From Tech is Becoming Emotionally Intelligent, and it’s Big Business [David Pring-Mill: SingularityHub]
Affectiva is uniquely positioned to profit from this “emotional economy.” The company has already created the world’s largest emotion database. “We’ve analyzed a little bit over 4.7 million faces in 75 countries,” said Zijderveld. “This is data first and foremost, it’s data gathered with consent. So everyone has opted in to have their faces analyzed.”
The vastness of that database is essential for deep learning approaches. The software would be inaccurate if the data was inadequate. According to Zijderveld, “If you don’t have massive amounts of data of people of all ages, genders, and ethnicities, then your algorithms are going to be pretty biased.”
This massive database has already revealed cultural insights into how people express emotion. Zijderveld explained, “Obviously everyone knows that women are more expressive than men. But our data confirms that, but not only that, it can also show that women smile longer. They tend to smile more often. There’s also regional differences.”
Hrm. “Obviously everyone knows that women are more expressive than men” ? I guess I wasn’t thinking that, but maybe it’s a cultural stereotype.
We should be aware of what’s behind this finding of women smiling more than men. From Deutsch, LeBaron, & Fryer, 1987:
Women have been observed to smile more than men in a variety of social contexts. In order to investigate the consequences of this sex difference for the way men and women are perceived, male and female college students rated the characteristics of men and women depicted in verbal descriptions accompanied by photographs in which they either smiled or did not smile. In control conditions these targets were rated without accompanying photographs. The findings showed that the absence of smiles had a greater impact on perceptions of women than on perceptions of men. When not smiling, women were perceived as less happy, less carefree and less relaxed than were men. Moreover, nonsmiling women were rated less happv, less warm, less relaxed and less carefree than the average woman, whereas smiling men were rated more favorably on those traits than the average man. These results suggest that different standards are applied to men and women. If women fail to perform expressive and warm nonverbal behavior, they will be evaluated more harshly than men.
Women are pressured to smile and look pleasant. They are punished if they don’t do this. Bitchy resting face is a thing because women are expected to smile. The parallel for men — resting asshole face — has not enjoyed widespread popularity because men can look serious without being accused of being angry or depressed.
When I was younger, I was routinely told to “Smile!” by men, usually those much older than I was. Often they’d say “Smile! It’s not so bad!” When I wasn’t smiling I was rarely upset in any way. I was usually thinking about something intensely, as I am wont to do. It was annoying and patronizing when they commanded me to smile. It seemed they just wanted me to look pretty and pleasant for them. This is the way that our culture imposes a requirement of smiling onto women.
What does this all matter for emotional artificial intelligence? For one it demonstrates that facial expressions are not hard-wired displays of emotion. But if you’re been reading emotion / know, you knew that already! It’s a useful demonstration of how culture and emotion and expression interact.
I was curious to see if there’s research evidence that women are more expressive than men. Here’s a study from 2004 (Simon & Nath, 2004) that found that women and men in the U.S. do not differ in self-reports of everyday subjective feelings and do significantly differ in their emotional expressiveness. This study looked at the General Social Survey from 1996 and found that women were more likely than men to disagree with the statements, “I keep my emotions to myself” and “When anxious, I try not to worry anyone else.” Men and women did not differ on the statement “When I’m angry, I let people know.” Men and women also did not differ on the statements, “I don’t tell friends something upsetting,” “I try to be pleasant so as not to upset others,” and “I’m not afraid to show my feelings.” So there was some support for women’s being more emotionally expressive than men, but it was constrained to certain types of expressiveness, not a global tendency towards more expressiveness.
I imagine there are facial expressions related to emotion that men are more likely to show. Anger? Perhaps not. Simon & Nath, 2004 did expect to find that men would endorse the item “When I’m angry, I let people know” more often than women. They didn’t find this. They also didn’t find differences in the self-reported frequency of experience of anger between men and women, which is something else they thought they’d see, based on cultural theories of emotion in sway at the time the paper was written.