[NASW] Talking machines

by Stephanie Chasteen on October 26, 2008

[CASW New Horizons: How talking machines can manipulate our brains, Clifford Nass]

I usually hate hearing about robots, I find pure technogeek stuff incredibly boring. But talk to me about research on the human brain and behavior and I’m rapt. This was an incredible talk with a constellation of research studies on how speech and interactivity with robots or other electronic “assistants” affect people and how we react to them.

First, people are incredibly strongly wired to react to language. By the age of one day, infants can distinguish the sound of speech, and by 3 days old they can distinguish their native tongue from other languages. At hte age of 18 months, they learn a new word, on average, every 2-3 hours. Wow! We use voices to identify how someone’s feeling, their gender (within 1 second!), age, and personality. We’re so good at analyzing voices that if the high frequencies characteristic of female voices are removed electronically from a voice recording, we can still determine the gender of the voice.

Talking cars

He did a bunch of research on what kind of voice is best to have giving directions and instructions to drivers. He played us the recordings of the two voices used for the study, both recorded by the same young woman. One was chirpy and professional. The other one was so Eeyore-like that we all burst out laughing. BUT, he found that when people were unhappy, they preferred listening to that morose voice, and had fewer accidents, liked the voice better, and were less distracted. After all, when you’re bummed out, do you want to hear some chirpy person reminding you that you’re down and they’re not?

Gender stereotyping

I’ve written about stereotyping and stereotype threat before. There is also a lot of gender stereotyping that occurs on the basis of voice. In fact, male and female voices are processed in different portions of our brain! He told an incredible story about a car with a talking GPS in Germany, with a female voice, that had to be recalled because male drivers wouldn’t take directions from a woman! In his study, he randomly assigned a male or female “avatar” to subjects as they tok an electronic math quiz. The picture of their cartoon-like male or female avatar showed up on the screen, and they were told they were working with two others on the quiz — both of whom were the opposite gender from the test subject’s avatar. He found that even with this randomly assigned gender, that all the normal effects of gender showed up — regardless of the subject’s actual gender! In particular, those with a male avatar did better on the math test (that’s actual performance on the test!), thought they did better, and were more competitive. Wow.

Remaking Microsoft’s Help Assistant

In an effect he called the movie critic effect, people see disagreeable people as more intelligent than those who are agreeable all the time. After all, a movie critic that hates all the movies is seen as smarter than one that likes all of them. They also believe people when they’re self-critical. So, if you’re modest, people will like you, but not necessarily see you as being very smart. He found, appropriately, that when a voice recognition system blamed the user for not speaking clearly, it was seen as more intelligent than if it blamed itself for not correctly processing the user’s voice.

On a similar note, if someone seems to not mirror your emotions, this can be annoying. If someone is cheery in the face of your frustration, you just want to punch them in the nose. He applied this same principle to the oft-reviled Microsoft Help Assistant. When the Help Assistant asks “Was my answer helpful?” users have the choice to answer “yes” or “no.” When they answered “no” the Help Assistant replied “That makes me really angry! Let’s tell Microsoft how poor their software is” and opened a Mail Window for you to send an email to the company. As you typed, the paperclip urged, “Don’t hold back! Let them really have it.” They found that users loved the new paperclip, though they then hated Microsoft. Finding the right scapegoat is important. Note that CNN reported on the killing of the overly cute Help Assistant about 10 years ago. Thank god.

Multitasking

Another fascinating area of study, he found that there are fundamental differences in the brains of people who are high multitaskers. This is particularly relevant as I sit here blogging on this conference, and other science writers are twittering. All classical psychology indicates that we shouldn’t be able to learn well while multitasking because we can’t attend to more than one thing at a time, our attention is selective. He wasn’t able to give many details on this research because it’s currently in press for Science magazine, but he found that high multitaskers show basic and fundamental cognitive differences in their distractability, information filtering, pattern recognition, and memory capacity. He used standard psychological tools like the Stroop test which tap into underlying cognitive abilities. These aren’t just changes in how people think about learning and thinking, but in how they process information on a deep and unconscious level. What does this mean for our teaching strategies? Something must change. People are unwilling to give up any of their access to media (look at me, blogging and twittering and reading my email while attending to this fascinating lecture), and this will have ramifications for teachers and society. We don’t know how education should look if it’s true that high multitaskers learn in fundamentally different ways than traditional learners!

Leave a Comment

Previous post:

Next post: