I completely ignored the experiment purpose and the instructions given.
What I enjoy about puzzle games is discovering rules that emerge from the internal mechanics. Mastering those rules feels like learning "the language" of the puzzle. Once you speak "the language" you can solve them much faster.
Based on solely the increase or decrease of correct pixels indicator in the top left, I was able to solve the "bat" in 14:26 (stopwatch)
It goes against the idea of the experiment, but if you include level numbers, it would help me boast about my competitive puzzle skills :)
This is what I call "bat", the original shape was different, but it paused on this frame after completion:
