RichardKennaway comments on The Blue-Minimizing Robot - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (159)
Well, your robot example was an intuition pump constructed so as to be as close as possible to stimulus-response nature. If you consider something only slightly more complicated the distinction may become clearer: a room thermostat. Physically ripped out of its context, you can see it as a stimulus-response device. Temperature at sensor goes above threshold --> close a switch, temperature falls below threshold --> open the switch. You can set the temperature of the sensor to anything you like, and observe the resulting behaviour of the switch. Pure S-R.
In context, though, the thermostat has the effect of keeping the room temperature constant. You can no longer set the temperature of the sensor to anything you like. Put a candle near it, and the temperature of the rest of the room will fall while the sensor remains at a constant temperature. Use a strong enough heat source or cold source, and you will be able to overwhelm the control system's efforts to maintain a constant temperature, but this fails to tell you anything about how the control system works normally. Do the analogous thing to a living organism and you either kill it or put it under such stress that whatever you observe is unlikely to tell you much about its normal operation -- and biology and psychology should be about how organisms work, not how they fail under torture.
Did you know that lab rats are normally starved until they have lost 20% of their free-feeding weight, before using them in behavioural experiments?
Here's a general block diagram of a control system. The controller is the part above the dotted line and its environment the part below (what would be called the plant in an industrial context). R = reference, P = perception, O = output, D = disturbance (everything in the environment besides O that affects the perception). I have deliberately drawn this to look symmetrical, but the contents of those two boxes makes its functioning asymmetrical. P remains close to R, but O and D need have no visible relationship at all.
When you are dealing with a living organism, R is somewhere inside it. You probably cannot measure it even if you know it exist. (E.g. just what and where, physically, is the set point for deep body temperature in a mammal? Not an easy question to answer.) You may or may not know what P is -- what the organism is actually sensing. It is important to realise that when you perform an experiment on an animal, you have no way of setting P. All you can do is create a disturbance D that may influence P. D, from a behavioural point of view, is the "stimulus" and O, the creature's action on its environment, is the "response". the behaviourist description of the situation is this:
This is simply wrong. The system does not work like that and cannot be understood like that. It may look as if D causes O, but that is like thinking that a candle put in a certain place chills the room, a fact that will seem mysterious and paradoxical when you do not know that the thermostat is present, and will only be explained by discovering the actual mechanism, discarding the second diagram in favour of the first. No amount of data collection will help until one has made that change. This is why correlations are so lamentably low in psychological experiments.
No, I've never used any of those systems. I prefer a medium in which I can take my time to work out exactly what I want to say.
Okay, we agree that the simple robot described here is behaviorist and the thermostat is PCT. And I certainly see where you're coming from with the rats being PCT because hunger only works as a motivator if you're hungry. But I do have a few questions:
There are some things behaviorism can explain pretty well that I don't know how to model in PCT. For example, consider heroin addiction. An animal can go its whole life not wanting heroin until it's exposed to some. Then suddenly heroin becomes extraordinarily motivating and it will preferentially choose shots of heroin to food, water, or almost anything else. What is the PCT explanation of that?
I'm not entirely sure which correlation studies you're talking about here; most psych studies I read are done in an RCT type design and so use p-values rather than r-values; they can easily end up with p < .001 if they get a large sample and a good hypothesis. Some social psych studies work off of correlations (eg correlation between being observer-rated attractiveness and observer-rated competence at a skill); correlations are "lamentably low" in social psychology because high level processes (like opinion formation, social interaction, etc.) have a lot of noise. Are there any PCT studies of these sorts of processes (not simple motor coordination problems) that have any higher correlation than standard models do? Any with even the same level of correlation?
What's the difference between control theory and stimulus-response in a context? For example, if we use a simplified version of hunger in which the hormone leptin is produced in response to hunger and the hormone ghrelin is produced in response to satiety, we can explain this in two ways: the body is trying to PCT itself to the perfect balance of leptin and ghrelin, or in the context of the stimulus leptin the response of eating is rewarded and in the context of the stimulus ghrelin the response of eating is punished. Are these the same theory, or are there experiments that would distinguish between them? Do you know of any?
Does PCT still need reinforcement learning to explain why animals use some strategies and not others to achieve equilibrium? For example, when a rat in a Skinner box is hungry (ie its satiety variable has deviated in the direction of hunger), and then it presses a lever and gets a food pellet and its satiety variable goes back to its reference range, would PCTists explain that as getting rewarded for pressing the lever and expect it to press the lever again next time it's hungry?
Just a brief note to say that I do intend to get back to this, but I've been largely offline since the end of last week, and will be very busy at least until the end of this month on things other than LessWrong. I would like to say a lot more about PCT here than I have in the past (here, here, and in various comments), but these things take me long periods of concentrated effort to write.
BTW, one of the things I'm busy with is PCT itself, and I'll be in Boulder, Colorado for a PCT-related meeting 28-31 July, and staying on there for a few days. Anyone around there then?