I just finished reading The Media Equation by Reeves and Nass and Wired for Speech by Nass and Brave.
Fascinating! And probably useful whenever I get to doing some voice, animation and/or video-including interfaces.
From the Media Equation: Have some people watch news and entertainment programs on a TV with a sign on top of it "News Television" and news and entertainment programs on an identical TV right next to it with a sign on top of it labeled "Entertainment Television". Mix up which people see which programs on which TV. Ask people to rate the programs they watched, and they rated the news programs shown on the "News" TV more enjoyable, trustworthy, etc. than the ones shown on the "Entertainment" TV and vice versa. Stunning to me.
From Wired for Speech: Have some people listen to the same five positive book reviews read either by a single voice or by five different voices. The people who hear the same reviews read by five different people rate the books as more appealing, etc. than the people who hear the identical content read by a single voice. Again, stunning.
What follows are some of my notes to myself to remind me of the findings in brief.
From The Media Equation. (Generally, substitute "computer" for "person" in social psychology findings and get essentially the same results, hence 'equation'.)
Manners
Politeness: People give more positive and homogenous answers when a computer asks the person to evaluate the computer than when another computer asks to evaluate the first computer. Grice's Maxims for politeness: quality, quantity, relevance and clarity.
Interpersonal distance: Pictures of people presented closer will have more intense evaluations, pay more attention to them and remember them better.
Flattery: People flattered by a computer will believe they performed better and like the computer more, whether or not the praise is warranted. People will think they did better when a computer criticizes them baselessly than with cause. People will like a computer better when it praises them than when it criticizes them.
Judging: Performance praised by a computer will be judged better than criticized performance. Computer self-praise will be judged less valid than when praise comes from another computer. A computer than praises another computer will be liked more than a computer that praises itself. A computer that criticizes another computer will be liked less than a computer that criticizes itself; it will also be judged more intelligent.
Personality
Personality of characters: Most important are dominance/submissiveness and friendliness/unfriendliness, and they are readily identified by people.
Personality of interfaces: From text, people will be able to differentiate dominant from submissive interfaces and judge how similar or different they are from themselves and correspondingly like more or less the interface.
Imitating personality: People like a compuer that starts out opposite them in the dominant/submissive range and becomes more like them than one that is like them consistently.
Emotion
Negativity: People don't like negative media, but they pay more attention to it and remember it better. They remember material better when presented after negative media and worse when presented before negative media.
Social Roles
Specialists: Content from a source labeled "specialist" will be judged superior than from a "generalist" source.
Teammates: People teamed with a computer will feel more similar to the computer, judge it better, cooperate and agree more than people not teamed with it. (Teaming is as simple as a colored wrist band matching a colored bezel on the screen. Teaming can be peer, superior, inferior.)
