I am at the BCI Meeting 2010 in Asilomar, California. This is a meeting of over 250 people with an interest in BCIs and related systems. There are people from many different universities, companies, media entities, grant institutions, and a few others I haven’t figured out yet. Several of us from the Future BNCI consortium are here, as well as persons who work in related EU projects.
The conference seems to be going very well. There are a lot of engaging talks and posters, and a strong sense of excitement and anticipation. I would go so far as to say that our field is dancing around a Kuhnian paradigm shift. If so, then the next International Meeting should be even more exciting. If someone presents (for example) a large scale clinical validation of a BCI for rehabilitation, then the field will change dramatically. And a lot of people are working toward BCI for stroke recovery, as well as autism, addiction, ADHD, and psychopathy. There is also a lot of work toward practical BCI systems, which involve dry electrodes or are otherwise easier to acquire, set up, and use.
Much of the conference involves workshops. In each workshop, about 8-20 people discuss specific issues, and then present their views to the entire plenary session. This is similar to the workshops from the 2005 conference, but on a much larger scale. I am watching all this eagerly, since we’ll have a similar workshop structure at our September FBNCI conference.
Conferences are another important part of a scientific research career. Science does not proceed in a vacuum. Scientists need to know about research that other scientists produce - as well as other things such as what works, what does't work, new tools available, or new job opportunities. Similarly, scientists need to tell other people about their research to gain publicity, defend their ideas, and be visible. In addition to just exchanging information, getting together helps to develop professional relationships, and direct discussions at conferences often lead to new research collaborations, grant proposals, scientific articles, etc.
Last year, there was one large international BCI conference, which was hosted by the BCI group in Berlin. The "Links" section of this page has links to the lab and to online talks from that conference. This year, we already had the First TOBI Workshop in Graz in Feb. 2010. About 135 people attended. There will be another international BCI conference beginning on May 31 that will be about twice as large. More information can be found elsewhere on this website.
What do people do at these conferences? Most conferences include these elements:
1) Talks. Scientists present a specific research study, a series of studies, a new idea or model, an argument or perspective, a review of work across different groups relating to a specific research direction, etc. These talks are usually followed by a brief question and answer session.
2) Posters. Scientists will bring posters with text, pictures, and sketches that describe their work. The conference organizers provide space where conference attendees can mount their posters, and the schedule includes "poster sessions" in which people will walk around the poster area, view different posters, and discuss them with their authors.
3) Discussion. Scientists will talk about one of the talks, the posters, or a specific topic that is usually announced ahead of time. Sometimes, the schedule will include discussion periods. These discussions may involve smaller groups. For example, panels or workshops may discuss specific topics that relate to a larger theme of the conference, and include experts in those topics.
Of course, regardless of the schedule, a lot of discussion just happens informally, such as during poster sessions, after talks, during coffee breaks, of just bumping in to people in the lobby. Conference attendees will also meet for lunch, dinner, and other typical social events. Many people at conferences are personal friends and/or current or former co-workers or collaborators, and are happy to meet and catch up.
4) Satellite events. Many conference feature satellite events. These are events that are not officially part of the conference, but may be of interest to conference attendees and address related issues. Satellite events usually occur just before or after the conference. These events may include the above elements like talks, posters, and discussions. Some satellite events are etraining events, where people can learn specific advanced material such as the latest new technologies in brain imagine or new drugs to treat certain disroderes.
Scientists have mixed feelings about conferences. On the one hand, they are essential events in a research career for many reasons noted above. Aside from professional reasons, it can be fun to travel and to see old friends, and learning about the latest research is fun to people who really enjoy it. On the other hand, conferences take a lot of time. They usuall last a few days. Plus, you have the added time of traveling, preparing, and recovering. Hence, most experienced scientists are very selective about the conferences they attend, and try to get the most benefit out of each one.
A career in BCI research involves a lot of work beyond running experiments. For example, you have to buy equipment, develop and configure software, get ethical approvals so you can conduct research with human subjects, write grant proposals to get money for research projects, and present your experiments and other work by writing paper, giving talks at conferences, visiting other research groups, etc. This blog entry focuses only on one core element of the job: conducting experiments.
The scientific method has four stages: make observations, form a hypothesis, test the hypothesis, and revise your hypothesis. A realworld scientific experiment typically does follow those stages, but most of the work is in the third stage. Making observations is easy. Forming hypothesis is easy. Testing and proving anything is quite hard. The last stage involves, at worst, admitting you're wrong, and then, at least you still proved something. The worst outcome of a scientific research project is none - if the results do not tell you anything, then you wasted your time, and probably did not design the study very well.
Once you have a hypothesis based on reasonable observations, you have to develop an infrastructure to test the hypothesis. For laboratory research, you need a building, electricity, time, some training, perhaps other staff, and other obvious things. You may need to buy hardware or software, and configure it. You need to develop a protocol - a document that describes what the participants in your research study have to do. Again, you want to develop a protocol such that your results can clearly prove or disprove your hypothesis. There should be different conditions in your protocol so you can study the effects of changing one or more factors (called controlled independent variables). You should probably test your protocol, hardware, software, and testing environment with a few initial participants, which is called pilot testing.
When your experiment is ready, you can run participants. A typical session proceeds as follows. The experimenter will show the person around the recording environment. The participant can see the electrode cap, computer, and other equipment. The experimenter explains the protocol and ethical issues and answers any questions. The participant then signs a consent form and is prepared for EEG recording. The experimenter puts an electrode cap on the particpant's head and squirts electrode gel under each electrode to get a good contact. The procedure typically takes about 20 minutes, although this can vary with many factors, such as the number of electrodes. Then, the participant follows the protocol, performing certain mental tasks at specific times or in response to cues like tones from speakers or arrows on a monitor. Afterward, the participant can wash his or her hair, and may complete a questionnaire before leaving. Participants are often paid for their participation.
After collecting data from enough subjects, the experimenter and/or colleagues must analyze the data. Which conditions produced which results? What is the best way to present these results - a graph, chart, table, etc? Do statistical tests validate the significance of your results? In addition to presenting the results, researchers also need to discuss them and why they matter. Are the results consistent with the hypothesis? Are the results consistent with other projects? What do these results mean for future research? Finally, these results must be shared with the scientific community through scientific papers, conference talks, or other mechanisms.
Here is an example of how these stages were followed in a classic research study in the literature. Prof. Neuper and colleagues made a couple related observations. First, some people seem to have trouble using BCIs that involve imagined movement. Second, some people interpret instructions differently. Some people think about moving their own hand, while other people think about watching someone move the hand. Prof. Neuper and her colleagues developed the hypothesis that different instructions could affect performance with BCIs that involve imagined movement. They developed a protocol to explore this hypothesis. People would participate in four different conditions. They would perform a different movement-related task in each condition. Fourteen people participated in ths study. Results showed that subjects performed better when they imagined moving than when they imagined watching movement. Therefore, the study produced useful results that could help revise the hypothesis, and help provide more accurate instructions to subjects in future studies, which could improve accuracy. The results were presented in a 2005 publication and other articles, as well as a conference presentation.