Machine works, flies are not learning

The old software is installed with the new hardware and I’ll be testing the new software in parallel to experiments with the old software. Technically, everything seems to be working now, but the first six flies I tested did not show learning after training:

However, avoidance looks fairly ok, only the last training period gives away that something still isn’t quite the way it should be:

So I’ll go ahead and train a new batch of flies this week. I’ll focus on training/avoidance and make sure it is perfect, before I shorten the training.