Faculty

Tom Griffiths

Tom Griffiths, Lab Director

(webpage)


Postdocs

Joshua Abbott

Joshua Abbott

(webpage) My research interests lie at the intersection of machine learning and computational models of cognition. In particular, I explore ways in which people reason with semantic knowledge; distinguishing between the structured semantic representations people use in memory and the processes for search and inference over these representations.


Dawn Chen

Dawn Chen

(webpage) I am interested in how humans learn and extract structure from their complex and noisy environment, often even without conscious effort, and how machines could emulate this remarkable behavior. My work combines computational modeling and behavioral experiments to shed light on how humans learn a variety of representations, including relational concepts necessary for analogy and heuristics for efficiently solving problems in a given domain. I have also begun to investigate the ways in which representations from deep learning capture and deviate from human cognition. I aim to apply the insights that I learn to improving math and science education and building more flexible and autonomous intelligent machines.


Aida Nematzadeh

Aida Nematzadeh

(webpage) My research interests are in the areas of cognitive modeling of language and computational semantics. I view human cognition as a complex computational process: the goal of my research is to provide a better understanding of the computational mechanisms underlying the human ability to learn and organize information. The primary focus of my research has been to explain how children learn their language, without explicitly being taught, through observation of and interaction with other people. Currently, I am exploring how semantic representations learned by neural networks perform in predicting human behavior.


Alexandra Paxton

Alexandra Paxton

(webpage) I am interested in exploring human communication in data-rich environments. From capitalizing on large-scale real-world corpora to capturing multimodal experimental data, my research seeks to understand how context changes communication dynamics. Broadly, my work integrates computational and social perspectives to understand interpersonal interaction as a nonlinear dynamical system. Relatedly, I also develop research methods to facilitate quantitative research on interaction, and I am currently working as part of the Center for Data on the Mind to foster the application of big data to cognitive questions.


Daniel Reichman

Daniel Reichman

(webpage) I'm a theoretical computer scientist interested in average case analysis of algorithms. My research has touched several fields including random graphs, cascading dynamics over networks, percolation theory and computer vision. I'm especially intrigued by computational problems related to human and artificial intelligence and interdisciplinary research involving theoretical computer science, AI and cognitive science.


Jordan Suchow

Jordan Suchow

(webpage) My research examines the perceptual and cognitive processes that affect our ability to see, remember, and learn from the world around us, and then develops technologies and techniques to overcome these limitations. The hallmark of this work is combining empirical methods from some of the best-understood domains of cognitive science, such as vision and memory, with formal frameworks developed outside the field — mathematical biology, machine learning, computer science, and network science. My work additionally makes contact with applications in computer vision, human factors & ergonomics, design, and technologies for learning and achieving mastery. The output of my research takes varied forms — journal articles, methods, software, visual demonstrations, and inventions.


Graduate Students

David Bourgin

David Bourgin

Creativity is considered by most to be an essential component of human intelligence. Where do new ideas come from? How do we search for and identify concepts during brainstorming? How do we develop creative intuition? In my research I employ techniques from machine learning and statistics to develop and evaluate computational accounts of these and other phenomena. Most recently I have been interested in improvisation as a window into creative processing and have been working with musicians from Berkeley's Center for New Music and Audio Technology to develop formal models for aspects of this process.


Fred Callaway

Fred Callaway

Intelligent agents must continually respond to and learn from their environment. Mathematical models from Bayesian statistics and reinforcement learning can provide optimal solutions to these problems; but they are often intractable to compute. How do humans find good approximations to these optimal solutions using limited neural resources? In particular, how do they balance the competing goals of learning, deciding, and conserving resources? I aim to study this question with game-based empirical experiments and computational models inspired by machine learning algorithms.


Rachit Dubey

Rachit Dubey

(webpage) Curiosity is one of the hallmarks of human intelligence and is crucial to scientific discovery and invention - yet our understanding of curiosity remains quite limited. What is the function of curiosity? How does curiosity develop? My goal is to better understand such aspects of curiosity and also explore how curiosity relates to cognitive processes such as creativity and metareasoning. By studying curiosity through a computational lens, I intend to develop a better theoretical foundation of curiosity which can then lead to applications in various pedagogical settings.


Monica Gates

Monica Gates

(webpage) Listen in on any two people talking on the street, and you'll witness one of the greatest feats of human evolution. Alice may be telling a story about what Bob said to Carol, but within every sparse sentence, omitted intention, and nevertheless correct reaction, we see evidence of social inference: how we interact and learn about people. I'm interested in approaching social cognition from a computational perspective, using probabilistic models and large-scale web-based crowdsourcing to investigate the computational goals and algorithms driving the social mind. (outreach site)


Jessica Hamrick

Jessica Hamrick

(webpage) My interests lie at the intersection between cognitive science and artificial intelligence. Broadly, I want to understand how people integrate perception and reasoning to understand the world around them. I am particularly interested in computationally specifying the algorithms underlying reasoning, as well as the knowledge representations accessed by those algorithms, by drawing on methods from machine learning and statistics. I have most recently focused on how people reason about every day physical events, modeling this "intuitive physics" through knowledge-rich simulations consistent with Newtonian physics.


Rachel Jansen

Rachel Jansen

(email) My research uses methods from machine learning and probabilistic modeling to study mathematics education and cognition. I am specifically interested in understanding more about how people learn math so that I may work towards improving both teaching practices and online educational tools. One branch of my research is centered around math learning in adults using an adaptive online algebra tutor. I use this technology to explore ways in which we can influence motivation and alter students' perceptions of mathematics, to ultimately remove emotional and psychological barriers so that more people may appreciate and excel at the subject. I also work with children in preschools and science museums in the hope of discovering how they develop an idea of what math is, or a "math concept" and how this interacts with their feelings towards it.


Thomas Langlois

Thomas Langlois

My research interests are centered primarily around human perceptual processes, and visual perception in particular. I combine methods from probabilistic modeling, machine learning, computer vision and psychophysics in order to uncover human inductive biases in this domain. I am currently using iterated learning techniques combined with non-parametric statistical methods to uncover spatial memory priors and prototype effects in a variety of visual (and non-visual) domains. I am also interested in visual preferences (aesthetic biases), and multi-modal perception. A long term research goal of mine is to understand why humans exhibit these perceptual biases in terms of a rational analysis (Anderson), and ultimately to determine how an understanding of human perceptual representations (including inductive biases) can inform the development of improved artificial intelligence systems.


Falk Lieder

Falk Lieder

(webpage) My research in computational positive psychology combines mathematical modeling, behavioral experiments, machine learning, and artificial intelligence to help people make better decisions. I approach this goal from four complementary angles: effective cognitive strategies, strategy selection learning, cognitive training, and cognitive prosthetics. To build a solid theoretical framework for effective interventions I develop a mathematical theory of bounded rationality: resource-rationality. I am currently testing and refining this theory by applying it to explain cognitive biases and how people choose cognitive strategies. I will apply this theory to foster adaptive flexibility in decision making under uncertainty and discover effective problem solving strategies. My long-term goal is to translate the results of my research into practical strategies, tools, and interventions that help people realize their potential. Let's make the world a smarter place!


Stephan Meylan

Stephan Meylan

(webpage) I investigate the complex relationship between natural languages and cognition, particularly how commonalities in the world's languages emerge from domain-general learning and processing. I am particularly interested in how people learn and use so-called 'closed-class' words (articles, pronouns, prepositions, etc.). Current work includes inferring children's early grammatical representations from developmental corpora, characterizing sound change in a probabilistic framework, and developing computational models of word sense induction. In my research I use (mostly Bayesian) models of concept learning, large-scale web-based experiments, and methods from natural language processing.


Joshua Peterson

Joshua Peterson

(webpage) I am interested in how intelligent systems learn, represent, and integrate complex information, particularly in the domains of vision, music, education, and their intersections. Most recently, I've begun to explore the correspondence between representations learned by humans and those learned by state-of-the-art machine learning algorithms, with the dual aim of informing both classic psychological theories and human-centric AI.


Undergraduate Honors Thesis Students

Alumni and Long-Distance Affiliates

Joe Austerweil

Joe Austerweil

(webpage) As a computational cognitive psychologist, I explore questions at the intersection of perception and higher-level cognition. I use recent advances in statistics and computer science to formulate ideal observer models to see how they solve these problems and then test the model predictions using traditional behavioral experimentation. This method yields novel machine learning methods and leads to the discovery of new psychological principles. I have three main lines of research: (1) understanding how representations are constructed, (2) exploring how concept learning affects similarity, and (3) investigating the interconnection of perceptual reference frames and higher-level cognition.


Vincent Berthiaume

Vincent Berthiaume

(webpage) I seek to uncover mechanisms of cognitive development by combining infant studies and computational models of normal and abnormal development. While my previous computational endeavors have structive neural networks, I am now looking into the possibility of creating a bayesian/connectionist hybrid developmental framework.


Wesley Baraff Bonawitz

Liz Bonawitz

(webpage) My research bridges two traditions: Cognitive Development and Computational Modeling. By working with both methods, I hope to understand the structure of children's early causal beliefs, how evidence and prior beliefs interact to affect children's learning, and additional developmental processes that influence children's belief revision.


Daphna Buchsbaum

Daphna Buchsbaum

(webpage) I'm interested in how children and adults understand and learn from other people's behavior, and ultimately developing intelligent computer programs with some of these same social learning abilities. Currently, my research is focused on the problem of action segmentation: when faced with a continuous stream of behavior, how do we identify individual, meaningful actions? How do we decide what the effects of these action sequences are? I'm exploring how both preschoolers and adults use causal structure and statistical patterns in human motion to help understand other's behavior, as well as using Bayesian computational approaches to try and model human action understanding.


Kevin Canini

Kevin Canini

(webpage) I'm interested in using tools from statistical machine learning to build probabilistic models of human cognition. My thesis work focuses on developing nonparametric Bayesian statistical models of the ways that people learn and represent categories of objects. I'm also broadly interested in topic modeling and other dimensionality reduction techniques.


Daniel Chada

Daniel Chada

I am interested in how computer systems and models can help explore aspects of human cognition. This endeavor leads to the intersection between cognitive science and artificial intelligence. I aim to explore the emergent aspects, as well as the high (computational) level phenomena of human information processing. How do memory, perception, analogy-making and representation interplay to allow for the range of characteristics that humans display? How do we build and manipulate structured representations with such flexibility? How are we able to make and understand analogies of such immense conceptual complexity with ease and accuracy?


Naomi Feldman

Naomi Feldman

My interests are in speech perception and language acquisition. I'm using computational and behavioral methods to look at how people organize speech sounds into categories and how those categories affect their ability to perceive differences between sounds.


Sharon Goldwater

Sharon Goldwater

My research interests include language acquisition, computational linguistics, phonology, and morphology.


Chris Holdgraf

Chris Holdgraf

(webpage) I hope to understand the mind from many different perspectives, with the goal of linking higher-level theories of the mind with our understanding of lower-level neuronal functioning and systems. Some of my broad research interests are learning, perception, decision-making, and creativity. I also love writing about science and and getting other people as excited about the natural world as I am!


Anne Hsu

Anne Hsu

(webpage) I am interested in human statistical learning using computational modelling and web-based experiments. In particular I've been investigating language learning and models of categorization. I've also been examining how people learn omission exceptions to general rules in cognitive domains.


Tiffany Hwu

Tiffany Hwu

My current research interests are in using computational models of cognition to improve music recommendation systems. I'm also interested in brain-computer interfaces and other means of using technology to enhance human performance.


Nori Jacoby

Nori Jacoby

(webpage) I'm interested in exploring the role of culture in auditory perception. My current work uses iterated learning alongside classical psychophysical methods to characterize perceptual biases in music and speech rhythms in various populations ranging from Westerners to the Tsimané, an Amazonian foraging-farming society in Bolivia. I'm also working on computational modeling of synchronization and entrainment in jembe drum ensembles in Mali. My previous work focused on the mathematical modeling of sensorimotor synchronization in the form of tapping experiments as well as the application of machine-learning techniques to model aspects of musical syntax ranging from tonal harmony to birdsong and the perception of musical form.


Chris Lucas

Chris Lucas

(webpage) I'm interested in how and to what extent the abstract knowledge that constrains human induction is acquired. My current research focuses on causal induction, but I'm also interested in categorization, language, and the neural machinery behind human induction.


Jay Martin

Jay Martin

(webpage) I am interested in using probabilistic models to investigate the structure of natural categories.


Luke Maurits

Luke Maurits

(webpage) My research to date has focused on functional psycholinguistics, using mathematical and computational models to explore how the syntax of language is influenced by two important requirements of language: facilitating efficient and robust communication between agents, and integrating closely with non-linguistic representations in the rest of the mind. More generally I am interested in finding ways in which seemingly arbitrary aspects of language may in fact represent rational/optimal/adapative solutions to certain computational problems. I hope to expand my research to non-linguistic subjects in the near future: I'm interested in issues of semantic representation (especially "language of thought" ideas) and theory formation.


Thomas Morgan

Thomas Morgan

(webpage) My interests lie in human psychological evolution with regards to culture and the impact culture has on phenotypic and genetic evolution. I am particularly interested in the nature and evolution of the psychological mechanisms that underlie how individuals learn from others and allow human culture to evolve cumulatively. My previous work has investigated the strategic use of social information in adults as well as in young children and the role of teaching and language in the transmission of Oldowan lithic technology. My work in the Computational Cognitive Science Lab investigates how individuals copy complex motor behaviors and how different means of information transmission impact the ratcheting of cumulative culture.


Michael Pacer

Michael Pacer

(webpage) People figure out what matters - they can toss away heaps of junk data while finding invaluable treasures of predictable regularity. Obviously, this isn't always easy; if it were, we scientists would be out of work. Still, even very young children are remarkably good at this, and even formal science often begins as informal intuition. Particularly astonishing is our facility in reasoning about high-dimensional spaces, like those involved in causal and social inference (not to mention the dimensional explosion resulting from combining the two). These issues (and those like them) are what drive me and inspire my research.


Avi Press

Avi Press

(webpage) My interests lie between cognitive science and artificial intelligence. How do people (and how can we make machines) effortlessly learn the vastly complex concepts and relationships that exist in our everyday lives? My current project is concerned with this very question, and aims to explore how theory and concept learning take place in real time.


Anna Rafferty

Anna Rafferty

(webpage) I'm interested in probabilistic models of human learning and how we can apply those models to automatically teach people more effectively. My recent work has involved modeling pedagogical problems as well as looking at iterated learning of language and the exploration of inductive biases.


Florencia Reali

Florencia Reali

(webpage) My research combines behavioral experiments and probabilistic models to study various aspects of language learning and processing. I am also interested in exploring some theoretical aspects of language evolution, including the interaction between cultural transmission, biological adaptation and individual learning.


Adam Sanborn

Adam Sanborn

I am interested in how perceptual categories are built and structured and how people use these categories to make decisions. Using Bayesian methods and behavioral experiments, I am developing methods for efficiently learning about natural categories, as well as exploring rational models of categorization and intuitive dynamics.


Benj Shapiro

Benj Shapiro

My interests lie at the intersection of statistical learning theory and computational linguistics. I am also broadly interested in the advancement of decision-making techniques for combinations of natural language and medical data, especially big data systems for predicting the impact of genetic mutations on disease. My current work explores the theory of uniform information density and its impact on communication and language evolution.


Lei Shi

Lei Shi

I'm interested in using Bayesian methods to model human cognitive processes. My current project is making connections between exemplar models and Monte Carlo sampling methods. When I have time, I also wonder how Bayes' rule is implemented in the brain.


Andrew Whalen

Andrew Whalen

I'm interested in computational models of social learning. My research currently involves population level models of human behavior and social learning. I am interested in the role that social learning might play in cultural and language evolution.


Joseph Jay Williams

Joseph Jay Williams

(webpage) I'm interested in understanding the role that explanation plays in learning, by drawing on computational modeling, verbal-conceptual theories, and empirical research.


Frank Wood

Frank Wood

My research effort is directed towards both contributing models and algorithms to the field of statistical machine learning and figuring out how the brain works.


Jing Xu

Jing Xu

I'm interested in understanding how cognitive control plays its role in people's daily interaction with the external world, from coordinating and controlling our sensory-motor system to learning and switching among tasks and categories. Particularly, at this interface, how the prior knowledge and constraints in people's mind influence people's behavior. I'm using computational and mathematical approaches to model behavioral and neural empirical data.


Saiwing Yeung

Saiwing Yeung

In everyday life people regularly have to make inferences that go beyond the data that are observable, by drawing on their prior beliefs and knowledge. I attempt to characterize these beliefs and knowledge, including those in the form of culture, that underlies these inductive inferences and to explain how they might be learned, applied, and transmitted. I am currently working on projects that study causal reasoning, trend prediction, and learning of probabilistic data.


Julia Ying

Julia Ying

I am interested in iterative learning in context of cultural transmission, as well as the Sapir-Whorf hypothesis. My research specifically focuses on discovering the prior expectations people hold towards music and how that expectation influences the shaping and transmission of music.


© 2017 Computational Cognitive Science Lab  |  Department of Psychology  |  University of California, Berkeley