Epistemic status: Thinking in public[1]
Warning: Moderate infohazard
Here's an experiment you can do to see why this hypothesis might be true:
Put part of your body under water that is moderately colder than you'd like. Direct your attention towards the sensation. If you're anything like me, you'll notice that your attention will automatically pull away from it.
Here's an experience I expect you've had that's also illustrative:
Maybe you've stubbed your toe or you have a stomache ache. It hurts. After a while, you stop noticing the pain, but if you redirect your attention back there you notice it again.
This leads to a very important question:
When you weren't noticing the pain were you:
- Not experiencing suffering OR
- Experiencing suffering (likely at a lower level though) and not noticing it
Notice that if we don't make the right call here, it'll mess up a bunch of our attempt to run calculations be they on human welfare, animal welfare or AI welfare.
Additionally, if 2 is true then:
- There is much more suffering in the world than we think
- We should be at least somewhat skeptical of anesthesia being effective[2]
- Chronic health conditions are much worse than what sufferers believe
- Mild pain that last for a long time, such as when you have a cold or eat a bit too much is much worse than people typically believe. I suspect that social perceptions around the degree of pain involved anchor us quite strongly.
Even if 2 is false, it doesn't mean that we are out of the woods. Insofar as humans automatically look away from pain[3] (which I believe to be true), you'll still experience brief amounts of suffering when you accidentally notice something painful and then flinch away. Often this will be so quick that you won't even remember that you experienced pain.
Some people (those who focus on higher-order cognitive functions as the core element of suffering) will say that this isn't real suffering or that it is only mild suffering at worst. I don't want to dismiss this insofar as I find Global Workspace Theory plausible AND insofar as I find it plausible that transmitting suffering to the rest of the brain causes suffering to spread to other parts of the brain[4].
However, this isn't obviously true either, so let's consider what it would mean if this latest hypothesis (about higher-order cognitive functions) were false. In this case, these repeated quick flashes of pain could add up[5] to a rather significant amount of total pain, especially since if you don't consciously notice that the action gave you a little zap, it would be very easy to accidentally turn your attention to it a few more times before your brain subconsciously learns to avoid looking that direction.
Honestly, the argument I've just presented is feeling a bit weaker than I expected it to be when I started writing it. Nonetheless, I still believe it to be true. I'm not willing to fully explore this strand of thought publicly at this time as I believe it's important to be very conservative around infohazards. As such, it feels more responsible to drip out some information, see what effects it has and then consider writing more if doing so feels robustly net-positive. That said, I might be willing to share some information with researchers early, if I feel that they have good practises around infohazards.
- ^
I've had these ideas for more than a year, but I've been too busy to write them up and uncertain whether I even should.
I eventually decided that (potential infohazard risks aside) it would be better to write them up quickly, despite the flaws this would inevitably introduce, rather than never writing anything at all.
I further decided that it would be better to write the post and then consider the infohazard risks of a specific, concrete article, rather than abstractly.
- ^
Eugene writes: "With regards to general anaesthesia it is true that parts of the brain are still processing pain stimuli which causes an increase in blood pressure and heart rate. That's why opiates are often still given during GA. But is anyone really suffering during that?"
- ^
Eugene writes: "I don't think this is true for severe pain... Sufferers of chronic pain often try to learn mindfulness techniques to direct their attention elsewhere. A key biologic function of pain is to capture attention esp if severe.."
This is a good point, though I would say that we have both a tendency to look at suffering and then a tendency to look towards it.
How does this pan out? One option would to assume both biases roughly cancel out. This wouldn't be impossible, but I don't buy it. Why? Well, well insofar as we have a tendency to look at suffering, I would say that this doesn't just bias our judgements upwards, but that acknowleding this tendency should also increase our estimates of how much suffering we will experience due to the ability of one small thing to ruin our experience even if everything else is going well.Isn't there a symmetrical argument where our tendency to look away from suffering biases our estimate downwards, but reduces how much suffering we experience? Yes, well-spotted! However, even if we don't directly suffer when we aren't looking at the pain, I suspect that a part of us knows that we are looking away from it and that this leads to suffering in and of itself.
(My accounting here is very handwavey, but this post isn't aiming to be the definite word on these matters, but merely to pushback against traditional assumptions. I'm not necessarily that satisfied with the analysis that I've shared here—but also, this is kind of a side-quest for me, so it's hard for me to dedicate the time to do this justice.Maybe the one key idea which I wish to promote here is that I don't believe we're likely to be able to do good work on the topic of suffering without engaging with the phenomenology of it, where this word is understood in a broad sense. That seems to matter more than any particular conclusion).
