, others much less so. The findings, the researchers say, points to subtle yet consistent differences in fake news — which favors exaggerations and superlatives — and real news, which leans more toward conservative word choices.

“Fake news is a threat for democracy,” Boix says. “In our lab, our objective isn’t just to push science forward, but also to use technologies to help society. … It would be powerful to have tools for users or companies that could provide an assessment of whether news is fake or not.”

The paper’s other co-authors are Sophia Latessa, an undergraduate student in CBMM; and Georgios Evangelopoulos, a researcher in CBMM, the McGovern Institute of Brain Research, and the Laboratory for Computational and Statistical Learning.

Limiting bias

The researchers’ model is a convolutional neural network that trains on a dataset of fake news and real news. For training and testing, the researchers used a popular fake news research dataset, called Kaggle, which contains around 12,000 fake news sample articles from 244 different websites. They also compiled a dataset of real news samples, using more than 2,000 from the New York Times and more than 9,000 from The Guardian.

In training, the model captures the language of an article as “word embeddings,” where words are represented as vectors — basically, arrays of numbers — with words of similar semantic meanings clustered closer together. In doing so, it captures triplets of words as patterns that provide some context — such as, say, a negative comment about a political party. Given a new article, the model scans the text for similar patterns and sends them over a series of layers. A final output layer determines the probability of each pattern: real or fake.

The researchers first trained and tested the model in the traditional way, using the same topics. But they thought this might create an inherent bias in the model, since certain topics are more often the subject of fake or real news. For example, fake news stories are generally more likely to include the words “Trump” and “Clinton.”

“But that’s not what we wanted,” O’Brien says. “That just shows topics that are strongly weighting in fake and real news. … We wanted to find the actual patterns in language that are indicative of those.”

Next, the researchers trained the model on all topics without any mention of the word “Trump,” and tested the model only on samples that had been set aside from the training data and that did contain the word “Trump.” While the traditional approach reached 93-percent accuracy, the second approach reached 87-percent accuracy. This accuracy gap, the researchers say, highlights the importance of using topics held out from the training process, to ensure the model can generalize what it has learned to new topics.

More research needed

To open the black box, the researchers then retraced their steps. Each time the model makes a prediction about a word triplet, a certain part of the model activates, depending on if the triplet is more likely from a real or fake news story. The researchers designed a method to retrace each prediction back to its designated part and then find the exact words that made it activate.

More research is needed to determine how useful this information is to readers, Boix says. In the future, the model could potentially be combined with, say, automated fact-checkers and other tools to give readers an edge in combating misinformation. After some refining, the model could also be the basis of a browser extension or app that alerts readers to potential fake news language.

“If I just give you an article, and highlight those patterns in the article as you’re reading, you could assess if the article is more or less fake,” he says. “It would be kind of like a warning to say, ‘Hey, maybe there is something strange here.’”

Joining the dots in large neural datasets

by Sabbi Lall |

You might have played ‘join the dots’, a puzzle where numbers guide you to draw until a complete picture emerges. But imagine a complex underlying image with no numbers to guide the sequence of joining. This is a problem that challenges scientists who work with large amounts of neural data. Sometimes they can align data to a stereotyped behavior, and thus define a sequence of neuronal activity underlying navigation of a maze or singing of a song learned and repeated across generations of birds. But most natural behavior is not stereotyped, and when it comes to sleeping, imagining, and other higher order activities, there is not even a physical behavioral readout for alignment. Michale Fee and colleagues have now developed an algorithm, seqNMF, that can recognize relevant sequences of neural activity, even when there is no guide to align to, such as an overt sequence of behaviors or notes.

“This method allows you to extract structure from the internal life of the brain without being forced to make reference to inputs or output,” says Michale Fee, a neuroscientist at the McGovern Institute at MIT, Associate Department Head and Glen V. and Phyllis F. Dorflinger Professor of Neuroscience in the Department of Brain and Cognitive Sciences, and investigator with the Simons Collaboration on the Global Brain. Fee conducted the study in collaboration with Mark S. Goldman of the University of California, Davis.

In order to achieve this task, the authors of the study, co-led by Emily L. Mackevicius and Andrew H. Bahle of the McGovern Institute,  took a process called convolutional non-negative matrix factorization (NMF), a tool that allows extraction of sparse, but important, features from complex and noisy data, and developed it so that it can be used to extract sequences over time that are related to a learned behavior or song. The new algorithm also relies on repetition, but tell-tale repetitions of neural activity rather than simplistic repetitions in the animal’s behavior. seqNMF can follow repeated sequences of firing over time that are not tied to a specific external reference time framework, and can extract relevant sequences of neural firing in an unsupervised fashion without the researcher supplying prior information.

In the current study, the authors initially applied and honed the system on synthetic datasets. These datasets started to show them that the algorithm could “join the dots” without additional informational input. When seqNMF performed well in these tests, they applied it to available open source data from rats, finding that they could extract sequences of neural firing in the hippocampus that are relevant to finding a water reward in a maze.

Having passed these initial tests, the authors upped the ante and challenged seqNMF to find relevant neural activity sequences in a non-stereotypical behavior: improvised singing by zebra finches that have not learned the signature songs of their species (untutored birds). The authors analyzed neural data from the HVC, a region of the bird brain previously linked to song learning. Since normal adult bird songs are stereotyped, the researchers could align neural activity with features in the song itself for well-tutored birds. Fee and colleagues then turned to untutored birds and found that they still had repeated neural sequences related to the “improvised” song, that are reminiscent of the tutored birds, but the patterns are messier. Indeed, the brain of the untutored bird will even initiate two distinct neural signatures at the same time, but seqNMF is able to see past the resulting neural cacophony, and decipher that multiple patterns are present but overlapping. Being able to find these levels of order in such neural datasets is near impossible using previous methods of analysis.

seqNMF can be applied, potentially, to any neural activity, and the researchers are now testing whether the algorithm can indeed be generalized to extract information from other types of neural data. In other words, now that it’s clear that seqNMF can find a relevant sequence of neural activity for a non-stereotypical behavior, scientists can examine whether the neural basis of behaviors in other organisms and even for activities such as sleep and imagination can be extracted. Indeed, seqNMF is available on GitHub for researchers to apply to their own questions of interest.

Ila Fiete

Neural Coding and Dynamics

Ila Fiete builds tools and mathematical models to expand our knowledge of the brain’s computations. Specifically, her lab focuses on how the brain develops and reshapes its neural connections to perform high-level computations, like those involved in memory and learning. The Fiete lab applies cutting-edge theoretical and quantitative methods—wielding the vast capabilities of computational models, informed by mathematics, machine learning, and physics—digging deeper into how the brain represents and manipulates information. Through these strategies, Fiete hopes to shed new light onto the neural ensembles behind learning, integration of new information, inference-making, and spatial navigation.

Her lab’s findings are pushing the frontiers of neuroscience—while advancing the utility of computational tools in this space—and are building a more robust understanding of complex brain processes.

Josh McDermott

The Science of Hearing

Hearing enables us to make sense of our whereabouts, understand the emotional state of others, and enjoy musical experiences. Acoustic information relays vital cues about the world—yet much of the sophisticated brain system that decodes this information is poorly understood.

Josh McDermott’s research is in search of foundational principles of sound perception. Groundbreaking discoveries from the McDermott lab have clarified how people hear and process sounds. His research informs new treatments for those with hearing loss, and paves the way for machine systems that emulate the human ability to recognize and interpret sound. McDermott’s lab has also pioneered new approaches for understanding music perception. His lab deconstructs the neural ensembles that allow us to appreciate music, while also studying the often striking variation that can occur across cultures.

Virtual Tour of McDermott Lab

Rebecca Saxe

Mind Reading

How do we think about the thoughts of other people? How are some thoughts universal and others specific to a culture or an individual?

Rebecca Saxe is tackling these and other thorny questions surrounding human thought in adults, children, and infants. Leveraging behavioral testing, brain imaging, and computational modeling, her lab is focusing on a diverse set of research questions including what people learn from punishment, the role of generosity in social relationships, and the navigation and language abilities in toddlers. The team is also using computational models to deconstruct complex thought processes, such as how humans predict the emotions of others. This research not only expands the junction of sociology and neuroscience, but also unravels—and gives clarity to—the social threads that form the fabric of society.

Virtual Tour of Saxe Lab

Mehrdad Jazayeri

Neurobiology of Mental Computations

How does the brain give rise to the mind? How do neurons, circuits, and synapses in the brain encode knowledge about objects, events, and other structural and causal relationships in the environment? Research in Mehrdad Jazayeri’s lab brings together ideas from cognitive science, neuroscience, and machine learning with experimental data in humans, animals, and computer models to develop a computational understanding of how the brain create internal representations, or models, of the external world.

Nancy Kanwisher

Architecture of the Mind

What is the nature of the human mind? Philosophers have debated this question for centuries, but Nancy Kanwisher approaches this question empirically, using brain imaging to look for components of the human mind that reside in particular regions of the brain. Her lab has identified cortical regions that are selectively engaged in the perception of faces, places, and bodies, and other regions specialized for uniquely human functions including the music, language, and thinking about other people’s thoughts. More recently, her lab has begun using artificial neural networks to unpack these findings and examine why, from a computational standpoint, the brain exhibits functional specification in the first place.

Tomaso Poggio

Engineering Intelligence

Tomaso Poggio is one of the founders of computational neuroscience. He pioneered a model of the fly’s visual system as well as of human stereovision. His research has always been interdisciplinary, bridging brains and computers. It is now focused on the mathematics of deep learning and on the computational neuroscience of the visual cortex. Poggio also introduced using an approach called regularization theory to computational vision, made key contributions to the biophysics of computation and to learning theory, and developed an influential model of recognition in the visual cortex. Research in the Poggio lab is guided by the belief that understanding learning is at the heart of understanding both biological and artificial intelligence. Learning is therefore the route to understanding how the human brain works and for making intelligent machines.

Mark Harnett

Listening to Neurons

Mark Harnett studies how the biophysical features of individual neurons, including ion channels, receptors, and membrane electrical properties, endow neural circuits with the ability to process information and perform the complex computations that underlie behavior. As part of this work, the Harnett lab was the first to describe the physiological properties of human dendrites, the elaborate tree-like structures through which neurons receive the vast majority of their synaptic inputs. Harnett also examines how computations are instantiated in neural circuits to produce complex behaviors such as spatial navigation.

Virtual Tour of Harnett Lab