The opening assumption is that the agents in my game think and act like they are human. Agents are going to have people-like limitations:
Agents are going to have people-like needs:
My thought is that to do this properly, we should probably walk through the life of a human being, an re-create most of the milestones in intellect that science has charted. By running through, in order, we might even re-construct the idiosyncrasy of the human mind. That is by no means a complete model of the development of a consciousness, nor even one that is bound by what is know by cognitive science. It is a working theory that I'm using as the foundation to explore the matters further, and to put together a simplified theory of mind.
Step 1: Newborn. Newborns have no concepts. They just understand (via nerves and reflexes) when they are uncomfortable. When they are uncomfortable, they cry. Let us establish that there is a central concept of self, and that self has a state of contentment. When that state of contentment is false, there is a desire to communicate that condition to the outside.
Step 2: Infant. Infants start to realize mother is the source of comfort from the outside. And that mother is one of several other people. (Including, possibly, father.) Infants start learning that they can perform actions to alter the outside. They also learn that in addition to other people there are things.
Step 3: Toddler. Toddlers learn that they own certain possessions. Toddlers learn that they can start to address their own wants. And when they can't address their own wants they can bargain with another party to address that want. Toddlers also learn that certain activities and treats make them happy.
Step 4: Child. Children learn the inside of the outside is a place called home. They also learn that there are different rules for inside the home and outside the home. Children also learn to deal with familiar people and strangers. Also they learn to accept guidance from individuals other than mother and father, namely teacher.
Step 5: Adult. Adults learn that life ends at death, and that most things are finite. They learn that sex creates new life. They learn that they can exchange money with other people for goods and services. And vice versa. They also learn that some individuals cheat, rape, steal and kill. They guard themselves, their family, and their property from violence and crime. They also learn that people who commit a crime are punished.
The next step is to build a data structure to encapsulate this simplified truth into a way that I can start subjecting to stimulus and get a response. With luck, the final form will be flexible enough to accept a database of truth, and a personality will pop out. But... baby steps.
Compared to modeling the mind, modeling the body is actually somewhat easy. Role playing games have been building such models for years. Most of the systems out there center around devising a metric for health, expressed as a number, and then providing ways to add or subtract from that health in response to events in the game. There are also metrics for the ability to reflect or prevent damage from various sources, as well as to inflict damage. The games also provide models for physical limitations and impediments. If a hand is needed to operate a sword, and the character's hand is injured, their ability to wield a sword is impaired. If the character is under the affect of a cone of silence, they can't cast magical spells.
Beyond the body, we need to consult modern psychology. And fortunately there is a wealth of research to build on. The famous theory applicable to this discussion is Maslow's hierarchy of needs. In that hierarchy he defines 5 components of psychological health, and the priority of each. The theory also surmises that the longer an important need is unfulfilled, the more likely it is to cause a behavior change or disruption. Also some needs become more urgent over time as they are unfulfilled.
There is an important point here. A mental model is not enough. We need to have some sort of drive that is triggered by the state of this mental model. Those drives need to have behaviors, and those behaviors are designed to resolve imbalances in their model.
Aha... I think I'm finally back on track!
For the purpose of the Epic of Gilgamesh, there are certain things that I think we can just ignore. Hunger. Thirst. Sanitation. Sleep. These are time based items that are tedious to model and tedious to deal with. Perhaps on a citywide scale we can model a food shortage, but individuals in the game we will just assume are handling this issue on their own. How many books in literature take you through the character's daily meals and bathroom habits? I rest my case.
I'll also point out that after playing Fallout 76 for a few weeks, I'm a bit tired of dealing with hunger and food bars. That was a cute mechanic at the beginning. It's getting really old now.
For non-player humans, we can take a page from Maslow. They are just looking for a safe environment where they can fit in, feel secure, make money, fall in love, and do whatever seems to content them. I realize it's not all that simple, but from 40,000 feet, that is a complete answer. There are also going to be plenty of non-player humans who can't achieve one or more of those goals. And that will drive their behavior.
The question then becomes, when I start modeling Gods, what is their motivation. Gods in this era really seem to go out of their way to encourage mortals worship them. Including in that behavior:
Let us assume in our model that Gods get some sort of reward for the attentions of human worshippers. Maybe it's like a food. Or a spiritual currency. Or maybe just bragging rights. Gods really seem to like (on the border of need) worshippers. At the same time, Gods are very competitive with the other Gods for human attention. Heck, just look at that Yahweh character. Sheesh.
There also seem to be occasions in ancient times where the Gods engaged in physical interactions with mortals. Namely Sex. And namely how demigods (like the player's character) are made. Gods also seem to enjoy sex with the other Gods, thus the Sumerian and Greek Pantheons.
Gods also seem to alter their behavior the longer their needs are neglected, so we can assume the same sort of need/fulfillment reflex mechanism we see in humans. They just have different needs and different priorities.
Each character, be it mortal or god, also seems to have an individual "need". The character Gilgamesh is driven by a need to live forever. The God Zeus is a notorious philanderer. What if that was caused by some sort of internal imbalance?
Long story short, it looks like there is a common cognitive architecture I can use to model both human and divine characters. And even better, that architecture is going to be a dynamic one, that can drive behavior.