Re: Evolution of consciousness

On 02/05/2024 19:03, Mark Isaak wrote:

On 5/2/24 6:21 AM, Arkalen wrote:

On 30/04/2024 01:36, Mark Isaak wrote:

My views on the evolution of consciousness are starting to gel.
>
1. Rudimentary nervous systems evolve.
2. Brains evolve, capable of memory and of decisions other than reflex.
3. Those decisions probably work better if the brain has a model of the world to work with. So such a model evolves.
4. Some creatures live socially. Their brains need a model of that important aspect of the world: the fellow beings one lives with, including how they think.
5. So we've now got a model of minds. How about if we apply it to *our own mind*? That might make our thinking about interactions with others' minds more efficient.
6. Viola! Consciousness!
>
Does that make sense to people?  Is it time for me to write a book on the subject? (Do you think publishers will want the book to be more than 106 words long?)
>
There's also the problem of testing it. I'm open to suggestions there, too. Step 4 implies that the model of how we think need not agree with how we think, much as the mental model of our world is flat, not spherical. This has at least some confirmation (e.g., blindness to many biases). More would be better.

>
Have you seen my thread on Michael Tomasello's "The Evolution of Agency"? I think the book would interest you. If you want more detail I have a post somewhere in that thread summarizing its arguments, I'd be happy to hear your take.

 I have seen it, but I don't remember particular points.
 I just came across reference to another book by Michael S.S. Graziano, _Consciousness and the Social Brain_, which appears to make an argument similar to mine above (particularly steps 4 and 5).
 

Basically (if you don't mind me going on about it again) he proposes a scheme similar to what you did but more specific, fleshed-out and (IMO) convincing. It revolves around the notion of "agents" or "agency" which Tomasello defines as a system that achieves goals via a feedback-control mechanisms where the system perceives aspects of the environment, compares them to the desired goal, engages in behaviors meant to bring it closer to the goal, checks the environment again, and loops this way until the goal is achieved.
His parallels to your steps might be:
1) rudimentary nervous systems evolve that coordinate perception with behavior on a stimulus-response basis but not the feedback-control system involved in true agency.
2) brains evolve that do implement such a feedback-control system [I'm not sure in the book he explicitly associates it with brains, but he does associate it with vertebrates which do have distinct brains as a feature so I'll say it's close enough for a paraphrase]
He doesn't have a parallel to your step "3" because models of the world are implicit in all of the cognitive models he presents, in fact the differences in he calls "experiential niches" (which could be thought of as "world models") are pretty important. So for example he points out that with agency comes the mechanism of *attention* (i.e. you orient your perceptions in specific ways depending on what goals you're working towards and where you're currently at in working towards them) which implies experiences of an outside world and internal states that are or aren't in sync, full of things that are relevant/irrelevant, good/bad etc.
4) He does bring in social living as a possible cause of his next step in the evolution of agency that he sets at early mammals: the appearance of a feedback-control system applied on top of the previous one to monitor and control the goal-seeking process itself (he sees social living as a driver for this because of the competition between peers would induce a benefit in more flexible, efficient decision-making). These early mammals would be able to not only perceive the world, pick a behavior to fulfill a goal and shut everything down in case of danger (as he describes lizards doing), but mentally play out possible behaviors and flexibly inhibit some in favor of others depending on which they anticipate working out best. This would introduce into the "world model" or "experiential niche" notions of goals, behaviors and cause-and-effect relationships between the two. I don't think he introduces models of other *minds* at this step per se although it's a bit like world models - they're implicit in several steps it's more of a question of what aspect of minds is being modelled.
5) I do think there is still some similarity between your 5 and the next level of agency Tomasello suggests, although he sets it at great apes and you seem to set it at humans (then again many would argue great apes are conscious and I don't think Tomasello would disagree). He proposes an extra metacognitive feedback-control system monitoring the lower ones allowing control not only over the behaviors taken in service of a goal but of the goals themselves, and an understanding of cause-and-effect in general and not only as concerns one's own actions. It also induces an understanding of others as being agents with goals they behave in service of.
6) While he does think of 5 as the ability to reason and I'm pretty sure would call it "consciousness" he does have 2 other steps separating humans from that, which involve collective agency. He proposes the critical difference between humans and other great apes is the ability to coordinate as part of a group that itself fits the criteria for being an agent - with collective goals, the ability to monitor their completion and act and self-regulate in service of them. He sees this as coming in two parts, first the ability to coordinate pairwise to achieve specific tasks (somewhere in hominid evolution - he gives several examples illustrating how strikingly worse chimpanzees are at basic cooperation than even human children) and then the ability to function as part of a larger community with shared norms that allow coordination with strangers (which he sets early in the evolution of our own species). He talks about this inducing a kind of triple mental model of agency, the "self" agent (the individual's goals, parallel to the sense of agency of other great apes), the "role" agent (the goals implied by one's role in some collective enterprise) and the "collective" agent (the goals of the collective enterprise itself). He then talks about how various aspects of our experience like culture, morality etc follow from that.
I think it's interesting how this suggests a difference between having a model of one's own mind, having a model of others' minds, and having a model of *mind in general* that's then applied to oneself and others. "Models of the world" and "models of the mind" really collapses a lot of functionality and variability and I think Tomasello's model does a better job of separating out different potential strands and honing in on those that actually account for how we resemble and differ from other species.
I also like how this model justifies that the last step, and only the last step, is truly self-reflective. All the other steps involve taking a system at a certain level of agency and adding a monitoring/control level, resulting in a system that's aware of itself *as a system of the lower level*. That last step is the only one that involves the system monitoring/controlling a level *above* itself, and indeed being able to monitor/control any arbitrary system of agency at all (given any combination of humans can display collective agency and a human can be part of multiple collective agencies at any given time). Meaning the recursion ends there, it's the only agent model that can model itself as being the level it is.