Using Physiological Arousal as a Marker for Emotional Experiences

 Summer 2010 - Present

As customers, products and services are an experience. Standing in a long line for a coffee may be a part of a frustrating experience. Seeing a picture of your family during a phone call may be a happy one. How do we measure these emotional experiences with products and services?

In this study, the LEGO group and I measured people's electrodermal activity—a marker of an arousing event—to enhance post-interviews about user's experiences: people provided deeper, more informative stories when asked to describe specific arousing moments.

Specifically, I designed and ran an experiment measuring the experiences of 50 different people with a new LEGO product. I conducted open ended interviews and asked people about specific moments when their EDA increased. I generated an internal paper identifying 17 key insights on how the LEGO Groups new product affects people's emotions. I worked with designers, writers, and administrators in realizing this project. In addition, to a report, I taught the LEGO group how to use these tools independently for future projects.

Masters Thesis: IN‐SITU MEASUREMENT OF ELECTRODERMAL ACTIVITY DURING OCCUPATIONAL THERAPY

Click on Picture for PDF
(Summer 2009- Spring 2010)

How could knowing a child’s electrodermal activity (EDA) during therapy change occupational therapy practices? As a joint project between the MIT Media Lab and the Sensory Processing Disorder Foundation, we measured the EDA responses of 22 children participating in occupational therapy for 77 hours. We saw children lying in a ball pit and their EDA correspondingly decreasing; we saw children’s EDA rapidly increase while they were swinging; and we saw many other patterns of responses from different therapeutic activities. Could we show how different guided activities changed children’s arousal with a statistical test? Analysis suggests that many factors play a role in changing children’s EDA, and these factors could not all be taken into account with our low sample size – making any statistical analysis inconclusive. As more tools begin to measure physiological data over longer periods of time, in less controlled settings, we should not forget to build a base understanding at the individual case level.

Three Key Factors that Influence EDA in Observational Studies of Occupational Therapy

Click here for Full PDF

Fall 2010

This poster was based off of my Masters Thesis

Selected Writings of 2009

Alexithymia Grant Application 

Using MANOVA to detect SPD  

Learning to Read EDA Pre-Trial 2009 

IMFAR Abstract 2009

Presentations 2009

Speaker for Bank of America, Customer Service Summit. “Using Emotions to improve service”. February 3, 2009. Charlotte, North Carolina

Keynote Speaker for the SPD Foundation 9th International Symposium. “Taking Away the Guesswork:Objective Measurement of Internal State During Occupational Therapy”. October 8-10,2009. Chicago, Illinois.

Keynote Speaker for Think, Create, Engineer. “The Power of Being a Creative Engineer”.  October 29-Novermber 1, 2009.  Karnataka, India.

Academic Posters 2009

Imfar 2009
 Using Machine Learning can we separate children with SPD and ASD?

ACII 2009 
Measuring Arousal Almost Anywhere

Are you Excited?

Summer 2009- Spring 2010  with Dr. Rosalind Picard and Dr. Lucy Miller

Have you ever worked with a child and he just went out of control- running around, not paying attention, or just completely spaced out?  Children with ADHD, Autism, and Sensory Processing Disorder can struggle with this daily.  Many of these children I worked with could not go into a regular classroom as they would just "act out".  As you can imagine their behavior can devastate their self confidence and social life.

For my Master Thesis, I measured 85 hours of Electrodermal Activity  on both the child and therapist during hour long therapy sessions.  While, I am still analyzing the data, initial results show that some activities (in this picture swinging) are effective in lowering the child's arousal.  While this coincides with therapist's theories, this evidence can be shown to parents and teachers to help them understand how our behavior is linked to our internal state which can be controlled in fun ways.

The research I have done so far was included in the application for Sensory Processing Disorder to be a part of the DSM IV.

I worked closely with Dr. Lucy J. Miller and 5 amazing interns over the Summer of 2009 to make this project work.

Weekly Survey Measures in Clinical Settings

Fall 2009- Spring 2010  with Dr. Rosalind Picard and Dr. John Weisz

One of the hardest challenges psychological clinics face is determining the effectiveness of their interventions.  This makes being creative and trying new solutions a difficult prospect.  Working with Proffesor John Weisz and the Judge Baker Children's Center, I have provided technical solutions to their new endeavor.  The group hopes to have parents fill out surveys weekly reporting how their symptoms are doing.

While the concept is extremely simple, the challenge is to motivate busy parents to fill out the information.  I am supposed to enter my workout minutes each day and I hardly ever do that. How can we motivate parents to do this?  Using principles of Motivating Technologies by BJ Fogg, I hope to introduce new ways to motivate parents.  Simple solutions could involve having a picture that grows and changes each time a survey is filled out.  You can also compare a user's input to the average.

Unlike traditional engineering though, I hope to find solutions by working with parents who do struggle.  I know what motivates me but finding out what motivates them is my end goal!  By developing the solution off of real needs (instead of theoretical needs), I hope to find an effective solution to this basic problem.

Measureable results for Occupational Therapy

Winter 2009- Spring 2009 with Dr. Lucy J. Miller and Dr. Rosalind Picard

Working with the SPD Foundation, I experimented with how our sensors can work with Occupational Therapists.

I tested our sensors vs. their gold standard sensors and then made observations on how the children's Electrodermal Activity changed while in therapy.

One child got extremely nervous everytime he had to speak into a microphone, but bowling and drinking water made him calm way down.

Another child would go into sparadic shifts with his EDA going way below normal and then way above normal.

While these observations in themselves do not lead to conclusions, they helped me develop a thesis topic and also to observe how functional our sensors are.

I appreciated this project because the therapists I was working with were excited (even more than I was). They had so many questions they wanted answer, they dreamt about how the technology could be used. Instead of forcing an awkward solution, I was providing means for therapists to answer their own questions.

Measuring Emotional Responses with Wireless Electro-Dermal Activity

Fall 2008-Present

Advisor: Dr. Rosalind Picard

Click here for Research Web Page

Using features similar to a lie detector: Electrodermal Activity, Heart Rate, Temperature, and Vibration

I have helped develop wrist bands that can output these measurements live.

This project is currently being used with various projects including baby shoes and students with autism

I personally developed the software that can read the incoming data and flexibly display it in a variety of formats.

The laptop used is the XO (One laptop Per a Child). Not only is it affordable, but it is also very durable and can take quite a beating, which is great for 24 hour measurements!

EDA has been an expensive tool where people are required to be in a stringent setting: don't move sit personally still.

By making affordable, wireless, sensors, I hope to uncover new ideas for psychology by providing live measurements that can be taken 24 hours a day.

High Speed Digital Design

Fall 2005, Spring 2006

Working with Dr. Melinda Piket-May, I spent a year researching electric magnetic wave simulations. I received: $1000 Learning Opportunity Grant and an $800 UROP grant to work on a programming environment for complex EM analysis. I used the LC programming language, which was developed as part of the CRAY institute.

IBM research

Summer 2006

Working Dr. Joan Mitchell at IBM, I analyzed the effects of color management on grayscale printing. Working on an IBM speed team (3 interns and 2 employees), to help IBM analyze the benefits of switching from color to grayscale.

As mass printing is switching to color printing, large amounts of investment will need to be made to switch company's logos/brouchures/etc. from grayscale to color. To alleviate this overhead cost, IBM is encouraging its clients to submit color images and still print them in gray, so that the transition will be more seamless.

IBM was curious to see how the quality of the image would change when color images were printed in grayscale. This involved using "color management" that shifted images from color to grayscale.

My role was to develop algorithms to print the colored images. As this was not on the market yet, I had to develop my own printer code (like assembly language) to send color images across and have them print using the "color management" feature. This was quite challenging as I had to be resourceful on how to print AFP code - often times talking to the people who created the standards for the code for advice.

In the end I was able to create a Java program, that would create a customized AFP code that could then be sent to programs.

I helped develop a Flash presentation for the final project.

Overall, our groups research was published at IBM and is now used by sales represenatives to show the

Experimenter Effect with EDA

Spring 2007

One of my favorite classes I ever took looked very thoroughly at all the research done on Psychic research. Feomm telekinetic to precognition, we analyzed the research done with thorough scientific reasoning. As a side project I ran my own Psy experiment. We tested the effects of remote staring. I constructed a galvanic skin response that could read people's physical response to stimuli (this same contraption is used in lie detector tests).

Recruiting over 100 people, including 4 PhD's, we ran our own official psychic experiment. In the end we were invited to present our results to the Society of Scientific Exploration at a conference in Utah. Below I have some of programs I developed for this research. Whether you believe in psychic phenomena or not, hopefully you can appreciate my MatLab code and Flash skills:

Flash Presentation

This presentation was used to demonstrate our results at a scientific conference. I don't know how much sense it will make without me explaining the experiment :).

Matlab Data Analysis

This Matlab file took a LARGE amount of data, and formatted with various forms of Fourier Transforms and derivatives. We were then able to analyze our results visually and numerically.

Python User Interface

This user interface made it so various doctoral physicists could run these experiments. (Which they did).

Wireless Tracking through Robotics

Summer 2005

Working with Dr. Dirk Grunwald and Dr. Greg Grudic I spent Summer 2005 developing a robot to detect wireless internet signals.

• recieved $4,500 NSF grant for undergraduate research
• developed a planning and control system for the ER1 robot
• collected data to be used with a Nearest Neighbor Algorithm.
  
• helped show how individuals could be tracked through wireless internet signal strength

We used the nearest neighbor algorithm to predict specific locations where the robot would be. Our research showed that wireless signals bounce too much for precise location.

Through this research I bumped into many obstacles: the robot's direction was skewed on carpet, developing the interface, and analyzing the loads of data I received.


I programmed the robot to talk with people from remote text commands. In addition the robot could sing and dance which was fun to do while collecting data to unsuspecting prospective students.

The above data shows what my robot analyzed. Each one of these was one wireless stations strength, with over 200 actually being analyzed. As you can see the variablity of strength made the data fairly inconsistant. If I were to continue on this experiment, I would attempt to use more sophisticated machine learning algorithms (perhaps Randomized Forrests) to deal with the chaos of the data.