Maybe Dayan's "Twenty-Five Lessons from Computational Neuromodulation"?
01.03.2026 17:37
👍 2
🔁 0
💬 0
📌 0
Maybe Dayan's "Twenty-Five Lessons from Computational Neuromodulation"?
Thanks! And yes, the essay is nice, not only on the confounder issues. I think its approach to exercising for health is on point.
Do you have a gift link, by any chance?
I dislike summary RoB scores in general. They often fail to convey meaningful information. Potential sources of bias assessed by different items usually vary in their potential impact and implications. It makes more sense to identify common risks and assess their potential effects on the results.
Precisamos reformar o ensino superior para a era da inteligência artificial?
Spoiler: não.
doi.org/10.64628/ADE...
I think most of what we know of the neural mechanisms of behavior points clearly to that. I also don't see clear alternatives
In this case the relevant structure is in our body, especially our brain, and while untangling the full causal chain is currently unfeasible for complex behaviors, I don't think most people disagree that the structure is a crucial *part* of the answer for what caused the behavior.
I am not sure I understood your point, Vincent.
What threat? My point is precisely that your indeterminism framework produces nothing that wasn't already there with determinism. If there is a threat, indeterminism is not its solution.
You are implying that emergence of macroscopic control structures is impossible under determinism but becomes possible if we add randomness. I don't see how this is supposed to work. The origins of life remain a mystery in both scenarios.
In a sense, maybe. But I think that would be somewhat different from the common use of the feedback loop concept.
I got scrambled with the tread function on the comment you replied to, but take a look at my subsequent comments there. But succinctly: entire scientific fields show this could happen - cybernetics, control systems theory, most of physiology. Indeterminism may be true, but is unnecessary for agency.
6/ ... it does not really preclude their existence in a deterministic world. As you pointed out, life's structure evolved to partially insulte it from (or even explore) lower level fluctuations. The source of these fluctuations - deterministic or not - seems tangential.
5/ What matters is whether higher level/structural constraints on lower levels dynamics are possible, and they are - with or without determinism. So even if you are right about the universe being fundamentally underdetermined, your argument only shows that agents are possible in such universe...
4/ I see nothing in determinism that precludes any of this. To give a more biologically relevant example, Brownian motion is consistent with both determinism and true randomness, but in neither case does it preclude cells from applying structural constraints to it and produce controlled behavior.
3/ To give a silly example, if the dynamics of an explosion is perfectly determined, you could still build a canon by constraining it. Does a steam engine requires fundamental randomness to work? Do negative feedback loops only produce goal-directed behavior under indeterminism?
2/ Let as assume that randomness in the lower levels is actually epistemic - like in classical statistical mechanics, or the kind of apparent randomness we know can emerge from nonlinear deterministic systems. You could still apply higher level constraints to it and build an agent.
1/ I agree with your argument in that, assuming ontological randomness at the lower leves, you can build agents through top-down constraints. Were I disagree is I that believe we can get the same result starting from determinism.
Yes, fully agree. I think your final remark about whether one could have decided otherwise is particularly on point.
And cybernetics and systems control theory provide many examples of sophisticated, goal-directed systems that do not require indeterminism either. This doesn't mean the universe is deterministic, but it shows that if it were, top down causation and agency would still be meaningful constructs.
There are also countless examples of devices from everyday life where higher level constraints produce work from a system whose lower level regularities wouldn't do useful work by themselves, and we don't need indeterminacy to explain them. We can certainly understand car engines under determinism.
I agree - that is what *agency* is about. But when that agent was built strongly influenced by genes and early environments that they had no control over, any claims of freedom, while still meaningful in my opinion, must be significantly constrained.
I may be biased, as I came to neuroscience from physiology, which abounds with mechanisms of regulation/control, often with a cybernetic flavour, that can be (and usually are) conceptualized in a deterministic framework. So I don't see how top down control should be impossible under determinism.
3/ What remains is the question of freedom. An agent can act according to its internal structure - in accordance with itself - but this structure emerges shaped (at least in part) by factors outside the agent's control, even if we accept indeterminism. The (in)compatibilism debate remains.
2/ But the "causal slack" this indeterminacy is meant to create would also exist in a deterministic universe and could produce agency - and if the argument works with determinism, it is essentially compatibilist.
1/ I'm yet to read the paper, but I saw a recent talk from Kevin where he discusses it at length, in the context of his other recent work. I must say it felt like an herculean effort to avoid the label of compatibilist.
I don't think this argument makes sense. Determinism does not imply the microscopic world is ordered from the perspective of higer levels, which is what matters for agents. Top-down constraints and control can exist both with determinism and indeterminsm.
Dopamine ≠ pleasure
