An Alien at a Social-Psycho Conference

Miriam Redi
SocialDynamics
Published in
4 min readFeb 9, 2017

The Annual Meeting of the Society for Personality and Social Psychology (SPSP) is an annual convention gathering the whole community of social psychology researchers.

“Who — ho I am a legal Alien …”

It’s an impressive concentration of brilliant research. And, out there, I was like a legal alien. Like Sting’s English Man in New York. A computer scientist in a social psychology conference. Getting surprised because people are very well dressed, and there is a line at the ladies room ;) Excited by mysterious terminology (“rumination”). Asking any kind of strange questions, because, you know, I am an alien.

This is what I was able to absorb.

The automated image analysis session.

I’ll start with the easy part. Our session at SPSP provided a preview of what computer vision techniques can offer to research psychologists how to integrate these methods into their research programs. Samuel Gosling @SamGoslingPsych was the chairman of the session.

[my usual stuff] Subjective Machine Vision can detect visual properties beyond semantics. I kicked off talking about the new frontiers of subjective machine vision: understanding people’s aesthetic preferences, feelings, personality and other psychological characteristics from the pictures they create, share and like.

Comparison of visual sentiment models across Flickr language communities

[portable vision] Wearable cameras combined with computer vision help studying people’s daily habits. More interestingly, @RyneSherman showed how computer vision can be used to infer people’s activities from wearable cameras.

[Faces are amazing] People Sexual orientation can be predicted from faces. Finally, Michal Kosinski delivered a brilliant presentation on face analysis and its potential for social psychology: neural networks can predict many things from people’s faces, including sexual orientation!! Michal is one of the main contributors of the myPersonality project: a dataset of million of Facebook users annotated with their personality traits.

More Lessons on Images

Now, the conference was way to big to see everything. Although I would have loved to. So I had to filter the impressive amount of research works out there with the keywords — guess what — “visual”, “image”, “pictures”. A brief summary of interesting things that caught the alien eye’s attention.

[the power of images] Visual content is more effective than textual content to improve people’s attitude toward minorities. A nice study using this “Modern Racism” as an evaluation metric. Charles Chue et al. from Purdue University

[funny one] Jesus’s face has different features for different people. There are “Warm VS Cold Jesuses” Chris Silver et al. from University of Tennessee

Average faces of Jesus from Cold to Warm!

[cross-culture] Facial expressions change across cultures. Europeans look more excited than Asians in interviews. European employers expect more excitement in interviewees than Asian employers. The “Compassion” facial expression differs between American and Germans! Bencharit et al., Stanford

German VS American compassionate expression

[scary] Republican newspapers depict dark skin politicians as darker-skinned compared to Democratic newspapers. Nesbitt, Universit of Nevada.

Skin color scale used by the authors to measure “darkness” of candidate’s skin on newspapers.

[magic] Time slows down when you are in nature! This is probably the most beautiful piece of work. A crowdsourcing study exposes subjects to urban vs nature scenes and then asks them to count up to one minute without watch. The time of people exposed to urban scenes goes faster! Davydenko et al., Carleton University

W

Message to the Tech Community

[Best Message Award] People see AI technology as positive when it “restores”, i.e. it solves people’s problem. When the aim of the technology is to solve new problems or give “super powers”, technology is perceived as immoral. Newman et al., University of Southern California

--

--

Miriam Redi
SocialDynamics

mad astonaut, vision research scientist, teaching machines to see the invisible