3 ms·

> [Agency is] built into the brain's fabric and architecture. Because even neurons have agency It seems like a borderline category mistake to claim that becaus

> [Agency is] built into the brain's fabric and architecture. Because even neurons have agency It seems like a borderline category mistake to claim that because the components of some system have property P then that system is somehow better fitted to have property P itself. There are exceptions (like P="reliability", or P="speed", thinking of transistors in a chip, perhaps) but in general, it doesn't seem to follow. Few transistors have Turing completeness, for example, but could be components of a system possessing that property. I don't see prima face why Agency should belong to the class of properties that transfer upwards from component to composed system. Anthropomorphizing neurons by ascribing "selfishness", "adventurousness", and so on to them seems to compound this error in an effort to persuade. I doubt Dennett intends this in the excerpted quote. Anyway, the "heterarchy of agents" view of mind doesn't need personality-endowed neurons, and has some compelling features which are explained pretty well in the article. Regards "level 4, the self", though, I wonder how well evidenced a "prime-ministerial agent" type of self is. The article seems to take that on faith versus alternate models such as a set of competing high-level agents that compete or cooperate to occupy the driver's seat. Equating this with abnormal states of affairs like schizophrenia or demonic possession seems a bit facile.. (even disregarding the implied antique three-faces-of-Eve view of schizophrenia.) Just because there aren't gross changes in behaviour or mentation, how would you know that a single top-level agent is consistently running the show? We're all familiar with mood-swings and role strain, but our strongly held belief in a ruling self makes us attribute these to other factors than agent swap-out. I think it's equally plausible that pathological multi-agency could be ascribed to abnormally divergent agents, rather than the mere existence of more than one top-level agent. Maybe swapping top-level agents in and out is SOP, and a uniquely privileged Self should be treated with caution as a possible metaphysical fiction.

ZeroFries 12y ago
If we agree that system A possesses agency, then either the components possess it, or at some level of abstraction it emerges from the relations of the components. The property of wetness emerges from the relations and mechanics of water molecules interacting with other water molecules. A hydrogen or oxygen element does not posses wetness, nor does it emerge from the interaction between a hydrogen and 2 oxygen. However, wetness is a binary category, where as agency as defined in the article is a continuous category. We say that a slave has some agency, but less agency than their master. If we define 0 as no agency whatsoever and 1 as ultimate agency, then it's likely that everything in the universe has agency 0 < x < 1. This of course still doesn't show that agency necessarily increases as you go up the abstraction stack. Does a city have more agency than it's citizens? What about the set of all living things? I would say that the molecules of a rock have more agency than the rock itself. I think the agency property really breaks down when you look at it hard enough.
vilhelm_s 12y ago
Clearly it can't be agents "all the way down" (in any case, the component parts of the neurons do not have agency), but the idea that it will be easier to evolve a high-level agent if the components themselves are agent-like seems coherent. It would be interesting if this could be sharpened into a testable prediction, in some kind of toy model. I could see it go the other way too--for example large companies and states struggle to act as coherent agents, because the individuals and business units inside them scheme against each other, pursuing their own agendas. In order to maintain high-level agency they need to suppress low-level agency, e.g. armies training their soldiers to blindly follow orders.

ZeroFries 12y ago

If we agree that system A possesses agency, then either the components possess it, or at some level of abstraction it emerges from the relations of the components. The property of wetness emerges from the relations and mechanics of water molecules interacting with other water molecules. A hydrogen or oxygen element does not posses wetness, nor does it emerge from the interaction between a hydrogen and 2 oxygen. However, wetness is a binary category, where as agency as defined in the article is a continuous category. We say that a slave has some agency, but less agency than their master. If we define 0 as no agency whatsoever and 1 as ultimate agency, then it's likely that everything in the universe has agency 0 < x < 1. This of course still doesn't show that agency necessarily increases as you go up the abstraction stack. Does a city have more agency than it's citizens? What about the set of all living things? I would say that the molecules of a rock have more agency than the rock itself. I think the agency property really breaks down when you look at it hard enough.

vilhelm_s 12y ago

Clearly it can't be agents "all the way down" (in any case, the component parts of the neurons do not have agency), but the idea that it will be easier to evolve a high-level agent if the components themselves are agent-like seems coherent. It would be interesting if this could be sharpened into a testable prediction, in some kind of toy model. I could see it go the other way too--for example large companies and states struggle to act as coherent agents, because the individuals and business units inside them scheme against each other, pursuing their own agendas. In order to maintain high-level agency they need to suppress low-level agency, e.g. armies training their soldiers to blindly follow orders.