Undergraduate Research: The Temporal Dynamics of Learning Center at UCSD

The Temporal Dynamics of Learning Center at UCSD


The Temporal Dynamics of Learning Center’s (TDLC) Research Experience for Undergraduates (REU) Site program is now accepting applications for the 2013-14 academic year.  The traineeship is nine months in duration, from September, 2013, to June, 2014. Trainees are expected to work 10 hours per week in their host lab, and will receive a $5,000 stipend for their time, plus additional funds ($1,100) to cover travel expenses to an academic conference. The program is focused on training undergraduates to perform research at the highest level in our center’s labs, located at the University of California, San Diego and the Salk Institute for Biological Studies.


The intellectual focus of the research experience will be based upon the research focus of the Temporal Dynamics of Learning Center (TDLC): the role of time and timing in learning. This is an especially broad topic, reflecting the size and scope of the center. We study timing in learning at multiple temporal scales – from the millisecond level as reflected in spike-timing dependent plasticity in synaptic modification, up to the month and year level as reflected in long-term effects of the spacing of study episodes in fact learning. Our research also spans many spatial scales and systems, from the scale of neurons up to the scale of social interactions between teachers and students.

The REU program’s trainee projects span a broad range of disciplines, represented by the following UCSD departments:

Cognitive Science


Computer Science and Engineering

Electrical and Computer Engineering


Program trainees will also receive professional development training, including a GRE preparation course and a research ethics workshop, among other activities.

Please visit our website, where you will find detailed information about the program: http://www.tdlc-reu.ucsd.edu/index.html.

To go directly to the application page, please follow this link: http://www.tdlc-reu.ucsd.edu/application/application-overview.html.

Any questions about the program can be directed to the program coordinator, Andrew Kovacevic, at akovacev@eng.ucsd.edu or (858) 822-1981


The following projects will be available for the 2013-14 TDLC REU program (The projects are always in flux, so they are subject to change and therefore may not be exactly as described below by the time you begin the program):

Theme(s): Behavioral Neuroscience, Neurobiology of Learning and Memory, Sequence Learning.
Mentor: Andrea A. Chiba, PhD; Associate Professor of Cognitive Science, Program in Neuroscience, Computational Neuroscience Program, and Interdisciplinary Program in Cognitive Science. Science Director: Temporal Dynamics of Learning Center.
Rationale/Motivation: Due to the pervasive nature of time, much of our sensory input takes the form of sequences of events. Our primary sensory cortices participate in multiple forms of coding, in order to accurately represent the timing of these sensory events. A complete characterization of the way in which sensory sequences are coded by the cerebral cortices would represent a leap in understanding how the dynamics of our sensory world are coded in the brain.
Objective: The scientific objective of this project is to gain further understanding regarding the interplay of the sensory thalamus, the sensory cortex and the reticular nucleus of the thalamus in coding temporally modulated sensory (auditory or visual) sequences of events across different brain states.
Specific Role of the REU Student: The REU student will be expected to gain a conceptual understanding of basic sensory processing in the brain. The REU student will work alongside an advanced graduate student in setting experimental parameters for the sequential presentation of sensory events. The student will assist in rodent neurosurgery; learn rodent neuroanatomical techniques, including retrograde tracing, immunocytochemistry and basic histochemistry. The student will assist in undertaking electrophysiological recording experiments. The student will assist in spike sorting and waveform analyses, in addition to basic analysis of electrophysiological data.

Using compounds to reverse long-term potentiation and study memory reorganization in a rodent model of human memory impairment
Theme(s): Memory Modulation
Mentor: Robert E. Clark, Ph.D.<http://psychiatry.ucsd.edu/faculty/rclark.html>, Professor of Psychiatry
Rationale / Motivation: A fundamental question about memory is how it is organized and stored in the brain. Life experiences are not formed and stored instantaneously in a form that persists as long-term memory. Instead, new memories are transformed gradually from a labile state, where they are vulnerable to disruption, to a more permanent state in which they can persist indefinitely and are resistant to disruption. This transformation is often referred to as consolidation and involves reorganization at both synaptic and brain system levels. System consolidation, the focus of the Clark laboratory, is a prolonged process that can take days, weeks, or longer and involves a gradual reorganization of multiple brain circuits that support long-term memory; this process appears to involve time-dependent modifications in circuits that support memory storage and recall.

The Clark laboratory is currently using a zeta-inhibitory peptide (ZIP) to study systems consolidation in rats. The ZIP infusion method does not inactivate or damage the target structure, but rather it reverses established long-term potentiation (LTP). LTP is a persistent enhancement of synaptic transmission widely studied as a physiological model of memory. Accordingly, if training an animal induces LTP in, for example, the hippocampus, and if LTP is a critical part of the memory representation, then reversing LTP should impair memory retention.

Rats are used as a model system for human memory function. Rats are trained on various tests to measure memory. Then ZIP is infused directly into different brain structures during brain surgery. Later memory retention is tested. The brains are then analyzed.
Objective: We are seeking to understand the dynamic nature of memory and memory reorganization across time.
Specific Role of the REU Student: Students will be engaged in all aspects of these tasks, including neurosurgery, histology, and behavioral testing.

Eye Movements for Visual Tasks: Models and Experiments
Theme(s): Computational Cognitive Neuroscience
Mentor: Garrison W. Cottrell Ph.D<http://www-cse.ucsd.edu/users/gary/>., Professor of Computer Science and Engineering, and Director, The Temporal Dynamics of Learning Center
Rationale/Motivation: As a result of having a foveated retina, we actively move our eyes to direct the highest resolution of our visual processing towards interesting things. We make about three saccades each second; it is a decision we make about 172,000 times a day. Hence it is of great interest to understand how we decide where to look and how we integrate the information acquired during a fixation. We have been approaching this issue through building computational models that acquire visual information through fixations, and integrate that information in order to recognize objects. We are also conducting human experiments using eye trackers to test our models.
Objective: Our goal is to understand how the human visual system chooses fixation points in an image, how visual routines for fixations are learned, and how visual representations are learned.
Specific Role of the REU Student: A computationally oriented student will learn how to build and analyze computational models of visual processes. A behaviorally oriented student will learn how to conduct and analyze human experiments using our eye trackers.

Human Computation for Annotating Audio Clips
Theme(s): Human-Computer Interaction, Applied Statistics
Mentor: Gert R.G. Lanckriet, Ph.D.<http://www.ece.ucsd.edu/~glanckriet/>, Assistant Professor of Electrical and Computer Engineering
Rationale/Motivation: Audio data is inherently temporal. It is currently still a major challenge, however, to develop useful features for audio data classification. To build sparse statistical models for audio classification, a sufficiently large collection of accurate training data, i.e., audio clips with high-quality annotations, is essential. Applying data mining techniques to music-related, social or other web sites doesn’t provide the required accuracy. On the other hand, controlled human surveys provide high-quality data but are prohibitively expensive to provide the required quantity. Combining the large-scale strength of online data collection with the precision and reliability of human annotation motivates an alternative approach, based on human computation. This emerging field of research leverages the vast human connectivity offered by the Internet to gather incremental, high-quality data from huge numbers of online participants. Participants are motivated to contribute for free by disguising tasks as online games.
Objective: The goal is to design and implement an appealing human computation game that attracts broad interest, and harnesses the “wisdom of the crowd” to collect reliable annotations for 20-30,000 audio clips.
Specific Role of the REU Student: The REU student will first design a simple but attractive online game that revolves around annotating audio clips. After designing the game, the student will implement and deploy it. Third, the student will learn some basic data analysis techniques and apply those to the data collected through her/his game.

Computational Analysis of nonverbal behavior in adaptive teaching
Theme(s): Human-computer interaction, adaptive tutoring systems, nonverbal behavior, computer science, cognitive science
Mentors: Marian Bartlett, Ph.D.<http://mplab.ucsd.edu/~marni/index.html>, Associate Research Professor; Javier Movellan, Ph.D.<http://mplab.ucsd.edu/>, Full Project Scientist, Institute of Neural Computation, UCSD.
Rationale/motivation:To date, there is a paucity of empirical data to support understanding nonverbal behavior in teaching at a computational level. This project will help advance the science of learning by improving our understanding of the dynamics of nonverbal behavior in teaching at a computational level, across multiple time scales: From low-level facial movements on the time scale of tens of milliseconds, to cognitive and affective processes with time scales of seconds, to higher level strategic behaviors operating at longer time scales.
Objective: The goal of this project is to develop computational models of the nonverbal behavior and interactive strategies observed during face-to-face teaching. These computational models will serve as a foundation for a new generation of embodied teaching agents that approximate the benefits of face-to-face human tutoring.
Specific role of the REU student: Participating REU students will be involved in the collection and analysis of student-teacher interaction data. This will include analysis of facial expressions from video, as well as motion capture data and eye movement data. Students will also participate in video data annotation.

Sensorimotor Learning of Facial Expressions: A Novel Intervention for Autism
Theme(s): Autism, facial expression, computer science, cognitive science, psychology
Mentors: Marian Bartlett, Ph.D.<http://mplab.ucsd.edu/~marni/index.html>, Associate Research Professor; Javier Movellan, Ph.D.<http://mplab.ucsd.edu/>, Full Project Scientist, Institute of Neural Computation, UCSD.
Rationale/motivation: Children with autism spectrum disorders (ASD) are impaired in their ability to produce and perceive dynamic facial expressions. Automated facial expression recognition technology opens up new possibilities for clinical research, assessment, and intervention systems.
Objective: The goal is to develop a computer assisted intervention system focusing on facial expression production to enhance the facial expression skills of children with ASD. In this project, automated facial expression recognition will be employed for the development of training exercises for facial expression production. In addition, the project will also characterize differences between facial expressions of children with ASD and typically developing children using statistical models, explore interdependencies between voluntary and spontaneous facial expression production systems, and measure changes in physiological responses to expression stimuli, which may be related to empathy and social functioning, including expression mirroring and mu suppression.
Specific role of the REU student: Participating REU students will develop facial expression intervention games using automated facial expression recognition technology developed in our laboratory. They will also participate in the development and analysis of tests of facial expression recognition and production for children with autism as well as typically developing children.

Brain Dynamics, Action, and Learning
Theme(s): Neuroscience, Bioengineering, Neural Computation
Mentor: Howard Poizner, Ph.D.<http://inc2.ucsd.edu/poizner/>, Research Scientist, Institute of Neural Computation, UCSD and Director, the TDLC Motion Capture/Brain Dynamics Facility.
Rationale/Motivation: Neither spike timing studies in animals nor hemodynamic studies in humans can capture the rapid cortical dynamics accompanying either sensorimotor or reward-based learning. We are using our newly developed NSF Motion Capture/Brain Dynamics Facility to combine high-resolution EEG analysis with fine-grained temporal analysis of motor performance in sensorimotor and reward-based learning tasks. Our tasks can be conducted in 3D, multimodal, immersive virtual environments for complete experimental control. We model the learning process using such algorithms as the temporal difference learning model, and examine the relationships between behavioral choice, movement kinematics, EEG patterns, and parameters of the model.
Objective: To elucidate the brain dynamics that underlie decisions, actions, and reward.
Specific Role of the REU Student: Participating REU students will learn about EEG recording and analysis, 3D multi-joint motion capture, creation of virtual environments, and/or computational modeling as they perform research on the brain dynamics underlying action and learning.

From Sequential Patterns of Caregiver Behavior to Infant Social Learning and Social Emotions
Themes: Social development, time-series analysis, sympathetic nervous system activation, social neuroscience, infant language development
Mentor: Gedeon Deák, Ph.D.<http://www.cogsci.ucsd.edu/~deak/>, Associate Professor of Cognitive Science and Human Development, UCSD
Rationale/Motivation: During the first 24 months human infants learn complex skills for social interaction. This project uses new methods to explore the mechanisms by which infants interact with adults. By collecting electroencephalogram (EEG), electrocardiogram (ECG), and motion capture of infants and parents while they are interacting in social games, we assess the relations among brain, behavior, and body during “in situ,” meaningful interactions.
Objective: To demonstrate and model a new “active cognitive-neuroscince” paradigm for studying social development in infants. The goal is to test theories of social-learning and early development of social perception and action, by cross-referencing data from the brain (EEG), peripheral signals of emotion and arousal (ECG), and patterns of movement and emotional expressions. The sources of individual differences in infants’ social skills are considered.
Specific Role of the REU Student: The REU student will be expected to gain a conceptual understanding of EEG methodology and joint attention literature. The student will work alongside an advanced graduate student and a trained research associate. The student will assist in preparing equipment for sessions and helping to run experimental sessions. The student will also learn basic analyses of electrophysiological data and coding of observational data. Finally, the REU student will attend weekly meetings with faculty mentor and research associates, and lab Journal Club meetings.

Role of Fast-spiking Interneurons in Generating Gamma Oscillations
Themes: Systems neuroscience, neural mechanisms of behavior.
Mentor: Terrence Sejnowski, Ph.D.<http://www.salk.edu/faculty/sejnowski.html>, Professor of Biology, UCSD; Professor, Salk Institute, and Margarita Behrens, Ph.D., staff scientist, Salk Institute.
Rationale/Motivation:Parvalbumin (PV) positive interneurons are involved in the generation of gamma oscillations, which regulate working memory and information transmission between cortical areas. In particular, synaptic inhibition from PV-interneurons controls the firing rates of pyramidal neurons, synchronizes spikes within populations of neurons, and participates in the development of executive functions associated with prefrontal brain regions.
Objective: Characterize the gamma oscillations in transgenic mice with altered NMDA receptors in PV-interneurons.
Specific Role of the REU Student: The student will participate in EEG recordings from transgenic mice with altered NMDA-subtype of glutamate receptor on the FS-interneurons and testing of these mice on behavioral assays for sensory, motor and cognitive functions.

Salience and learned associations in visual search
Themes: Systems and computational neuroscience.
Mentor: Terrence Sejnowski, Ph.D.<http://www.salk.edu/faculty/sejnowski.html>, Professor of Biology, UCSD; Professor, Salk Institute, and Leanne Chukoskie, Ph.D., Asst. Project Scientist, UCSD.Rationale/Motivation: Traditional studies of how people choose where to look consider the low-level salience information in the scene almost exclusively despite the tremendous influence high-level factors brought by the particular task. People use scene gist information to narrow the region of search, for example, looking along the side of the road to find street signs. These different factors come together seamlessly to produce the saccadic eye movements that we produce 3
times a second.
Objective: Characterize the contribution of low and high-level information to visual search in traditional and novel paradigms.
Specific Role of the REU Student: The student will assist in collecting eyetracking data, analyzing and modeling the data. Extensions of the analyses and model may also be applied to rat search data that is being collected in comparable tasks.

Can we predict whether someone will remember the next presented item from their EEG? — Applications of Brain Computer Interfaces
Mentor: Virginia de Sa, Ph.D.<http://www.cogsci.ucsd.edu/~desa/>, Associate Professor of Cognitive Science and Member of the Graduate Program in Neurosciences.
Themes: Bioengineering, Machine Learning, Electrophysiology
Rationale/Motivation: Brain-computer interfaces (BCIs) have led to developments in single-trial analysis and classification of EEG signals. We can use these in applications for healthy people by monitoring their EEG in real time. In this application we want to see if we can improve people’s memory of items by presenting them at times when we believe (from their EEG) they are best able to remember them.
Objective: We propose to investigate the EEG dynamics before, during and after presentation of items (pictures and words), analyze the signals, and create machine learning classifiers to predict when subjects are in a “good brain state” for remembering items.
Specific Role of the REU Student: The student will assist graduate students in designing the experiments, preparing the subject (placing and testing electrodes), recording electrophysiological data, and analyzing the data. Students with a good mathematical and computer programming background will benefit most from this opportunity.