Thursday, April 16, 2009

Parsimony and Internalist Theories of Value

I have argued that internal state theories of value are flawed.

Internal state theories are theories that hold that all value is to be found in a particular internal state – happiness, pleasure, contentment, satisfaction (in the hedonist sense). However, I have expressed two concerns about internal state theories.

First, we need to have an explanation as to why an agent can be so constituted so as to be concerned about an internal state, but cannot be constituted so as to be concerned about any external states (except insofar as they are useful for realizing the internal state).

Second, if all value is found in an internal state then if a brain is in that state, nothing that is going on in the external world matters (as long as it can be prevented from dealing with that state). This is true by definition. If the internal state is the only thing that matters, then external states do not matter.

There are several stories raised as objections to internal state theories, but they all have they same structure. A part of the story is aimed at getting the brain in the required state. The rest of the story is premised on the fact that if internal states are all that matter, then external states do not matter. So, it alters the external world, then asks the question, “Are you telling me this internal state does not matter? As long as I have created the required internal state, I am done?

The story I use deals with that of a person who is given two options. Option 1: the agent will be made to feel happy by being made to believe her child is well off when, in fact, the child is being tortured. Option 2: The agent is made to feel miserable by being made to believe that the child is being tortured while the child is made better off. In this story, the false belief is used to put the mind in the right state, but is divorced from the external world.

Against this, Richard Carrier wrote:

Hence in actual psychological fact, Option 2 is only selected because people are happier being the sort of person who would choose Option 2 (and would not be happy being the sort of person who would choose Option 1). Were this not the case, they would never choose Option 2 (and I suspect this would bear out experimentally: e.g. people who do choose Option 1 will be shown upon testing to have no care what sort of person this makes them, whereas those who choose Option 2 will be shown upon testing to care very much about that).

However, according to internal state theory, our agent doesn’t care about being a particular type of person. Our agent cares about realizing a particular state. Being a particular type of person is only relevant insofar as it helps to bring about or preserve the internal state. If we can create the internal state through any method at all, that is the only thing that matters.

Another response that Carrier gives is:

people are basing their choice on the covert assumption that they will remember what they chose, even despite consciously thinking they are taking into account the condition that they won't. Like many counter-intuitive decision-situations, people might be making an error in choosing Option 2, which could only be overcome by rationally working out the options, in order to overcome the counter-intuitiveness of the choice.

I understand this to mean, “Yeah, people claim that they would prefer Option 2 over Option 1. But this is because they are making a mistake. They are ignoring the fact that they will be made to forget their choice and acting as if they will remember their choice. If people took the proposition that they would forget their choice seriously, they would in fact choose personal happiness over the well-being of the child.”

So, no parent actually cares about her children. She only cares about her own happiness and the well-being of her children is merely of instrumental value.

Here, we have two competing theories. One theory postulates that the person is reporting accurately what she would choose in such a situation. The other theory postulates that the person is not reporting accurately what she would do because we are going to introduce into our theory another entity – a mistake. By introducing the assumption of a new entity (that the first theory has no need to postulate), the internal state theorist gets the answer he wants.

However, I want to ask what reason we have to grant the internal state theorist this new entity? This new entity is just another complexity that we have no reason to add, unless we are compelled to do so by the facts. In the absence of compelling evidence, we are at liberty to throw out the postulate that this additional entity exists and to go with our original theory. We explain the fact that the agent picks Option 2 because the agent actually prefers Option 2, rather than explaining it by saying that the agent actually prefers Option 1 but picks Option 2 on the basis of a mistake.

It is always possible to invent a more complex theory that yields the same results as a less complex theory. The mere fact that a more complex theory is available does not count as an argument against the simpler theory. The person must argue that the more complex theory is necessary. In the case of the mistake hypothesis as it is used here, it is not necessary. The only person who has any reason to adopt it is because the person has adopted an internal state theory of value and needs the mistake hypothesis to make that theory conform to the way people actually behave.


Piero said...

"So, no parent actually cares about her children. She only cares about her own happiness and the well-being of her children is merely of instrumental value."I don't think this is a fair response to Carrier's objection. If I'm made to believe my child is safe and sound, I fell happy because I have no way of knowing that I'm wrong. So, in actual practice, I would choose the "wrong" alternative because I cannot know it is wrong. Tje situation is somewhat analogous to the following: I'm driving in the city, and I see a child carelessly crossing the road in pursuit of a ball. I bring the car to a sudden stop, and as a result the car following me has to manoeuver in order to avoid a crash, and the driver loses control of the car, killing seven bystanders. Was my action wrong? How could I possibly know?

Richard Carrier said...

Fyfe has completely misconstrued the remarks he is quoting: I was identifying a design flaw in his proposed experiment, not proposing a theoretical model. I was saying he needs to overcome that design flaw before the experimental results he cites can be relevant to his argument.

Parsimony has nothing to do with this.