Robot Theory of Mind

[Black Rose]

In this thread, I will explain the article I found about a brain model of the theory of mind in robots.

We have an attention of what others know and what we know. This is recurrent in memory. When we see that others have a false belief this creates a conflict in the neuronal circuits. So we pay attention to the other and ourselves at the same time. The outside information of what the other knows and the internal information of what we know create a pathway difference. We know what we know and they know what they know. So if they have a different belief this inhibits the path we had about our belief yet we know we have our belief and theirs is false.

We switch the path to a new circuit, the one where our belief is current (we remember what we know) then we switch back to the path where they held their precious belief.

Thus we have (my belief -> their belief -> their belief changes -> my belief if different - look at their current belief)

This is a 4 path inhibitory circuit. (check figure 2)

-

I find it very interesting that with the model they said 2.5-year-old humans can do this but without language just by looking at the objects others have seen. This is an inbuilt circuit for predicting what other minds do. What matters is that because it is recurrent it can build up higher representations in working memory later on.

Frontiers | A Brain-Inspired Model of Theory of Mind

Frontiers | A Brain-Inspired Model of Theory of Mind

Theory of mind (ToM) is the ability to attribute mental states to oneself and others, and to understand that others have beliefs that are different from one'...

www.frontiersin.org

This was done in 2020 so they had powerful computers. Imagine what they could do if they put this model in a simulated human. And imagine what they could do if they modeled executive control in the prefrontal cortex.

Playstation 4 human:

Cognitive Control:

Executive Function Brain's Control Center​

 

[fractalwalrus]

Well, yes. A robot would likely have to use the processes you described here to infer the state of mind that their fellow mind-havers would be experiencing. Interestingly enough, Autistic humans struggle with this functionality: https://www.psychologytoday.com/us/...dren/202303/autism-and-the-theory-of-mind-tom. Really makes you think about robots and humans, and cognitive abilities in general.

 

[sushi]

how do you explain animals then, are their minds robotic or biological and irrational

 

[fractalwalrus]

how do you explain animals then, are their minds robotic or biological and irrational

If you are questioning whether animals can have executive function specifically, well some have performed research that would suggest so: https://www.mdpi.com/2076-2615/13/3/533. I would expect the extent of their abilities to vary by species, but they do have some capacity to perform certain tasks in a manner similar to humans. We would tend to classify animals as biological (which would imply that their brains are as well), but as for rational versus irrational, not even humans can meet that threshold entirely depending on one's definition. We are emotional creatures with the capacity for reason, but in order for an individual to "act rationally" they need a goal to be used as a metric for that behavior to be analyzed as rational. It is rational for one to satiate their hunger by eating, for example. However, I may not be rational for one to gourge upon food if their objective is weight-loss. As for robot vs human. humans tend to be the programmers of robots (for now), and often their designs for their creations are inspired by other natural designs (https://www.popularmechanics.com/science/animals/g28912650/animal-inspired-technologies/). I would expect that an engineer of a robot would likely attempt to mimic the processes of human cognition, though with silicon rather than neurons (in fact, computers already attempt this (they do have both working and long-term memory, but stating that binary is the language of the human brain is not accurate).

 

[sushi]

@fratalwalrus

the debate is like beast wars transformers, are they beasts or are they robots.

 

[fractalwalrus]

@fratalwalrus

the debate is like beast wars transformers, are they beasts or are they robots.

How would you differentiate between the two? If a being made of an elaborately arranged housing of oil capsules we refer to as cells makes one a beast, then animals are beasts. If you were implying that robot minds could not replicate the processes seen present in human minds, well that may be a matter of definition and classification. A robot can be argued to be able to perform the same analysis-based tasks that humans do, even if the internal mechanisms that precipitate said processes are different, unless you believe that there is something fundamentally different about the very nature of the task itself which depends on the pieces used to accomplish the task. If this is this case, then seeing as individual humans have brains with slight variations in arrangement or neuronal pathways, it would imply that each of us who does a task someone else has done before is actually not doing the same exact task at all because of the variation in how it was accomplished. So the question is, then, is one task said to be different when the outcome or the methodology of said task varies, or both?