Project on observational learning capabilities of birds-need help refining method

[deleted-442640]

Hello, board:

I write to ask for assistance refining the method of my experiment. Any feedback is appreciated. My project was focused on collecting data about how much of a certain cognitive skill (such as insight) various backyard bird species possess, but I now think it's infeasible to set up a video recording system, motion-activated or otherwise, to actually identify species, attempts, and success rate.

I am now attempting to answer this research question: to what extent does an exemplary image improve backyard birds success rates in completing a cognitively challenging task?

An example of such a task would be the water displacement paradigm, where a container with food floating on top of water can be manipulated by dropping objects into it, raising the water level and rendering the food accessible.

After an acclimation period where food is freely given at feeders, I would administer the paradigm using the same type of food. Feeders would feature a static picture or model of a bird carrying out the task, while control feeders would only feature the paradigm. The idea is to measure and compare the rates of successful food acquisition without needing to observe all attempts. The paradigm is ideally one that the bird needs either insight or observation to complete, not just trial and error.

The bait would be replenished, say, at the end of each day.

I see many problems with this. For example, if an individual or group masters the task early in the testing period, they could just repeat it within or throughout days and achieve a perfect acquisition rate. The probability that they find the control feeder could be mitigated with distance.

The water displacement task may not be optimal (any suggestions?)

I'm using black oil sunflower seed for its wide appeal so that a wide range of species will attempt the task. However, isn't there still the possibility that one feeder site will happen to attract a certain intelligent species that the others won't, skewing the results?

I still intend to observe which species can learn, but for limited periods of time while I personally observe the feeders.

With that said, how could I address some of those flaws and sources of error, which other ones do you see, and how could I fine-tune the method to produce more valuable results? Any general advice or feedback would also be appreciated.

Thank you very much.