The sky is falling: negativity bias in social transmission

For June’s edition of the journal club, we’re reading Bebbington et al. (2017): The sky is falling: evidence of a negativity bias in the social transmission of information.

The paper presents the results of a transmission chain experiment in which participants were tasked to reproduce a story that incorporated a series of positive, negative, and ambiguous events. Additionally, state anxiety was manipulated by showing videos to the participants, elevating or reducing the anxious emotion. Using mixed-effects models, the authors demonstrate that negative events are generally transmitted more often than positive events, and that ambiguous events increasingly receive negative interpretations over successive transmissions. State anxiety, however, does not influence the negativity. Taken together, the results are interpreted to show a general negativity bias in social transmission, gradually transforming a story until we are only left to say: The sky is falling!

Feel free to comment with your thoughts on the study below!




image source:


  • comment-avatar
    Thomas Müller 19 June 2017 (11:54)

    negativity bias
    A very nice experimental study, theoretically plausible and with some good results. One thought I had was whether it might have been better to include a baseline condition with regard to the anxiety manipulation, in that there would have been neutral, elevated, and reduced anxiety, respectively. That way the effect of the manipulation could have been tested against no manipulation as well, showing whether there is some impact on the outcome at all (which we can’t see with the employed contrast). However, as the two conditions do not differ even in this maximal contrast, at most there could have been an effect into one direction for both manipulations (interesting, nevertheless). Maybe if, instead of state anxiety, trait anxiety had been measured, there would be some better results for this part of the experiment. This has also been done in some of the cited literature, and could be supported theoretically because the study is looking for general and permanent effects rather than specific, mood-dependent ones.

    In a similar vein, if the transmission of neutral events had been compared to both positive and negative events as well, I would be confident that an additional positivity bias could be found. In a way, this is what the analysis on ambiguous events shows, because more events become negative and positive than stay ambiguous. Taken together, results like this could provide an interesting bias framework to explain why, say, rumors can become increasingly sensationalist (mostly negative, but sometimes also positive!) and distort an original ambiguous or neutral message.

  • comment-avatar
    Piers Kelly 20 June 2017 (14:15)

    The banality of truth
    I wonder what would have happened if the participants had been told the story was true? There is a good reason to favour negativity if you’re trying to spin an interesting fictional tale: problems make for interesting plots, while the absence of a problem means there is no opportunity for narrative resolution, be it happy, tragic or otherwise. When listening to a “true” story on the other hand, there is less expectation that the narrative will be intrinsically interesting. It’s OK for the truth to be banal, but boring fiction wastes everyone’s time.

  • comment-avatar
    Barbara Pavlek 30 June 2017 (12:26)

    distorted narratives
    This was an interesting paper concerned with a very current topic of a prevalence of negativity in news and on social media.

    I find the ambiguity resolving part of the study the most interesting one. I think it gives the strength to the main argument, but also opens some new possibilities which the authors mention in the discussion. The way humans transform and distort narratives (real or fictional) is interesting, but in my opinion even more interesting is asking why. What is the main factor (if there is one) causing changes in the narrative? Our personality traits? Our designated public / interlocutors? The personal tendency to tell truth or spin exciting lies?
    One thing the authors mention is that people who think the world is a dangerous place tend to expect more negativity in daily news and conversations with others. Questions like these sound like they could be asked from a life history theory perspective, and social media platforms could be a good environment to test them.

  • comment-avatar
    James Winters 6 July 2017 (13:29)

    Amplification of informational asymmetries
    I really loved reading this paper; the use of ambiguous and unambiguous story events was a simple yet elegant way to test the effects of positively- and negatively-valenced information. In many ways, this study reminded me of work by Moussaid, Brighton & Gaissmaier (2015): The amplification of risk in experimental diffusion chains (which was surprisingly not cited by Bebbington et al). Here, the perception of risk is amplified in transmission chains, even when the initial message contradicts preconceived risk judgements. Of particular relevance is that the authors looked at both the message content (i.e., what is actually said) and the message signal (i.e., whether the message is positive, negative or neutral in valence). They found that (i) negative statements tend to propagate across the chain more freely than positive statements; and (ii) the relative proportion of negative statements tends to increase at the expense of positive statements. Importantly, there were significant individual differences (with some choosing to amplify positive aspects over negative ones), which makes me wonder whether these effects are more to do with the perceived asymmetry of payoffs (i.e., information perceived with more gain than loss in terms of payoff is preferentially favoured). In the real world, it just happens to be the case that there is (usually) more to gain from acting on negative information. For example, a heuristic whereby you always act on negative information might have a big payoff in terms of survival, even if most of the time it is irrelevant and/or results in a small cost.