I’m aware that my brain may group things in ways that aren’t related to useful criteria or criteria I would endorse. My brain was doing this anyway before I wrote the post. Discussing it is an essential part of noticing it and self-modifying or compensating in some way.
How, specifically, do you think that having this discussion is arrogant and callous? What would have to be different about it for it not to be arrogant and callous?
My objection is that answers to (1) are being confused with answers to (2). In particular, a reductive (non-agenty) answer to (1) will tend to drift towards a reductive answer to (2).
The “arrogance” I see stems from the bias towards using non-reductive models when dealing with behaviors we approve of, and reductive models when dealing with behaviors we don’t approve of.
For example, consider a devout Mormon, who spends two years traveling in a foreign country on a religious mission. Is this person an agent? Those already sympathetic to Mormon beliefs will be more likely to advance an agent-like explanation of this behavior than someone who doesn’t believe in Mormon claims.
As another example, is Eliezer an agent? If you share his beliefs about UFAI, you probably think so. But if you think the whole AI/Singularity thing is nonsense, you’re more likely to think of Eliezer as just another time-wasting blogger best known for fan-fiction. Why doesn’t he get a real job? :P
Can a smoker be an agent? We tend to assume any unhealthy behavior has a non-agenty explanation, while healthy behaviors are agenty. We can’t imagine the mind of an agent who really doesn’t believe or care that smoking is unhealthy.
Your roommate doesn’t wash the dishes. Have you tried imagining a model of her as an agent, in which she acts in accordance with her own values and decides not to wash the dishes? If she places very little internal value on clean dishes, she may not be able to relate to the mind of an agent who places any value on washed dishes. She may even be modeling you as a non-agent with a quirky response to the stimulus “dirty-dishes”. (Did your parents never mistake a value they didn’t understand for non-agenty behavior on your part?)
People of one political ideology utterly fail to model those with other political ideologies as agents, even while considering themselves to be agents.
Are athletes agents? Becoming an athlete requires as much goal-directed behavior as becoming a mathematician. But mathematicians often have terribly condescending opinions of the agenty-ness of athletes.
In all these examples, value judgments precede the models we make of agency-ness, which will slide towards evaluations of actual agency-ness.
It’s very hard to create a non-reductive model of an agent with sufficiently alien values. Even defining agency in terms of “executive function” still relies on a comparison to your own mind (to decide what counts as goal-directed and how much it should count). Since a failure to build a model suggests that no model is possible, we will be biased towards only considering minds similar to our own as agents. The result in an strong in-group bias.
I hope this is a bit clearer. I don’t mean to criticize you for reporting patterns in your own thought, and I understand that’s not meant to make any normative claims. To report some patterns of my own, the “red flags” my brain noted reading the comments were the phrase “Heroic responsibility drive” and a bunch of really math-y types claiming they had everyone else figured out.
Agreed that you have to be very careful about letting your answers to (1) slide into your answers to (2). But I don’t think you c an do th
If she places very little internal value on clean dishes, she may not be able to relate to the mind of an agent who places any value on washed dishes. She may even be modeling you as a non-agent with a quirky response to the stimulus “dirty-dishes”.
Oh, she likes clean dishes all right. She nagged me about them plenty. It was just that her usual response to dirty dishes was “it’s too ughy to go in the kitchen so I just won’t cook either.” She actually verbalized this to me at some point. She also said (not in so many words) that she would prefer to be the sort of person who just washed dishes and got on with life. So there was more to “what she said” than saying to me that she would wash the dishes (which someone who didn’t care about dishes might say anyway for social reasons).
Obviously all people are agents to some degree, and can be agents to different degrees on different days depending on, say, tiredness or whether they’re around their parents. (I become noticeably less agenty around mine). But these distinctions aren’t actually what my brain perceives; my brain latches onto some information that in retrospect is probably relevant, like my roommate saying she wants to be the sort of person who just washes dishes but not washing any dishes, and things that aren’t relevant to agentiness, like impressiveness.
I’m aware that my brain may group things in ways that aren’t related to useful criteria or criteria I would endorse. My brain was doing this anyway before I wrote the post. Discussing it is an essential part of noticing it and self-modifying or compensating in some way.
How, specifically, do you think that having this discussion is arrogant and callous? What would have to be different about it for it not to be arrogant and callous?
Sure, I should have been more specific.
Here are two questions:
1) How do I model the minds of other people?
2) What are the minds of other people like?
My objection is that answers to (1) are being confused with answers to (2). In particular, a reductive (non-agenty) answer to (1) will tend to drift towards a reductive answer to (2).
The “arrogance” I see stems from the bias towards using non-reductive models when dealing with behaviors we approve of, and reductive models when dealing with behaviors we don’t approve of.
For example, consider a devout Mormon, who spends two years traveling in a foreign country on a religious mission. Is this person an agent? Those already sympathetic to Mormon beliefs will be more likely to advance an agent-like explanation of this behavior than someone who doesn’t believe in Mormon claims.
As another example, is Eliezer an agent? If you share his beliefs about UFAI, you probably think so. But if you think the whole AI/Singularity thing is nonsense, you’re more likely to think of Eliezer as just another time-wasting blogger best known for fan-fiction. Why doesn’t he get a real job? :P
Can a smoker be an agent? We tend to assume any unhealthy behavior has a non-agenty explanation, while healthy behaviors are agenty. We can’t imagine the mind of an agent who really doesn’t believe or care that smoking is unhealthy.
Your roommate doesn’t wash the dishes. Have you tried imagining a model of her as an agent, in which she acts in accordance with her own values and decides not to wash the dishes? If she places very little internal value on clean dishes, she may not be able to relate to the mind of an agent who places any value on washed dishes. She may even be modeling you as a non-agent with a quirky response to the stimulus “dirty-dishes”. (Did your parents never mistake a value they didn’t understand for non-agenty behavior on your part?)
People of one political ideology utterly fail to model those with other political ideologies as agents, even while considering themselves to be agents.
Are athletes agents? Becoming an athlete requires as much goal-directed behavior as becoming a mathematician. But mathematicians often have terribly condescending opinions of the agenty-ness of athletes.
In all these examples, value judgments precede the models we make of agency-ness, which will slide towards evaluations of actual agency-ness.
It’s very hard to create a non-reductive model of an agent with sufficiently alien values. Even defining agency in terms of “executive function” still relies on a comparison to your own mind (to decide what counts as goal-directed and how much it should count). Since a failure to build a model suggests that no model is possible, we will be biased towards only considering minds similar to our own as agents. The result in an strong in-group bias.
I hope this is a bit clearer. I don’t mean to criticize you for reporting patterns in your own thought, and I understand that’s not meant to make any normative claims. To report some patterns of my own, the “red flags” my brain noted reading the comments were the phrase “Heroic responsibility drive” and a bunch of really math-y types claiming they had everyone else figured out.
Agreed that you have to be very careful about letting your answers to (1) slide into your answers to (2). But I don’t think you c an do th
Oh, she likes clean dishes all right. She nagged me about them plenty. It was just that her usual response to dirty dishes was “it’s too ughy to go in the kitchen so I just won’t cook either.” She actually verbalized this to me at some point. She also said (not in so many words) that she would prefer to be the sort of person who just washed dishes and got on with life. So there was more to “what she said” than saying to me that she would wash the dishes (which someone who didn’t care about dishes might say anyway for social reasons).
Obviously all people are agents to some degree, and can be agents to different degrees on different days depending on, say, tiredness or whether they’re around their parents. (I become noticeably less agenty around mine). But these distinctions aren’t actually what my brain perceives; my brain latches onto some information that in retrospect is probably relevant, like my roommate saying she wants to be the sort of person who just washes dishes but not washing any dishes, and things that aren’t relevant to agentiness, like impressiveness.