Imagine the following. You’re moving house and you need some carboard boxes for your belongings. You know that a colleague at work recently moved and so you ask them whether they might have any boxes they could loan you for your own move. The colleague agrees without hesitation – it’s no big deal for them. You agree to collect the boxes at the weekend and to return them once you’ve unpacked. 

Now imagine something slightly different. You’re still moving house and you still request the favour, but this time your colleague hesitates before agreeing. There’s a slightly awkward moment where you’re not sure what they’re going to say – but they eventually agree that you can borrow their boxes. 

These two scenarios make us feel slightly differently towards the colleague – for good reason! Hesitating – or deliberating – before helping someone acts as a signal that the helper is strategically considering their options before investing. It suggests that perhaps they aren’t that intrinsically motivated to help and that they are considering the costs and benefits of doing so (‘Will it make me look bad if I refuse? How much effort would it involve for me?’) before making their move. And we don’t need to rely on thought experiments to show this: research conducted by Jillian Jordan and colleagues in 2016  showed that uncalculated helping serves as a stronger indicator of trustworthiness compared to calculated helping. In this experiment, helpers who chose to see what their own personal cost would be before offering help, or who took a long time to decide to help, were considered less trustworthy than those who acted more impulsively. People also seemed to be aware of the potential reputation costs of deliberating over help – and acted in a more uncalculated manner when they knew they were observed by others.  

In our study, we built on these findings by exploring deliberation in a punishment setting. Punishment occurs when one individual incurs a small cost to inflict a larger cost on someone who cheats. While this may not sound like a particularly prosocial thing to do, punishment can be deployed in settings where the punisher was not even the victim of the original harm and can help to increase cooperation within communities. Think, for example, of someone who intervenes to defend someone against a bully or a person who chases down a thief who steals someone’s handbag. This is what’s called ‘third-party punishment’ and, despite being a harmful behaviour, it is also often construed as a prosocial one. We were especially interested in the signal that this kind of punishment might send about the punisher and whether we might replicate the findings above, by showing that uncalculated or reflexive punishment would be a stronger signal of trustworthiness than more calculated or deliberative actions.

However, our predictions were not as straightforward as for the helping scenario. Helping is an unequivocally positive action – generating benefits to others. Punishment, on the other hand, is more ambiguous – as it is a fundamentally harmful act. This complicates interpretations of the punisher's motives: while some punishers might be motivated by a desire to help others or to restore fairness, others could simply enjoy harming others or causing strife.  This ambiguity also complicates how people might view those who punish reflexively compared to those who deliberate: unlike the helping setting, deliberating over punishment might also stem from the punisher considering the harm they are about to inflict on another individual (which might be viewed positively) and not just from what it will cost them to intervene. Maybe thinking carefully before harming someone would be seen as the ‘right thing’ to do – unlike helping, where the ‘right thing’ seems to be to help without thinking too much about it.

Buoyed with enthusiasm for our ‘straightforward’ research question, we submitted a plan for a registered report where we would replicate Jordan et al.’s study within a punishment context. And then things spiralled a little. During the review process, we also added in helping contexts to make comparisons between punishment and helping – and we also threw in another experimental condition for good measure, exploring whether deliberating over the impact to targets and not just the cost to self would be viewed especially positively by observers in a punishment setting. By the time we had completed the review process, we had a whopping 41 hypotheses over five experiments, a pre-registered N of 13,060, and the most unwieldy registered report design table you’ve ever seen. Our study was no longer straightforward, but it was comprehensive.

Study 1 was a replication attempt of Jordan et al. (2016) with an additional punishment condition -  this study therefore asked how considering personal costs during decision-making influenced perceptions of trustworthiness of helpers and punishers. Study 2 went on to ask how considering the impacts on targets affected perceptions of trustworthiness of helpers and punishers. We expected that deliberating about the personal costs of one's actions would signal untrustworthiness for both punishing and helping, while deliberating about the impact of those actions to others would serve as the distinguishing factor between helping and punishing behaviours. Specifically, we anticipated that individuals who deliberated about the impact of punishment would be trusted more, whereas those who deliberate about the impact of helping would be trusted less.

These expectations were only partly met. Although we largely replicated Jordan et al.’s original findings (uncalculated helping reliably signalled trustworthiness), this pattern did not extend to punishment (only individuals who made uncalculated decisions to refrain from punishment were perceived as more trustworthy than those who checked costs but then refrained). In Study 2, reflecting on the impact of helping had a more pronounced effect on the trust and trustworthiness perceptions of non-helpers than helpers. And our main hypothesis was completely unsupported: we found no evidence to indicate that considering the impact of punishment influenced perceived or actual trustworthiness.

So what did we learn from all this? It was reassuring to replicate what has become a classic finding in the field: that uncalculated help serves as a signal of trustworthiness. That our results didn’t support our predictions when it came to punishment suggests that there is still a lot that we don’t know about this behaviour – and how it is viewed by others. There is still much to learn about punishment in humans and we look forward to studies that address the many open questions that remain in future.

References

Jordan, J. J., Hoffman, M., Nowak, M. A., & Rand, D. G. (2016). Uncalculating cooperation is used to signal trustworthiness. Proceedings of the National Academy of Sciences, 113(31), 8658-8663. https://doi.org/10.1073/pnas.1601280113