How to Study Harm

Ethics and accuracy raise challenges for exploring harmful behaviors in the lab.

Posted Dec 05, 2019

Researchers often seek to study things that harm people. For me, it's aggression. For others, it's cancer. Yet our goal is the same: understanding something that harms people so that we might ultimately mitigate that harm. 

Studying harmful things is needed to make the world filled with less suffering, but it comes at a cost. Whereas scientists who study non-harmful things can readily recreate them in their controlled laboratories without ethical concerns, us harm researchers cannot do so. To study harm in the lab you have to recreate it, but this then entails that someone is harmed.

This is the harm science paradox that happens to all of us who study harmful things: because the thing you want to study is inherently harmful to people, you cannot ethically recreate it in the lab in its naturally-occurring form, because doing so would require doing real harm to people. For example, I want to understand violence but I cannot host a daily fight club in my lab because participants would be carted out on stretchers. The Milgram and Stanford Prison Experiments are striking examples of when our desire to accurately recreate harmful behaviors leads to unethical harm-doing to real people.

Source: Pixabay

So what can we do about this? As a harm researcher, it seems like you have 4 options:

1.    Don't study that harmful thing.

2.    Recreate a minimally-harmful version of it in the lab and hope that’s good enough.

3.    Recreate it fully in non-human animals that it's 'okay' to harm.

4.    Observe it outside the lab, as it naturally occurs in the real world.

We can't do #1 because violence needs to be understood and we need that understanding yesterday.

I won't do #3 due to an excessively soft spot in my heart for little creatures.

So I'm left with options #2 and #4.

In my work, I've focused on the simulated laboratory experiments inherent in option #2. Do they work? More specifically, do they accurately capture the intended target (aggression) in a way that will generalize out into the real world? Well, it depends on what you're trying to learn and on who you ask, but that's a topic I've already covered in detail.

Source: Pixabay

Option #4 because is messy and difficult because those are defining features of the real world. Labs allow control over details and allow you the infrastructure you need to pull off your experiments. 

I've started to look to cancer researchers as an example of a way forward, as they are in a similar boat as fellow harm-researchers. They can’t simply give people cancer and they can’t not study cancer (option #1), so they go with all of the other three options. They study manufactured versions of cancer in the lab that often don’t resemble conditions in the real world (option #2). They give cancer to rodents and other critters and study it in their bodies (option #3). They go out into the real world and study people with cancer in their homes and communities (option #4). But most importantly, they intermingle options #2 and #4. They bring people who have already naturally developed cancer, and bring them into their controlled labs to study. Or they take their carefully contrived experiments and find ways to bring them into the community. This hybrid approach is the most promising for harm researchers in psychology. We as harm researchers need to do a better job of bringing the real-world into the lab and the lab out into the real-world. I have begun to apply this model to my own work.

For me, this has meant recruiting people from the local community to complete our brain imaging studies and screening people who score high on measures of antisocial traits like psychopathy, instead of just recruiting undergraduate students. 

It has also meant doing my lab's first field study out at a 'Rage Room' to see how simulated acts of aggression influence short-term and long-term emotions. 

It has been much messier than our typical lab studies, but it has also been much more rewarding.