I think of the inhabitants of the “people” category as “main characters” rather than just people. They can be non-human, they can be sentient objects, cartoon characters, aliens, etc. The important thing for me is that they are viewed as “sentient subjects, capable of movement and intention.” This way they become the focal point of the scene and they initiate the animation of it.
Usually the more specific the person, the better. This doesn’t have to mean using a specific actual person like Johnny Depp. If you use a “pirate” that is initially a generic entity, but you are consistent with how you visualize that pirate you’ll have created a specific and unique image from that initially generic person. So always picturing him with a specific hat or bandana, a vest with a specific pattern or color, a peg leg maybe on a specific leg, a specific beard or mustache shape, yelling a consistent catchphrase, etc… The more you consistently work with that image the more it becomes that person for you. It may take less time to just try to picture Johnny Depp’s Jack Sparrow as your specific pirate, but if you can craft your own person, it could have more of a connection with you.
Something that helped me when learning the 1352 images for the Shadow System (many of them people) was to grab an image from google for some of my more generic words. For example, I had “session musician” and “diver.” Initially I struggled to visualize them until I grabbed an image off of google for each. This gave me a consistent visual and after a while those words (and their associated card pairs and numbers) just “became” those specific images. Before I did that, I found that I could recall the word just fine, but there was no real image that was coming to mind. If you use a flashcard app to practice your associations, I’d really recommend using pictures on the answer side so that you can lock in to a specific association.
As far as actions go, try to use ones that are active, that can affect an object, but that don’t rely on a specific object to adequately visualize. If you can picture a mime performing the action and you can clearly tell what it is, it usually is a good PAO action.
So, you mentioned “playing football” as an action. Ronaldo (P), Playing Football (A), Football (O). The “playing football” action could be problematic since it requires specific objects to really visualize. A “better” action option would just be “kicking.” Or, maybe more specifically “bicycle kicking.” This is an unambiguous visual action, that could be performed by anyone, on any object. You can picture someone bicycle kicking a guitar, or a donut, or a fish tank, or whatever.
“Playing guitar” would be better thought of as the action that the hands and fingers make while playing guitar, but without the guitar implicitly in the image. This way you can fill that space with a guitar or with the donut, the fish tank, etc. You could picture a person holding one of those objects like a guitar and “playing it.” It may require a change in practice habit to start visualizing the action on its own, without any objects at all, but I think it will help.
One last example, you mentioned “cooking” this is kind of an ambiguous action that would usually also be dependent on another object of some kind. It also could be a passive action, which usually isn’t very good for a PAO. Actions like “reading” or “thinking” don’t make for very visual images. A useful way to visualize “cooking” might be to modify it to something like “stirring” and visualize a person swirling their arm and hand around as if in a pot. The interaction with an object could be that the person takes their hand and shoves it into the object and starts swirling and mixing the object around. This could work with any object, but doesnt depend on a specific object to visualize. Keep a watch out for passive actions and try to adjust them to a more active and interactive form.
This brings us to the idea that you don’t have to create your action and object lists so that they naturally connect back to their standard PAO person. So, you don’t have to have a set like “Ronaldo playing soccer with a soccer ball.” In fact, the end goal of learning a system is to bypass that extra connection so that you directly link your action to the number and, separately, your object to the number.
This is easier to do when your action can be independent. If you could picture yourself miming the action, this is a good way to learn it because then you’re not creating any extra links to people or objects that you will need to break later
More on the idea if independent association here:
If you use a system where the number itself can generate an independent action word and an independent object word, you can learn the direct associations from the start. This is possible with something like the Major System, where you can generate all the elements needed for your list independently and they all follow the same encoding/decoding structure. The PAO elements for each number won’t really have anything to do with each other, but thats kind of the point and what the end goal of learning a system is anyway. It’s not for everyone, but if you do well with a more defined structure it might work for you.