Borissova et al. 2020 – MDMA, cooperation, and candles in the dark

Like Gabay et al. 2018 and Gabay et al. 2019, Borissova et al. 2020 is a randomized controlled trial looking at the effects of MDMA on cooperative decision-making.

But Borissova et al. didn't find the same increase in cooperativeness that Gabay et al. did. Why not?

The study had two goals:

We aimed firstly to extend and replicate the findings from a naturalistic study of recreational MDMA’s acute prosocial effects (Stewart et al. 2014) in a controlled laboratory experiment, using measures of empathy, trust and cooperative behaviour.

Secondly, we aimed to assess subacute changes to mood three days after MDMA administration.

Goal 1: do MDMA's prosocial effects replicate in an RCT?
Goal 2: do the infamous "mid-week blues" show up when MDMA is administered in a laboratory setting?

Their first goal is more relevant for our present purposes – it's the same big question that Gabay et al. looked into.

(As an aside, the Borissova study participants didn't report lower mood in the days after their MDMA session – some evidence that the "mid-week blues" come from taking MDMA alongside drinking, dancing and not getting enough sleep, rather than just from taking MDMA on its own. Sessa 2017 has a good discussion of this.)

Borissova et al. tested MDMA's potential prosocial effects by giving participants a bunch of social tasks: listening to emotional stories & rating how good the stories made them feel, looking at pictures of faces & rating how trustworthy they seemed, playing a "public project" game, a "dictator" game, and the ultimatum game we talked about last time. (The prisoner's dilemma wasn't one of the tasks, so we can't directly compare Borissova's results to Gabay et al. 2019.)

It's really good that Borissova et al. published these results, even though they complicate the picture. Our knowledge production apparatus doesn't really incentivize publishing null results as it's currently configured, so the file drawer remains a serious problem (exciting positive results are published & publicized while boring null results don't ever make it into the literature).

What's going on here? How is it that Gabay found strong prosocial effects and Borissova found none?

From the paper:

Some differences between the tasks in our study and previous ones yield important insights. In the ultimatum game, we asked participants to respond with the minimum offer they would accept. This – as opposed to the ‘direct-response’ method used by Gabay et al. 2018 where participants are presented with different offers to accept or reject – can lead to less punishment of unfair offers. Indeed, our participants were willing to accept offers that were below 40% of the total stake – below what is considered to be a ‘fair’ offer...

We also used a ‘one-shot’ task for our cooperative behaviour games. This contrasts to Gabay et al. 2018 and Gabay et al. 2019, where participants received feedback on their performance over multiple rounds. Such feedback might be important to more closely match real-world social interaction and allow for detection of an effect of MDMA. This again suggests there may be an important role for context, whereby true social feedback may be necessary to elicit effects of this drug – but this would need to be tested experimentally.

Two things here:

  1. The studies used different versions of the ultimatum game. In Gabay, participants played live with another person. In Borissova, participants were asked to think abstractly about the minimum offer they'd accept, but they didn't actually play the game with another person. If our hypothesis is that MDMA increases cooperation by boosting the salience of direct relationships with other people (which is about where we left things), then we should expect to see less effect on hypothetical reasoning about an abstract problem.

  2. In the Gabay experiment, participants played the ultimatum game dozens of times. In Borissova, they answered one hypothetical question about how they would play the game. This structure doesn't give any opportunity for players to change their behavior over time, as they see the dynamics of the game in action and learn more about their partner.

These are important points, but I don't think they let us straightforwardly side with the Gabay et al. results to say that MDMA definitely increases cooperative decision-making. And we also can't straightforwardly follow the Borissova et al. results to say that MDMA obviously doesn't have any effect on prosocial behavior.

The best takeaway at this point is something like "It's plausible that MDMA increases cooperative decision-making, but we're not sure yet. Running bigger RCTs on the question would help us learn a lot more."

But I notice that a part of me wants to compress things down into a simple answer... does MDMA increase cooperation, or doesn't it? It's uncomfortable to not know for sure.

I really do believe in science as a candle in the dark. We can use scientific processes to actually figure things out, which is amazing! Careful theorizing can generate interesting hypotheses, RCTs can determine the causality of many questions (even questions about complicated systems like human beings), and literature reviews & meta-analyses can roll up these results into an overall state of the art that continually improves over time.

But this wonderful process isn't straightforward, and there are many opportunities for motivated reasoning. A psychedelic skeptic can point to Borissova et al. 2020 to back up their point that psychedelics don't produce any real benefit, just a momentary good feeling. An MDMA advocate can pick up Gabay et al. 2018 to show that ecstasy can help us all get along better. Neither are wrong (in a narrow sense), but they're both missing the bigger picture.

A casual reader can go astray even if they're not trying to prove a particular point. If they bump into Borissova et al. 2020 first (or Kuypers et al. 2014, or Schmid et al. 2014), they'll learn that in reality, MDMA doesn't increase prosocial behavior. But if instead they bump into Gabay et al. 2018 first (or Stewart et al. 2014, or Kirkpatrick et al. 2015), they'll see that MDMA does in fact increase prosocial behavior.

So what's a poor ol' consumer of scientific research to do?

I think a good place to start is improving our ability to notice when we have impulses toward quick, sure-thing answers.

Reality has a surprising amount of detail, and in any domain it takes a lot of work to figure out what's going on. Headlines announcing new results will never convey the full story, and neither will paper abstracts. Knowing this, and then starting to notice when you're looking for a quick answer rather than actually trying to figure something out – this seems like a good first step.