Nishmar Cestero
Weekly Journal
Week 1: May 27-31
On Monday, I flew in from Boston and moved into a Columbia brownstone. I share my room with a roommate, but it's spacious with its own bathroom and kitchen. The only downside is that we need to provide everything from linens to cookware. However, the location is very convenient-- I'm just three streets away from the Columbia main entrances, and also within a two to three street radius of several supermarkets and an Asian convenience store.
I met my mentor Dr. Michelle Levine on Tuesday. Dr. Julia Hirschberg was away on a business trip this week; in fact, the lab was mostly empty. I spent the day getting paperwork in. It's important that I get my UNI, so I can complete an online human subjects training, since I will be running a lot of experiments. For the rest of the week, I read the grant proposal for the Deception Experiment and several research papers relating to it, and reviewed research methods.
Week 2: June 3-7
This week was very busy-- it's where the real work begins! On Monday morning I met with Dr. Hirschberg, Dr. Levine, and the other students working on the Deception Experiment. We talked logistics: the goal for the summer is to collect data and get to feature analysis. So that we could get familiar with the procedures involved in the experiment, the new interns and I participated in the experiment as test subjects. I was surprised to find out that the experiment usually runs for nearly two full hours. It's a really fun experiment to participate in, though much harder than it seems at first glance due to the high cognitive load (subjects are asked to press T and F keys to indicate whether they are telling the truth or lying within lies). It truly is very difficult to tell when someone you don't know is lying or not; I felt very uncertain about most of my answers. It turns out that (as expected of most people) I'm terrible at detecting lies. The rest of the week was spent on training. I learned the procedures and how to handle the equipment used in the experiment and ran many practice sessions.
Week 3: June 10-14
We scheduled our first pair of participants for this week and ran the experiment. Even though I've completed the necessary training at this point, I was not directly involved in running the experiment because I'm still not approved due to issues with the IRB protocol being outdated. There is a lot of interest in participation, and once I get permission we will start running more experiments. Meanwhile, I continued going through files in Praat and checking if they can be used. This week was also the first lab lunch of the summer, where I got to know some of the other people working here. We will be doing presentations about our respective projects in the future during this time.
Week 4: June 17-21
I'm finally on the protocol, so we scheduled and ran five experiments. Everything went very smoothly this week. Unfortunately, once we finished going through the sound files in Praat, we found that a lot of them can't be used because the gain on the mic was not adjusted properly to each participant. We have taken measures to make sure that the gain is set correctly by doing a volume check with each participant before the experiment.
Week 5: June 24-28
This week we ran more experiments. Unfortunately, it didn't go as smoothly as we had several participants who forgot about our appointment or had to cancel but didn't notify us. We prepared a brief presentation about our project based on the grant proposal, our mentor's own presentation, and our own experience running the experiments. You can view the presentation here: Deception Presentation .
Week 6: July 1-5
Interest has slowed down for the week. We only ran 3 experiments total. The rest of our time was mostly spent logging the data from the experiments. I ran a set of participants on Friday on my own for the first time (we usually run experiments with a partner). Things did not go completely smoothly due to reasons beyond my control. One of the participants arrived almost half an hour late and didn't understand that he was supposed to lie on the questions that we indicated; instead, he randomly lied and told the truth throughout the interview.
Week 7: July 8-12
I ran another experiment on my own on Monday, which went by a lot more smoothly. One of the participants did surprisingly well: he got the other participant to believe 10 out of 12 lies he told despite being thoroughly interviewed for an entire hour, and he made very accurate judgments about whether the other participant was lying or not. He might be the participant with the best overall performance so far, so it was interesting to be in the sound booth for that. We ran 6 experiments during the week. Also, scheduling has gotten more difficult now that we are sharing the sound booth with another experimenter. Soon, we'll be sharing with a third. Also, we have completely caught up with data logging at this point.
Week 8: July 15-19
This week was very productive. A lot of people signed up, and we were able to run 6 pairs. A couple of people canceled, but we were able to find participants from neighboring labs to fill their spots. I also recruited some students I met on Monday to participate in the experiment.
Week 9: July 22-26
We ran a total of 4 experiments this week (we had to cancel 2 of 6 scheduled). We had a participant who was very interested in the study and kindly offered to flyer in her dorm building, which is the international house. We've gotten a few emails from people who saw our flyers there, so she definitely helped us out a lot. Additionally, on Friday we had an interesting pair. One of the participants got his partner to believe all 12 out of 12 of his lies. This is probably the first time that happens.
Week 10: July 29- August 2
This is my final week on the DREU program. However, I will continue to work on the project as a summer employee for another two weeks. We ran 5 more pairs this week. During the summer we've gotten a total of 70 participants and jus over 21 hours of speech. We need to focus on getting more specific pair types. The breakdown for pair types is as follows:
- English, MF: 7
- English, MM: 2
- English, FF: 2
- Mandarin, MF: 3
- Mandarin, MM: 3
- Mandarin, FF: 1
- Eng/Mand, MF: 5
- Eng/Mand, MM: 4
- Eng/Mand, FF: 8
We don't have any results yet because this is a large study that will take several years. However, I'm very interested in this project and would love the opportunithy to work on it again next summer.