Columns Journey of a YI

Fail faster, fail better

Mayuri Rege

Mayuri Rege is a DST-INSPIRE Faculty in the Department of Microbiology at Ramnarain Ruia Autonomous College, Mumbai. She is one of the Young Investigators selected to attend YIM 2020 in Mahabalipuram. In this invited article, she writes about the importance of facing and learning from failure as a scientist, and how she tries to instil this value in the undergraduate students she teaches.

Mayuri Rege
Mayuri Rege 

Failure is an integral part of doing research. Whether it’s the story of Edison’s 1000 bulbs or Rosalind Franklin’s trials leading up to the famous Photo 51 of the DNA double helix, failure is a lesson we should have learnt the moment we embarked on the research track.

However, ask yourself, how often do experiments fail in a typical undergraduate practical lab? For example, before the session on PCR, the teachers will often ensure that the band is always present and at the right size even though PCRs regularly fail in real-life research. To add to this, reference books and publications that students refer to typically discuss only the successful finding and breakthroughs without any mention of the multiple failed attempts it took to get there. 

Thus, we do not adequately prepare young researchers to face the most frequent situation in a lab — a failed experiment! I reflect on my attempts to deal with this issue as I set up my lab primarily with undergraduate and postgraduate students in one of the top colleges in the country. 

My first encounter with failure was when I was a master’s student at the Tata Institute of Fundamental Research (TIFR), Mumbai. I wanted to understand what made males more susceptible to malaria compared to females using a mouse model. We wanted to check for differential compounds in samples obtained in the least invasive method, namely urine. 

Although it is pretty straightforward for humans, how does one collect mouse urine for analysis in a non-invasive manner? Since mice are not housetrained, we had to use expensive metabolic cages that had a special funnel design to collect the urine without disturbing the animals. However, I soon found out that our mice just would not pee in these cages! Bizarrely, they would just sit in one place, not drink any water and eventually die — even the control mice that weren’t infected. 

I changed multiple parameters — gave them different food, and widened the pore of the water feeder so they could drink’ more water but none of these made the mice happy’, leading to several failed experiments and this becoming a running joke amongst fellow students. Eventually, I figured out that the bar spacing on the bottom of these cages was too wide for our mice to move freely- explaining why they were perched in one place. Placing a smaller spaced jaali’ on the bottom immediately solved my problem and to my relief, I could finally collect mouse urine to perform my experiment. 

Perhaps the first step of getting students accustomed to failure is to shift the focus from getting good results’ to learning the research methodology. Teaching them to design experiments demonstrates the importance of controls that help to cross-check experimental conditions and give expected results. 

But the first time our experiment fails to give an expected’ result, my students are just dumbfounded – they haven’t ever encountered this possibility before! Their knee-jerk reaction is to simply repeat the experiment as is because they presume that they have done something wrong along the way. On probing them about what might have gone wrong, they only have vague hand-wavy answers and no concrete reasoning that justifies repeating the experiment. This observation was a turning point for me as I realized that students need to be taught to step back and interpret failures. 

Troubleshooting a failed experiment is what gets students to think critically because this time the answer is not already available on the internet. Importantly, I have to resist the urge to give them the answer and wait for them to arrive at the solution on their own. A natural consequence of this is that the students then proactively suggest what the next experiment should be – finally setting the scientific process in motion. 

Another aspect of young researchers dealing with failure in the lab is at an emotional level. These are students who have excelled at academics and I have to remind them to not take failure in the lab personally. A failed experiment isn’t a reflection of who you are as a person. Students who cultivate resilience in the face of failure and show the tenacity to work through it are the ones that actually succeed as future graduate students. Coping mechanisms might differ — my students will often burst out into hysterical bouts of laughter when their best-laid (experimental) plans fail and then work together to figure out the next steps. 

Designing experiments that can actively disprove your hypothesis is the only way to rigorously test your science. An experiment that discriminates between your favourite hypothesis and others and can disprove your hypothesis is more important than several peripheral experiments that support it. There are two facets to this – avoiding confirmation bias and saving time. 

An important skill to acquire as a young researcher is to identify that smoking gun for a given hypothesis – something that can happen if and only if the hypothesis is true. In practice, this is hard to achieve and such a smoking gun often involves a combination of observations. However, these are the experiments that must be done first in order to test the hypothesis effectively and fail quickly if we have to. This saves time that would have otherwise been wasted on inessential experiments.

How can we remove the fear of failure at the undergraduate level? One possibility is by changing the metric of performance for research done in colleges. We can reward original thought and proper research methodology, whether or not it leads to publishable results. While this is harder to assess, we can recognize undergraduate teachers who emphasize how to do research’ rather than getting their students to produce results that lead to publications in predatory journals. 

In our Department, we have implemented this through rigorous discussions of projects proposed by teachers by their peers. We also grade our students for clarity in formulating a hypothesis and engagement in discussions on experiments rather than just showing a result. Although it’s too early to tell if this works for a majority of institutions, we see a remarkable improvement in students’ ability to tackle problems after being subjected to frequent failures. 

By letting them fail early, undergraduate research can teach students the correct process of doing research, prepare them for graduate school in the future and meaningfully contribute to the research enterprise in India.

Written By

HBCSE, TIFR