It's poor AI. Wobbledogs are designed to be simple creatures, not fully aware of their own needs. They prioritise whims and interaction over said needs. This is why you'll often find a starving dog clinging to another dog. Personality traits also heavily affect this. If a dog is high energy and food-averse, they will have very little reason to independently seek food. A dog that has no traits is more likely to feed itself than one that has traits, unless it has the glutton trait. These traits take priority in a Wobbledog's impulses, leaving them to do things like sleep, play or other random things before eating.

While we can't control a Wobbledog's traits, we can encourage and discourage certain behaviours.

Reinforcements are what you get when praising or scolding an action. A dog's age affects how many times you need to praise or scold a specific action for the dog to acknowledge what you're telling it.
Puppies only need to be told what to do once in order to get what I call a "pip." Four "pips" maxes out either positive (praising) or negative (scolding) reinforcements.
There is also something to note, dogs will ignore a praise or scold action if they're already acknowledging one. It's best to wait until the reinforcement "bubbles" have disappeared to do it again.

Dogs earn "pips" as follows.
"Interaction" in this context refers to praising and scolding.
Puppies: Once per interaction.
Juveniles: Once per interaction.
Teens: Once per two interactions.
Young Adult: Once per two interactions.
Adult: Once per three interactions.
Ancient: Once per four interactions.

Using the reinforcements already mentioned, we can train our dogs as puppies and juveniles easily. By prioritising helpful training like "eat," "food," "sleep," or "toy," we can reduce the need to micro manage our dogs, and completely negate it after a few generations of training.

Dogs retain their training and even pass it on to offspring, making the max you can train a dog technically infinite, as you can overlap generations of training.

Some interactions are actually harmful to positively reinforce, like "grab." If a dog is trained to grab something, they'll grab anything. This includes food, poop, half-eaten food, dirt, toys, your roomba, and even other dogs. Because of how broad this is, you may find dogs harassing other dogs, dawdling, or even delaying your roomba (robot vacuum as it's called in-game) from cleaning up that cocoon puddle. For this reason, it's not always a walk in the park to train your dogs. For example, some things will overlap when training. If you train a dog to do anything with food; eat, throw or grab, both the relative action and "food" will be reinforced.

For this reason, you may want to break training up into specific patterns.

For food training, always do "eat," as the "food" action is broad. They'll do anything that they can with food, be it grab, throw, hoard or eat it. You want to train the specific action you're after.
When it comes to toys, you're not really going to win with one training session alone. The actions you can command a dog to do with a toy are limited, being only "throw" and "grab." Both of these are not great as they'll do these to anything, including other dogs, food and poop. For this reason, when training for toys, you want to break it into two steps.
1, you train "grab toy," maxing out both "toy" and "grab."
2, you introduce the dog to something else that doesn't have overlapping action. My best option is another dog. Tell your dog to grab the other dog and then scold it. Repeat this process until "grab" is either at neutral, or lower if you want it there. Once done, your dog will have max "toy" pips, and either no "grab" pips, or it'll have negative "grab" pips.

Sleep is the same as food, as there's no negative behaviour to get rid of, unless the dog is actively holding something when it falls asleep, as it will then praise "grab" as well.

My own personal setup is to get a few pieces of food, one toy, and a newly-hatched dog into a corner or fenced area. I make them eat the food and praise them twice per piece, once when they start and once when they finish. Then, once they've positively maxed "food" and "eat," I move onto the toy, making them grab it, praising them, waiting a few seconds and repeating the process. They won't take the action as valid if they're still going through an AI command. Waiting for the reinforcement bubbles to disappear before praising again is the answer to not having to repeat the process too many times. Once I've maxed "toy," I move onto sleep. I make sure the dog maxes that out, and then I introduce them to another dog. I have them grab the dog, and scold them. Repeat four times. This removes "grab" from their positive reinforcements. That way, they're maxed on necessities, but aren't encouraged to harass other dogs. Anything past that, like sit, speak, or roll over training is done as optional.

For dogs that can't walk, I only max out their sleep and food reinforcements, leaving entertainment to petting when needed. I generally also leave them in a closed-off box so they can't "buck" or wriggle away from food.



If any of that seemed partially repetitive, it's because I copied part of a comment I made on another post about why Wobbledogs starve themselves. That post is actually what inspired me to make this guide.

Dogs that can't walk may be a challenge, as they're not always capable of getting to things that will help with their needs. As mentioned in the previous section, locking them in a space where they can't leave food is a good way to manage them. However, if the food is too close to their face, the dog will try to back up, and will panic if it can't. If it's legless, it will try to "buck" away from the object in its face. For this reason, I rarely ever fill a pit with food, I usually only fill a corner of the pit and just make sure the dog is close to it. If this doesn't work with the specific dog, then I do fill the box/pit, scold bucking until they're maxed on negative reinforcement, and leave them alone. If this still doesn't work, then I use the gravity machine to pin them to the floor in a room other dogs can't get to, so nobody gets trapped.

The gravity machine is received after decorating the walls and floor of a room in a space theme. Any kind of space theme works, be it the spacelab wallpapers, moon floor, or even solar system diagram floor. But they both have to be decorated in a space theme at the same time. Once the gravity machine is unlocked, you can redecorate the room if desired.

These other guides aren't my own, but may help with other struggles or better explain things I've talked about here.

The Moth's "Wobbledogs Master Guide"
https://gtm.steamproxy.vip/sharedfiles/filedetails/?id=2569837279

Lemon's "How to disable dogs dying!"
https://gtm.steamproxy.vip/sharedfiles/filedetails/?id=2684114030

Destiny's "Forbidden Knowledge (Commands Master Guide)"
https://gtm.steamproxy.vip/sharedfiles/filedetails/?id=2866425429

Through experimentation, I've discovered a few other things that I'll work to compile after my school year concludes.

One thing I can note right now, is that ghosts seem to only need three reinforcements for one pip, though this does need further testing.

Another thing I've found is that you can easily train "dispense food" by just letting your dogs get hungry, and then trapping them with a food dispenser. Keep an eye on them, praise when they hit the button, and delete the food so they can't eat until they've maxed their reinforcement. Just be aware not to let them starve to death during this training.