Cognitive Science Notes
A Premature Word on Free Will
(Warning: Very rough, tentative thoughts. I’m still figuring this out and I’m not peeking at the official answer.)
You can only intervene on a system from the outside. To talk about causality, you have to partition the world into what is inside the system and what is outside. Otherwise, the world is just an automaton going from one state to the next.
So, why do humans think they have free will when obviously the world is just deterministically hopping states? Perhaps the answer lies in that our mind considers the rest of the world as the system and itself as outside the system. Then, it can talk about interventions on the world. Yes, the “interventions” it makes on the world are predetermined by its “state of mind” a second ago. But as far as it is concerned, different outputs it gives will lead to different states of the world.
I have to explain the feeling of “choice”. I feel like I can choose to walk to the left or to the right.
Evidence
Let’s list the evidence we have.
We notice a correlation between the thought “I have decided to turn left” and the action of turning left. (To be more precise, it’s the neural signals of turning left - you may not be able to move your legs.)
Similarly, we notice a three-part correlation between the initial thought “I don’t know whether I should go left or right”, the reasoning thoughts (“there’s icecream on the left and I like icecream more than chocolate”), and the final thought “I decide to go left”. These thoughts appear in that order.
I suspect that these thoughts we have are on the order of log statements in a program, incidental to the main execution. As the functions of the program get executed, they write some statements to the log. I suspect that we just get to hear (a bit of) what the machinery of the mind is doing. The log file - our sequence of thoughts - gets filled up as the machinery progresses.
This explains the correlations above: since the log file is updated about the decision as it is made, it will receive “decide to turn left” before your body actually turns left.
Also, initially, the log file contains the sensory inputs but an empty output because the machinery has just started. That is, the output message is “body could go left or right”. The log file then starts receiving log statements from the decision-making part of the machinery and still emptiness from the output because the result hasn’t been computed yet. Finally, the log file gets the result - your decision - because the machinery has finished cranking it.
Choices and Models
What are the “choices” that you see in your mind? Why is there a “choice” at all? How come you think you “can go left or right”? I think it’s our mind’s causal model of the world outside it. The choices are the various interventions that can be made on the system, as per the model. You will only execute one action, of course, but your model will show different actions and their consequences. Your model could be wrong, incomplete, or imprecise. Still, you will see choices in accordance with your model, not with “reality” itself.
So, I hypothesize, there is a part of your mind that uses sensory input to build and update a model of the world as best as it can with limited storage and processing power. The decision-making machinery uses both the sensory input and your current model to produce a decision, using some algorithm.
Your log file - your sequence of thoughts - receives first the sensory input (“icecream on the left, chocolate on the right”), then receives the possible interventions on the external world as per the model (“go left or right”), also the uncertainty about the reason (phrased as “should I go left or right? I don’t know yet. Let me decide.”), then the decision-making process as your mind judges the outputs of different interventions (“left means icecream: 20 utilons; right means chocolate: 15 utilons”), then the final decision (“I will go left”).
This seems reductionistic enough. It’s all (reasonably) computable stuff. You can imagine writing a very basic program that did the above. So what are the implications? What does it mean if “we” only receive messages of things that are being done by the brain and only intimated to “us” as a courtesy?
No. I think that “we” are actually the decision-making process, not the log file (the thoughts). Still not fully sure.
Changing your Mind
The biggest explanatory problem we face with “free will” is that we think we can “change our actions”. Or that we can decide to “make any choice we want”. We reject the idea that there is a single action we are going to take whether we like it or not, and that nothing we do can change it.
The idea of “no free will” says that if an all-knowing oracle told me I am going to turn left no matter what I do, then I can’t decide to go right; I will turn left. But we humans disagree with that vehemently.
What’s the problem with that headstrong belief? You could imagine an oracle that looked over the state of your mind and crunched the numbers and … (does this run into undecidability problems? TODO). Perhaps this is straightened out when you talk in terms of cognitive algorithms.
Let’s say that an oracle told you that you are going to turn left. Well, that’s actually an intervention on your mind, even if you don’t realize it. Now, given the sensory input you get, plus your cognitive algorithm, plus this above suggestion, we can predict what thoughts you will get and what your final action will be. If you learn of this further notion too, then given that fact, we can predict what your (possibly different) action will be.
So, as long as you’re not told what it is, you will end up taking a particular action. If you’re told what it is, you will still end up taking a particular action, just maybe not the same one.
Wait. What if you “think” you are going to take a particular action? What does that mean? When would you get such a “thought”? Your mind modeled itself and ran the model with a particular input and predicted that it would get a particular output. (TODO Question: If it can model itself, is it an algorithm? Will it become undecidable somehow? I’m not familiar enough with Automata theory to answer that.)
Still, the original retort was that an external mind cannot predict my mind’s decision. It can, but if it tells you, then it may turn out to be wrong. (It may turn out to be right too - my friend can tell me how I will react to a new movie and sometimes I will react that way. Sheldon: “I’m not nearly as mysterious as I think I am.”)
TODO: summary.
The Core
The key idea of free will is that you can “choose” any of your “possible” actions you “want”. You can even change it on the fly if someone asks you to do X instead of Y.
Related is the idea of responsibility: should “you” take responsibility for your actions? Isn’t it all predetermined? Why bother “doing” anything in particular? What is going to happen is going to happen and all.
Resources
Textbooks
Cognition in Action - approachable book about how cognitive science applies to everyday life.
Cognition by Daniel Reisberg - decent textbook
Also, look at the Handbook of Categorization in Cognitive Science.
At the MIT Encyclopedia of Cognitive Sciences - Categorization; Concepts; Scientific Thinking. (links are broken; google them.)
Try to get Formal Approaches to Categorization.
Check out Stevan Harnad (link)
Cognitive Psychology: a student’s handbook
Books on Cognitive Psychology for Learning
How we Learn by Benedict Carey
Making up the Mind by Chris Firth
Why Don’t Students like School by Daniel T. Willingham
Iconoclast by Gregory Burns (neuroscience approach)
Make it Stick: The Science of Successful Learning
comments powered by Disqus