Psychologists have long known that while B.F. Skinner is a founding father of behavioral psychology, some of the foundations he built his theories upon haven’t held up under the scrutiny of modern research.
One of Skinner’s core contributions to modern psychology was a theory called “operant conditioning.” In it, he believed that people could be motivated by four different types of stimuli: negative or positive reinforcement and negative or positive punishment.
Unfortunately, a lot of developers build online tools, services and frameworks that put their pop psychology beliefs into practice. So what did the researchers find when they examined the use of two of Skinner’s most popular operant conditioning tools in a few large online communities?
First, let’s run through what we mean by all these psychobabble terms. Most of us are familiar with punishment, which Skinner might consider a “positive punishment.” That’s when you add something aversive following a behavior you want to extinguish (such as a parent giving a child a spanking1, or an adult receiving a speeding ticket for driving too fast).
The opposite of this is negative punishment — the removal of something that a person values (such as taking away a child’s toy after a behavior a parent is trying to stop).
Positive reinforcement is the addition of something — like a reward — that occurs after a behavior you want to reinforce (such as when you give a child a treat for picking up his toys, or an adult receives a bonus at work for exceeding their goals). Negative reinforcement is the removal of an unpleasant stimuli to encourage future behavior (such as cleaning the trash out of your car to avoid a funky smell).2
Operant Conditioning in Online Communities
The Internet is full of a diverse range of online communities, ranging from Facebook, Twitter, reddit and 4chan to support groups and comments on blog posts. Communities vary just as widely about the tools they employ to encourage positive behavior by people who post or comment in them. Communities like reddit and Slashdot, for instance, use up-voting and down-voting by its users as a means to encourage positive contributions. Facebook only allows positive “Likes.” Psych Central’s support groups only allow “Thanks” or “Hugs” to be given to a post.
Researchers wanted to understand how these various reinforcement tools work in online communities. So over the course of 18 months, they examined the voting behavior of 1.8 million different users who cast 140 million votes on 42 million comments left on four large online news communities (CNN.com, Allkpop.com, Breitbart.com and IGN.com) to see what relationship those votes had on participants’ behavior. All fours sites use a comment moderation system called Disqus that allows users to both up- or down-vote comments they agree or disagree with.
Here’s what they found.
Negative evaluations increase posting frequency
The more negatively a commenter was evaluated, the more that commenter would post in the future. This is contrary to what operant conditioning theory would suggest. Punishment (in the increasing of negative votes for your comment) would suggest that a commenter would post less in the future. Instead, the researchers found commenters posted more than those who received positive evaluations. Surprisingly, no feedback at all slows down posters the most:
Moreover, when we examine the users who received no feedback on their posts, we find that they actually slow down. In particular, users who received no feedback write about 15% less frequently, while those who received positive feedback write 20% more frequently than before, and those who received negative feedback write 30% more frequently than before.
Negative evaluations impact future perceptions
The more negative evaluations you get (in the form of down-votes on your comments), the worse the community perceives you. “After a positive evaluation, future evaluations of an author’s posts do not differ significantly from before,” note the researchers. “However, after a negative evaluation, an author receives worse evaluations than before.”
Negative evaluations reduce future post quality
Once a person have been evaluated negatively on their comments, their future comments’ quality is likely to drop significantly. A positive evaluation does nothing for future post quality.
These results are interesting as they establish the effect of reward and punishment on the quality of a user’s future posts. Surprisingly, our findings are in a sense exactly the opposite than what we would expect under the operant conditioning framework. Rather than evaluations increasing the user’s post quality and steering the community towards higher quality discussions, we find that negative evaluations actually decrease post quality, with no clear trend for positive evaluations having an effect either way.
If you get negative votes, you’re more likely to give others negative votes
The researchers also found that if you get a negative evaluation on your comments, the following week you’re more likely to vote on other people’s comments more negatively. There was no change in participants’ voting behavior for positive comments.
Overall, punished users not only change their posting behavior, but also their voting behavior by becoming more likely to evaluate their fellow users negatively. Such behavior can percolate the detrimental effects of negative feedback through the community.
What Does This Mean for Online Communities?
The researchers found that negative feedback (in the form of down-voting) leads to significant negative changes in the user’s behavior. These changes are overall very negative to the online community in general, because the user is more likely to post more often with comments that are of lower quality. They’re also more likely to evaluate their fellow users more negatively in the future.
There are some limitations to mention, however. The researchers examined only four online news-oriented websites, where the sense of “community” may be very different than a Facebook or LinkedIn group, or an emotional support group. So we don’t know whether these results are generalizable to all online communities (my suspicion is that they may not be). The researchers only focused on one type of feedback — votes of comments — and they mostly ignored the content of the actual discussion taking place.
Rewards (in terms of up-votes and positive feedback from the community) appeared to have little effect in encouraging a greater quantity of future comments or increasing the quality of comments. But punishment seemed to have the exact opposite of its intended effect. Users who were punished (by receiving down-votes on their comments) actually posted more and of less quality in the future.
If you’re an online community that uses both up- and down-voting, this research should encourage you to examine your own community for similar data trends. And perhaps reconsider the use of down-voting.
Cheng, J, Danescu-Niculescu-Mizil, C. & Leskovec, J. (2014). How Community Feedback Shapes User Behavior (PDF). Association for the Advancement of Artificial Intelligence.
- Which no adult should ever do nowadays. [↩]
- Funny enough, Wikipedia’s entry on these two terms is exactly the opposite of what they claim. Ahh Wikipedia… [↩]