Re: "The Evolution of Agency" by Michael Tomasello

On 06/04/2024 13:14, LDagget wrote:

John Harshman wrote:
 

On 4/5/24 2:13 PM, Arkalen wrote:

On 05/04/2024 16:02, John Harshman wrote:

On 4/5/24 4:13 AM, Arkalen wrote:

Hello all,
>
Has anyone here read "The Evolution of Agency" by Michael Tomasello ?

 

Maybe really short to start with:
>
Tomasello defines "agency" as a kind of goal-seeking system: a system that has a goal, is capable of perceiving the environment, verifying whether the goal is met, if not deploying a behavior that would move the goal forward, and looping between verification/behavior until the goal is met. Analogy is a robot lawnmower.
>
>
He argues that animal evolution has seen progressive complexification of agency that basically involves adding layers, with higher ones monitoring/controlling the lower ones. He focuses on the history of human evolution specifically and takes some example organisms from lineages that presumably match the level of agency a human ancestor would have had (he acknowledges convergent evolution of various levels of agency in other lineages but leaves it at that).
>
The levels he describes are:
>
* no agency - nematodes. There isn't goal-seeking, just stimulus-response. The animal eats food if it runs into it, escapes danger if it's present etc but doesn't really *plan* or anticipate beyond its immediate environment

. . .

* first level of agency - early vertebrates. Basic goal-seeking: a lizard will have distinct goals at any given time (seeking food,

. . .

* second level - early mammals. They have a second layer of goal-seeking feedback-loop system that pilots the first in order to not just achieve

  . . .
 

* fourth level - humans. Tomasello argues that the human "secret sauce" is essentially collective agency - reasoning agents like great apes that

more snipping

Thanks for that. Sounds interesting. My greatest immediate apprehension is that it would take a truly huge amount of experimental and observational evidence to test the various aspects of the scenario, and I wonder how much of it has already been done.

 Seems to me that it's very anthropocentrically biased.
The suggestion that nemotodes don't have goals as described is odd.
They seek food, seek mating. Hell, bacteria have goal seeking behavior in terms of seeking food,
or fleeing toxins via chemotaxis. Just because we understand some
of these things in terms of simpler biochemistry means what?

The word "agency" can mean many things and the book is clearly about defining a specific set of phenomena; the fact the word "agency" is used by other people to mean something different doesn't impact the substance of the argument, at most it could make one question the wisdom of the vocabulary choices.
Here the difference he proposes between organisms with and without "agency" isn't that we know the biochemistry in one case and not in the other, it's a specific claim about how they function and how many degrees of freedom the individual organism has with respect to their evolutionary hardwiring.
He defines a "feedback-control system" for agency that has the following features:
- a goal
- behavior(s) suitable to reaching the goal
- perception that verifies whether the goal is achieved
- a feedback loop between them to repeat the behavior until the goal is achieved
The thing is, living things don't *have* to function with this kind of organization. You could have an organism that chemotaxes towards nutrients & absorbs all it encounters and chemotaxes away from toxins or chemicals associated with predators/bad environments and reproduces once it's reached a certain size, and evolutionarily speaking that's a perfectly cromulent organism; if it can survive and spread this way it will. Insofar as it has the "goal" of eating or breeding or avoiding predators however that goal is evolution's goal more than the individual's. The nature of the goal and the specific behaviors it engages in to meet it change over the generations via evolutionary processes. Insofar as there is a feedback loop between perception and behavior that optimizes things towards the goal, the "perception" is "how does this organism interact with its environment" and the feedback loop is "is this organism reproductively successful".
But if you have a living thing that *does* have this kind of organization then "goals" have a different definition for it. "Evolution" still has the "goal" of it eating but the way this is behaviorally implemented means the individual itself can be described as having this goal in a completely different sense, that manifests differently. You can even get differences between "evolution's goals" and "the individual's goals" - usually not with eating but for example you can have goal-driven animals evolve the drive to have sex, although from an evolutionary point of view the actual goal is reproduction.
Now whether these different behavioral organizations actually occur in different animals the way the book claims is a different question but it seems a reasonable claim to me. Although it's true he is a bit ambiguous about nematodes; here is his discussion of them after introducing early animal filter-feeders as an example of the former [disclaimer: I said "nematodes", it's actually C elegans which may well not be a nematode at all in which case my bad]:
"Not only do the chemosensory neurons detect either good or bad things and 'signal' the motor neurons to produce bodily contractions that propel the organism either forward or away from those things, but C elegans also uses the rate at which it is ingesting food, typically bacteria, to detect the location of richer and less rich clumps (Scholtz et al., 2017). Moreover, if a behavior such as forward movement brings a bad result (e.g., a noxious chemical), the creature can perform one of two actions to move away (Hart, 2006). C. elegans finds its food by moving around in its environment actively, sometimes even learning the location of food in novel environments after several encounters (Qin & Wheeler, 2007).
The behavior of C. elegans would thus seem to be organized in a more complex manner than that of unicellular organisms. The have different mechanisms for sensing things in the world and acting in response. Classically, the function of a nervous system is to connect separate mechanisms of perception and action, and ganglia are seats of this integration, so it would seem that the separate mechanisms of perception and action are integrated in C. elegans (and also, by inference, in early bilaterians). However, it is unlikely that there is also a comparison with some kind of internal goal to create direction: their locomotion is mostly random or stimulus driven (Scholz et al., 2017). And these organisms do not seem to exhibit anything that we would want to call behavioral control: they do not inhibit or otherwise control action execution, and what they learn is simply the location toward which to direct their hardwired movements. It is thus unlikely that early bilaterians, as modeled by C. elegans, were goal-directed, decision-making agents, only animate actors."

 We can induce mating behavior in sea slugs with peptide hormones,
or they can induce those hormones themselves through other pathways.
 It seems an attempt to over-emphasize the use of central nervous system control which downgrounds gut level control or other physiological
control schemes as inferior. But why?

"Inferior" is your take, not the book's. The book is also razor-focused on behavior, only mentioning nervous systems in the context of describing behavior. You could claim that the author betrays an illegitimate preference for central nervous systems in their choice of model animals; that lizards have much higher behavioral flexibility than C. elegans and also have a much more complex brain which makes it look like the two are associated but the author could have described the exact same behavioral differences using sea slugs instead of lizards.
I'm guessing that this claim would be incorrect though, and that central nervous systems are in fact associated with higher behavioral complexity. It seems like you might disagree ?

Again, smells anthropocentric.
We can recall that many significant neuropeptide hormones stem from
what were first gut peptide hormones. Is there some innate advantage
across all life to relocating control systems?

Are you aware of this paper on a hypothesis for the evolution of neurons ? It's possible you do as it does include a link between digestive molecules and neurotransmitters.
https://onlinelibrary.wiley.com/doi/full/10.1111/tops.12461
As for an advantage to using nervous systems for control systems an obvious one seems to be the ability to link up different functions of the organisms in arbitrary ways instead of function-constrained ones. I don't think hormonal systems can do that as flexibly but I'm happy to be proved wrong ("can induce mating behavior with hormones" definitely isn't sufficient).

Sure, it's worked out well for those who currently have done so, but
it seems to be working well in those creatures who haven't.

To a first approximation anything any species does works well by virtue of that species existing. "Working well" isn't the standard. This book looks at behavioral complexity/flexibility. It doesn't cast judgement on different kinds of behavioral organization being good or bad, it discusses what they are.

It would be curious to simply test these ideas with observations
of ants. A botanist who studies complex communities might also
have some interesting commentary.
 

Tomasello thinks ants and other social insects are also goal-directed agents, it's one of the examples of potential convergent evolution of the trait he mentions. Interestingly, that suggests he thinks other insects aren't. I think it's a subject that would definitely merit developing and challenging but it's not done in the book.