Social Contagion: Taking on the values of others
Are you wearing a mask? How do you feel when you see other people wearing masks? How do you feel when you read tweets or see instagram posts showing other people wearing masks? You may not realise it but learning the value preferences of other people changes your own preferences and makes you more like them.
Contagion isn't a great word to be throwing around during a pandemic... but the transmission and spread of beliefs and behaviours from one person to another can be highly beneficial from an evolutionary perspective. That said, there are situations where negative or harmful beliefs can be transmitted as well. Here, I'll introduce my PhD student's research (you can find here on twitter @ljthomas1991) where she made hundreds of people more patient/impulsive, risk-seeking/risk-aversive, just by showing them the alternative preferences of another person.
Social influence through the lens of behavioural economics
Last week, I posted about the benefits of using mathematical models instead of vague and often ill-defined psychological terms (Knowing me, Knowing you? Do Psychologists understand each other?). Behavioural economics offers an incredible resource for psychologists as they use mathematical models to describe how humans and other animals make value-based decisions. You may have seen this fun video of children doing something called 'The Marshmallow Test'. Can children restrain themselves from eating the first marshmallow in order to double their rewards with a second marshmallow?
This is an example of Intertemporal Discounting (also known as Delay Discounting). The longer we have to wait, the less valuable something becomes. Let's see an example using money.
Which of these offers to you prefer?
£5 now or £5 in a week
In this case everyone would choose £5 now rather than wait a week. This highlights how waiting for a reward actually reduces its value. But what about this offer?
Which of these offers to you prefer?
£5 now or £10 in a week
This is obviously a more difficult decision. For some people it's more valuable to wait a week for more money, whereas other people would see £5 now as being more valuable than £10 later. By asking lots of these kinds of questions we can run your answers through a simple mathematical formula to calculate your 'temporal discounting factor' - a single number that describes how much having to wait reduces the value of something. Temporal discounting factors are very reproducible and stable over time, and can be indicators of clinical conditions such as ADHD and addiction.
It's also worth noting that temporal discounting is just one type of task you can use here. There's also risk or probability discounting (£5 guaranteed vs 50% chance of winning £10), or effort discounting (£5 with no work vs £10 but you have to do some work). Here, effort can be either mental or physical. Each of these simple discounting tasks has a different mathematical formula to run the data through, and can help us to reduce the complex constructs of risk and apathy down to single value.
So here we have a series of robust mathematical models that can be used to explore impulsivity, risk, and apathy. How do we use this to study social influence? Easy, just watch and learn the choices of someone else! This is where it gets really interesting, because learning the value preferences of someone else makes you more like them.
Social influence of value preferences
Because we use a mathematical model to define participant behaviour, we can also use the same model to simulate the behaviour of other people. In all these experiments, participants are actually learning about two artificial computer agents (not real people) who are ±1 of the participants own discounting factor, i.e. 1 step more impulsive or 1 step more patient. This is great when thinking about experimental control because it means it doesn't matter whether the participant is impulsive or patient, risk-seeking or risk-aversive, apathetic or highly motivated. They're always exposed to someone that sits 1 step further away on the continuum.
I put together a quick fire round of questions and answers regarding contagion. If you've got a question then leave a comment and I can update this.
So if I learn that someone else makes different choices to me I become more like them?
Yes! It's an incredibly robust effect. The image below gives a breakdown of the data.
Does this kind of social influence only work for temporal discounting, i.e. impulsivity?
No, we (and other labs) also see the same effects for risk discounting. Participants became more risk-seeking or risk-aversive depending on the person they observed. There's also a large literature on emotional contagion, a core construct in empathy.
Does this change last forever?
No, it seems to be linked to exposure. The more you see someone else's preferences, the longer you maintain their beliefs. You eventually go back to your original value preferences. But this is still an area where there's lots of scope to learn more.
Are some people more sensitive to social influence than others?
Yes, looking at all the data in the graph below, we can see there's quite a spread of responses. After learning that someone else is more patient, most people become more patient (blue dots), but some don't change (grey), and some even go the opposite way and rebel (orange)!
Why is that?
Great question! As of yet we don't have a clue!! We compared contagion to the following variables and found no relationship:
Whether you believed the other person was real. This was about 50:50 but had no effect on contagion
Whether you noticed your choices change after learning about the other person.
Cognitive and Affective Empathy
Apathy and Motivation
Oh, so you're really not sure why some people take on the values of others more readily?
Unfortunately no, and it's something I'd love to get comments on from social psychologists with any ideas. One thing we found in a more recent study is that contagion is stronger when you're learning about a friend compared to a stranger. However, that's a topic for another blog =o)