Robtos: Trust me, I'll get you a beer!

Robots are becoming useful now that engineers are realizing it's all about the human interaction.



Summary of the Boston Globe article from the ACM News:

Robot May Furnish Lesson in Human Trust
Boston Globe (07/05/10) Johnson, Carolyn Y.

Boston-area scientists are using a new robot to study the signals that people use to decide whether to trust one another within minutes of meeting. "There should be some signal for trustworthiness that's subtle and hard to find, but [it is] there," says Northeastern University's David DeSteno. The robot, called Nexi, has advantages over human participants because people use subtle gestures, or engage in unintentional mimicry, that can be hard to measure or control. Nexi has many human expressions, but researchers can control every aspect of its behavior, which enables them to test what nonverbal cues might seem more or less trustworthy. The research also could help roboticists find ways to design machines that will be trusted partners for humans. The experiment is a collaboration between the Massachusetts Institute of Technology, Northeastern, and Cornell University. At the end of the experiment, researchers measured how trustworthy the participants found Nexi to be using an economic task in which they decided how many tokens to exchange with Nexi and predicted how many tokens Nexi would give them.

If that creeps you out too much, maybe just try having your computer to get you a beer.

The Real "Avatar" (The Last Airbender)

I saw The Last Airbender this weekend with my 13 year old and his friends. If you don't know, the movie is based on the cartoon called Avatar : The Last Airbender. I call it the "real" Avatar because Avatar the blockbuster movie hardly reflects the meaning of the word. The word "Avatar" comes from the Sanskrit word for "descent," meaning a spirit (or "God"), descending to the material world as a human. You may think such a notion is preposterous, or you may have deep faith that some Avatar is your personal savior (like Jesus Christ or Krishna), but in either case, an Avatar, fictional or factual, is more closely related to the Last Airbender than the Marine in the virtual blue suit.

As you might expect, I enjoyed The Last Airbender more than Avatar. Avatar was three hours of over-the-top drama. The Last Airbender was far more subtle. Instead of yelling and screaming and in-your-face interaction, Aang, the main character's strongest imperative was telling an animal "Be nice."

Noah Ringer, who played Aang, was phenomenal. I was expecting to be disappointed, as most were by Hayden Christenson playing the adult Anakin Skywalker. Monks with power are very difficult to play, but Ringer did what great actors Ewan MacGregor and Liam Neeson could not pull off - maintain a constant state of grace and even temper while exercising power in difficult circumstances (Alec Guinness, the original Obi-Wan Kenobi, maintained grace, but his style was better suited for stage than film).

The theatergoers, including my triad of 13 year olds, did not like the movie. They picked on the things missing from the cartoon, like the villian's missing ponytail or the fact that Aang was pronounced more properly like a mantra than like it was in the cartoon (rhymed with "hang"). After comparing Aang to the Jedi, I was met with disbelief by the 13 year olds. "Aang is not a monk - he's a warrior!" That led to a 15 minute diatribe about the Bhagavad Gita's ("What? What do you mean we've never heard of the most widely read book ever?!") Arjuna the Archer, Joseph Campbell's Hero with a Thousand Faces, and Star Wars. I think I got enough into their minds that it will spark their interest when they come across it again sometime.

China's Government Advocates for Workers

According to this IBD story, "Chinese workers are usually represented by government-sponsored unions. "

Now I suspect this may not be all it's cracked up to be since "Many recent labor disturbances were organized outside these communist-backed groups." But still can you just imagine a government advocating on behalf of workers? I suppose if you remember America in the 1930's or 1940's, you might remember, but in my lifetime, government has been 98% a tool of business interests - the capitalists - not labor. I know there was once or twice a raise in the minimum wage (fat chance of that happening again now that Ted Kennedy is dead). And I am well aware that those who rely on FOX News for their worldview think labor and teachers have an undue influence on government, but just that fact that the notion that the government advocates for labor is a foreign idea to most Americans (and anathema to many) shows how tilted the power in this democracy has become. Some say America is now Fascist, who by definition "seek to organize a nation on corporatist perspectives, values, and systems such as the political system and the economy."

Sure there's corruption in China, as there is in the U.S., but at the end of the day, the Communists truly do look out for the workers. This may be idealogical, but is likely just as much a matter of self-preservation - the Chinese government is well aware that without economic growth, the people would revolt and who knows what will happen then. In any case, it will be fascinating to see how the U.S. corporate-oriented government will compete with the Chinese worker-oriented government over the next 50 years.

Pre-Internet Publications

As I was writing a blog post on switching to the Mac, I started discussing my first programming on the Mac. Since it was pre-Internet (or at least pre-browser), I didn't think I'd find any about the articles that come out of my first professional post-B.S. work, but I did come up with the references via David A. Rosenbaum's C.V.:

Rosenbaum, D. A., Engelbrecht, S. E., Bushe, M. M., & Loukopoulos, L. D. (1993). A model for reaching control. Acta Psychologica, 82, 237-250.

Rosenbaum, D. A., Engelbrecht, S. E., Bushe, M. M., & Loukopoulos, L. D. (1993). Knowledge model for selecting and producing reaching movements. Journal of Motor Behavior, 25, 217-227.

Bushe, M. M., Vaughan, J., & Rosenbaum, D. A. (1994). Pascal external functions for Strawberry Tree's "Analog Connection Workbench." Behavior Research Methods, Instruments, and Computer, 26, 461-466.

This was some of the most memorable and fun work I've done. The model for reaching control was a computer model that acted like a baby at the start. This baby would randomly reach in different directions using different combinations of angles of various limbs. Each angle would be given a cost so that moving your torso was more costly than your wrist, and bending at angles that were difficult (like scratching your back) was also costly. After a few thousand trails, the model learned to reach for objects at novel locations using fluid lifelike movements. Lots of hours with Mac programming manuals always handy.

The "Pascal" paper was a different experiment. We were seeing how well someone could mimic tapping out a beep-beep rhythm with varying delaying between the beeps (.1-1.0 seconds or so) when heard aurally or seen visually on a screen. Aural won. The article was about writing Mac software to hook up with a hardware/software package called "Strawberry Tree." It involved some low level z80 processor instructions to pick up analog signals from a external board hooked up to a Mac. Can't say it was all that fun, but doing the dirty work made it publishable.

Ultimately, though I'm always dissatisfied by cognitive psychology experiments. They are too contrived. "Life is a bowl of concurrent schedules," as my behaviorist professor Dr. John Donahoe, liked to say. Looking back, I can honestly say that he was the most influential professor I had. I think I had the best work-study gig on campus as I wrote a neural network simulation for Dr. Donahoe. A harbinger was the nights that I was up until 2AM writing the visuals. Dr. Donahoe didn't think the visuals were that important, he really just wanted the numbers, but I wanted to see the network learning. I still remember how his face lit up when the dark lines lit up yellow as connections were made. An early positive reinforcement that has been reinforced many times since in my UI and visualization work.

Williams Syndrome

Williams Syndrome, is a genetic condition that causes a lack of social fear - people with Williams Syndrome are really friendly even with people they don't know. Fascinatingly, they also have no or nearly no racial bias, from a recent Discover Magazine article:

Typically, children start overtly gravitating towards their own ethnic groups from the tender age of three. Groups of people from all over the globe and all sorts of cultures show these biases. Even autistic children, who can have severe difficulties with social relationships, show signs of racial stereotypes. But Santos says that the Williams syndrome kids are the first group of humans devoid of such racial bias...

It's very interesting to think about the fact that we are born with a predilection to racial bias. It makes more sense when I think about what I read in Guns, Germs and Steel that said when two tribal people who don't know each other meet each other in Papua New Guinea, they will first start naming off people who are related to each other to see if the other person knows them. If they find a match, they are OK, if they don't find a match, they fight. I guess those conversations could last a very long time.

Serendipitously, I've never heard of Williams Syndrome until this morning. Then Dylan and I visited UMass for an Open House. I talked to Dylan about the story on the drive to UMass (hmmm, is this the kind of conversation that inspired Dylan to want to teach science?). Then not one, but two different speakers mentioned Williams Syndrome in passing during their speeches. Weird. One was talking about the opportunities at UMass and mentioned how he worked with local kids with Williams Syndrome. I forget the context the other speaker mentioned it in. Our tour guide came close to the concept as she was doing her thesis on social contact.

Why Alex Trabek is So Smart

Alex Trabek, the longtime host of Jeopardy! seems like a very smart guy. He knows the answers to so many questions, "Oh, I'm sorry, it was Victor Hugo." (He does get a bit swarmy when the answer is about something French).

Category: Education Level
The answer is : 6
The question: How many DAYS did Alex Trabek spend in college?

Now, of course, not going to college doesn't make you not smart, but could it be that we think Alex Trabek is smarter than he is?

I don't have the reference, but one thing I remember from my psychology (neuroscience) degree was that there were some studies done in the 80's that were in a game show format. Participants either asked trivia questions or gave answers, and they switch roles for different audiences. Audiences were asked to rate how smart they thought the participants were - both the questions givers and the answer givers. The audience thought the people asking the questions were smarter than the ones answering the questions, no matter who was asking. The numbers were quite extraordinary.

We have a strong tendency to think the person asking the questions is smart - which is why the crooks at Enron could convince other that they were the "Smartest Person in Room."

This came up when a friend and collegue, who is wicked smaht - he went to a Ivy League school, he's a genius,and he's very technically proficient - was asked some questions about algorithms in a techie interview and didn't quite have the answers (neither would I, I only minored in math/computers). Most experienced folks know that you can't figure out someone's ability by a few technical questions in an interview. That strategy only assesses the intersection of knowledge of the interviewer and the interviewee, both of which can be vast and non-intersecting. This tactic makes the questioner feel real good about themselves since they knew something the other person doesn't. It deflates the person being interviewed, who is already nervous.

I approach it differently. I look for what I know about what the person claims they know. I then drill down into what they did to see if they had a thought process to intelligently make the decisions they made given the constraints they were under. This gives the person confidence since they are talking about something they know, not something you pulled out of the air. It allows you to have a humane interaction, and assess the person under a situation that would be more normal. Hopefully, you also talk enough tech to realize how knowledgeable the person and if what their resume says holds water.