susu.exp wrote:Mr.Samsa wrote:Well it sort of differs depending on your exact definition of free will but all of them as far as I'm aware demand the existence of some agent that can direct behavior by overriding external variables. So positing absolute unpredictability is just as troublesome for free will as positing determinism because it's impossible for a free agent to change behavior when the inputs and outputs are unpredictable.
But here´s the hook: The unpredictability of the output stems from the behaviour of the agent, in the case of humans minds the stochasticity of neuronal processes. If I can not predict the behaviour of a person precisely knowing their brain state and the inputs, then there is something that chooses between various possible courses of actions with predictable probabilities. This something is free will.
I don't think the last part of your paragraph directly follows from the first half. Correct any part of this which you feels misrepresents you, but essentially what you seem to be saying is:
1) Unpredictability of systems is due to the processes of the system
2) We cannot be aware of every brain state, input, internal processes, etc of that system
Therefore, the system transcends the laws of nature by actively choosing a behavior that is independent of causal variables.
In other words,
1) Unpredictability of weather systems is due to the processes that underlie the creation of weather patterns
2) We cannot be aware of every single factor that influences the weather
Therefore, the weather has free will and actively chooses whether to create tornadoes in Antarctica or droughts in the Amazon etc.
It just doesn't follow, in my opinion. We've established that physical systems in the universe are ultimately unpredictable due to quantum uncertainty and chaos theory etc, and that they are probabilistic in nature - this part is fine. So then the strongest claim we can make is that complex systems are unpredictable. That's it. There's no need, nor reason, to posit free will.
susu.exp wrote:
For the same reason we´re calling it love when it happens to humans, but pair-bonding instinct when it happens to chimps. It boils down to the difference between observing a stochastic system and being a stochastic system...
Fair enough.. At least your position is consistent, I have no argument against that. I just think that when you're basically claiming that the weather, digestive processes, billiard balls, planets, etc all have "free will", then we're no longer discussing free will.
susu.exp wrote:Mr.Samsa wrote:References for genetic influences on behavior? That's a pretty broad topic, surely you've read Tinbergen's studies on fixed-action patterns etc? Beyond that, there are probably studies that are investigating the selection process of neural networks, but I'm not sure of any off the top of my head.
I´m aware that there are genetic influences on behaviour, what I was wondering was whether there were genetic influences on mice picking left or right in a T-maze with no prior history with that maze. It would be interesting (and have implications for all T-maze experiments made so far).
I'm not sure of any detailed studies into the genetic component, but there will always be certain genetic characteristics which affects how an organism responds in a given situation. For example, pigeons tend to have a slight bias toward colours whereas rats have a bias toward geometric shape. This doesn't affect results a whole lot as it's still possible to train pigeons to attend to the shapes of a stimulus, and rats to the colour, it's just that if you present them with a compound stimulus then certain biases tend to emerge presumably as certain relations/associations are easier to form due to some innate neuronal structure.
This sort of ties into a principle called "preparedness" which tries to explain why certain relations are easier to learn than others. In particular, a phenomenon called the Garcia Effect where, despite there a consistent finding that despite most relations require numerous pairings with a relatively short time delay, organisms will learn a taste-sickness association with a single pairing and delays around 24 hours. This is assumed to be an evolved trait, but I'm not sure how that explanation could be supported or falsified, so it is somewhat speculative, I guess.
susu.exp wrote:Mr.Samsa wrote:My apologies then. I thought your example was meant to imply that mice are actually unpredictable, to such a degree that even predators are unable to develop an evolutionary strategy to help catch them. The problem with optimal strategies in real life though, is that they necessarily ignore the environmental history of the organism - whilst it may be best for the mouse to alternate directions, it's not going to do this. The probability of it choosing left or right will be shaped almost entirely by it's history with each option.
True. However a part of the strategy of the mouse is that it merely alters probabilities. There are reasons for this, the primarily one is that evolution is a stochastic process as well and mixed strategies are only stable (even if they are ESS they would otherwise be removed by drift) if the mixing occurs by the organism picking the strategy with a certain probability each.
I'm not quite sure where your explanation is meant to contradict or be inconsistent with mine above it? Except in that you say the "organism picks the strategy" whereas I'd rather leave the "picking" claim out of it, so the probabilities simply predict the behavior.
susu.exp wrote:Mr.Samsa wrote:Determine as in "shape or influence", no? Are you objecting to the word based on philosophical grounds or does it have some mathematical connotations that I'm not aware of?
Mainly I´m objecting because "determine" at least implies determinism.
In a strict sense of the word, yes. But surely the fact that external variables shape or influence our behavior is undeniable, right? Or am I missing something here?
susu.exp wrote:Mr.Samsa wrote:Put another way - introducing the variables in a behavioral experiment are exactly the same as introducing variables into a physics experiment. Whilst it's true that on some deeper metaphysical level that the sloping gradient might not "determine" the direction of a rolling ball, we can say with some certainty that even if the ball "chose freely to roll down the hill" instead of up it, the slope was at least a factor in the end result. The same can be said of behavioral experiments.
I agree, in physics that ball experiment is so close to deterministic that it makes no practical difference. The same is usually not true in behavioural experiments.
Really? What level of predictability would you need before considering behavioral experiments 'close to deterministic'? Without even considering genetics or neuroscience at all, we can predict choice of all organisms reliably between 95-100% of the time. And this is from a particular branch of science that has only been around since the 70s. Given a few more decades, and an integration of behavioral, genetic and neuroscientific theories, then I'm sure we could bump up that average a bit.