Tuesday, September 29, 2009

Reading Response: Affective and Social Responses to Computing

In today’s age, the word “social” is thrown around with reckless abandon to describe the rich interpersonal connections that the Internet provides as a platform for us to engage with each other in many ways. Of notable popularity in the last few years have been social web applications like Facebook and Twitter. As a society, we aren’t quite sure what to think about the relationship between computers and human sociology. One thing is for sure: these issues are now becoming a point of household dialog. But researchers have been intrigued by humans’ uncanny ability to socialize for a long time. Consider our historical relationship to inanimate items; while it has been commonly acknowledged that humans may have always established emotional bonds with some objects- especially when they represent some person, event, or idea- that relationship could hardly be classified as social. However, as objects get anywhere even remotely close to the likes of a human, an amazing transition occurs where we begin to socialize with them as if they were humans, expecting manners, and judging their personality characteristics. In the paper “Computers are Social Actors: A Review of Current Research,” Clifford Nass and colleagues present five experiments that serve as evidence that computers are indeed social actors- whether intended or not.

The paper offers two sides that have commonly been taken when discussions have taken place regarding humans’ social behavior towards computers: An attribution of youth or ignorance on the part of the human (which assumes that humans whom act socially with computers are either young or- frankly- stupid, and therefore represents a small set of the human population) or an indication that users actually interact with the creators/programmers of the computer via proxy. But Nass's studies show that social reactions and affectations toward computers are not representative of a small set of the human population. In fact, it is quite common. And in either case, both of these sides assume that people’s behavior matches their beliefs, which is uncharacteristic of humans and turns out to be false.

Essentially, the experiments that Nass and his colleagues present offer several key takeaways:

  1. Users exhibit politeness to computers, and further expect politeness in reciprocity
  2. Users respond to computer personalities in the same way they respond to human personalities
  3. Users are susceptible to flattery from computers (sincere or not)
  4. Users apply gender stereotypes to computers
  5. Users do not react socially to computers simply because they are thinking of the creators/programmers via proxy (a direct contradiction to one of the two sides that many fall into)

One major theme that ran across all studies was that people were (sometimes defiantly) unaware that they had treated the computer as a social actor. After all, what rational person would think that politeness is important when interacting with computers? Yet time after time (and supposedly unconsciously), participants would treat the computers with manners- as if they hadn’t wanted to hurt the computer’s feelings. Participants rated computers as more likable if they matched personality characteristics. They found themselves wooed by the charming flattery of the computers, even if they were told that the computers didn’t know better. It was as if people couldn’t help but to treat computers like their long lost Aunt Maude.

Nass and colleagues built many of their premises for presented experimental studies on established results from sociology and communication findings. Cleverly inserting the word “computer” where originally “human” was used allowed them to bootstrap concepts previously thought to only apply to human-to-human interactions and instead apply them to computers. This may have presented some issues, especially while taking complicated social concepts like dimensions of personality into the lab. For instance, in the second experiment the researchers based their premises on studies showing that the more similar people are, the more likely they will be attracted to each other- especially true with respect to personality. However, in their study the researchers only measured the variable personality across one dimension: dominant vs. submissive. I find it difficult to rely on one dimension in order to validate the hypothesis that similar personalities between computers and humans lead to attraction.

Nonetheless, this particular study did offer a valuable implication in that it showed how words alone can be used to engender personality and foster a relationship between a person and computer- whether it was intended or not. Words! These design elements that we use ubiquitously in the things we create. The diction we choose to label something or request input is either attributing to a user’s affection for a computer, or deteriorating it whether we like it or not. The extended implications of this: What happens when we introduce other design elements that foster personality traits in computers (as we arguably already have been doing for a long time)? Anthropomorphic technology can now use facial expressions and embodied gestures to communicate with humans. Does this require even more careful thought on our behalf as we begin to use higher fidelity ways of communicating (and inherently conveying personality traits), or is our job as “designers of the social” getting easier?

To step back, how does the common acceptance of Internet usage now change how people interact with the computer- that hunk of metal that still sits on our desks and on our laps? Recalling applications like Facebook and Twitter, it now seems arguable that people actually are looking through the computer and reacting socially with human counterparts on the other side. It appears the proximity effect of the messenger that was illustrated to us in the fourth study isn’t so ironclad any longer, and perhaps new experiments need to be conducted to test this.

I wonder how the results for re-administration of the 5 studies would fare today, given the explosion of the Web 2.0’s cutesy landscape, riddled with rounded corners, bright colors, and plain language copy. Surely we rely on a lot more cues in our designs to allow people to judge a website’s personality compared to 1994- when this paper was written. People can go to many-a-web-app, and get a load of positive feedback for simply registering for a free account. Yet, one can’t help but wonder if the Web 2.0 movement has relied too much on the social nature that computers can play. Websites have gotten pretty good at increasing conversion rates, but on average, people stop using their free online accounts just two weeks after signing up. It’s as if they fall in love, but it turns out to be just a fling, or worse- a one night stand! Have we gotten so good at designing the social into web applications that we’ve become players?! It’s dreadful to think that we may have been spending too much time and effort trying to engage the social aspects of our interactions and forgetting to address the content of the interactions in the first place.

No comments: