More and more the behavioral neuroscience field is moving toward using highly efficient and high-content behavioral tracking software to record and analyze fly behavior (for questions about the best tracking systems check out ‘Which tracking system should I use?’). There is good reason for this: watching, hand-scoring and analyzing fly behavior can be extremely laborious, time-consuming, and let’s face it, mind-numbing work.
That being said, I think one of the most enjoyable things about being a scientist that studies fly behavior is watching the flies behave. I started my scientific career watching worms behave, and then Drosophila larvae behave. While this can be endlessly fascinating (and also effective at inducing a trance-like meditative state), when I started my postdoc I got to watch adult flies and it was a revelation. Their behavior is way more complex than we ever thought it was in the ‘olden-days’ when people basically used flies as handy in vivo test-tubes. Drosophila have complex natural environments where they need to find food and mates, and escape from predators. They are constantly evaluating and interacting with their environment (for more on this check out ‘How do Drosophila behave in the wild?’).
Should you watch fly behavior with your own eyes?
So, aside from the fact (yes, fact) that watching flies is fascinating, should you watch your flies behave in the lab as a regular practice? I unequivocally say “YES” to this for a number of reasons:
You will get to know your organism and gain an intuitive sense for what their typical behavior looks like.
Your familiarity with fly behavior will allow you to be able to spot when something seems ‘off’ that day.
You can easily get ideas on how to improve the behavior arena and environment.
You can gain ideas on what other behaviors to test.
Your brain is an amazing machine and you can pick up behaviors and responses that the tracking software will miss. This is especially important when performing mutant or circuit analysis. Although tracking systems are objective, sometimes it is difficult to interpret the behavior and watching the flies can give you an intuitive sense of what the behavior differences are.
Interpreting the massive amount of data from high-content tracking software will be easier.
You will learn an appreciation for tracking systems (I certainly did after watching hundreds of videos and tracing thousands of trajectories in the pre-tracking software days)
It’s a fantastic practice to do quality (or sanity) checks - every once and a while. By this I mean to check to see what you observe with your eyes in the videos matches the output of your tracking software. Tracking errors and mistakes are not uncommon.
When characterizing a new behavior in the lab it’s always a good idea to watch the flies because this can help you figure out the best metrics to measure.
Should you hand-score fly behavior?
In the pre-behavior tracking days, we all had to hand score behavior and I certainly did my fair share. If anyone could find a faster way to get the behavior data out, it would be me. That being said, I think there is also some merit to hand-scoring fly behavior. “But, but …. why???” do you ask? There’s a few reasons for this recommendation:
When characterizing a new behavior in the lab it’s always a good idea to hand-score the behavior (and watch the flies behave) because this can help you figure out the best metrics to measure and analyze.
Hand-scoring can be helpful for compare your observations to published data (remember, all the ‘old’ published data will be hand-scored and annotated.
Hand-scoring snippets of your videos can help you discover potential tracking errors
Hand-scoring behavior can help you get an idea on how to best train a new classifier for a behavior where one doesn’t exist.
Maybe I’m a bit old-school because I spent a lot of my childhood watching animals (especially insects) behave in northern boreal and alpine forests, and gained an appreciation for their complexity. And maybe because the labs I was trained in were focused on the intersection between genetics and psychology, behavioral neuroscience, and neuroethology. In any case, I recommend to all people joining my lab that they spend some time actually watching the flies until they gain an intuitive sense of the fly behavior they are interested in.
What are your thoughts on this?