A few notes on Predictability and Determination.
We have learned over the last few decades that a process can be ‘determined’ as a physical processes but not necessarily predictable. Weather, for example, may be inherently unpredictable beyond a certain rough level of precision given the theoretical limits on computation using systems made of matter. Human behavior modeled as the interactions of humans within a given physical and cultural environment might be equally complex and hence only partially predictable.
Three models: is there no such thing as a Free Lunch?
I’ve found it useful to work through the following thought experiment to tease out the various threads of meaning that I commonly see operating in discussion of Free Will. More complex examples might ultimately be more nuanced but I think the gist of can be captured with a very simple look at three versions of lunch.
Models
The constant – lunch.
1) Model 1: Direct. Imagine a buddy suggests lunch. I suggest a place. We meet for a burrito. This is simple meaningful behavior. I have made a choice to go add nutrition and some social interaction to my day. (This simple assumption will be called into question in Part 3.)
2) Model 2: Random. Imagine I now decide to add a randomizing element to the model above. I step outside the door and roll a 4 sided dice. Each of the 4 numbers is a direction and, if it is rolled, I walk in that direction to the next intersection. If a particular direction is blocked, I reroll until I get one that’s open. I do this until I have found a lunch spot. (My neighborhood is more or less a grid with lunch in all directions so this should work. As a failsafe, I limit this to 100 dice rolls and pack an emergency backup burrito.)
A few points —
- This has made me significantly less predictable. One could build a probability map of my possible locations but my exact location could not be predicted with any further certainty. If our definition of Free Will is behavior that cannot be precisely predicted in advance, I’m free! Also, notice I could introduce random elements at any level of seriousness… from what I have for lunch to whether to have a child, say, or jump off the bridge. These choices would have a more or less significant impact on my personal life and potentially human history. Call it Schrödinger’s life planning.
- At the same time, the process is completely a material processes.
- Theoretically, I guess, one might even be able to predict the outcome of the dice rolls by modeling muscles, dice shapes, and air currents if one had the right tools and computing power (which may be, as noted above, outside the theoretical limits of computation on a material substrate.) This observation is, of course, irrelevant inside the experiment because I couldn’t make that prediction myself when assigning pathways to dice outcomes and hence exert deliberate or unconscious control of the process. At the level of my ability to shape the outcome, once I commit to the game, my movement is random.
- The important observation is that this behavior is much much less meaningful than my actions in Model 1. If I don’t pack a backup, I may not get lunch at all.
- This raises an interesting point, I think. It is possible that, in a certain sense, paths may be more meaningful the more predictable they are! Less predictability would mean a disengagement from the essential concerns of my existence or making sub-optimal choices. In Model 1, I could go for unpredictability and lunch somewhere else ditching my friend… and have a less meaningful lunch.
3) Model 3: The uber-rationalist.
Imagine:
- I have decided to attempt life extension via calorie restriction. Since I am very concerned with the calories in a meal, my aim is to get essential nutrients with a minimum of calories.
- But, because both exercise and socializing are keys to long-term mental agility, I don’t want to stop my routine of walking out for social lunches.
- To compensate, I have gone so far as to survey most restaurants in the neighborhood and have built charts of meals by restaurant rated by nutrition over calories. (This example would seem less outré to me if I hadn’t had a friend that routinely pulled out a scale and weighed parts of his lunch.)
- My friend suggests Emily’s Spaghetti Shack. I suggest a sushi place a bit further down the street. We compromise on a Vietnamese place and meet for lunch.
Now, this outcome might be totally predictable, perhaps even more so than Model 1.
However, given the objectives I’ve set for myself, I am successfully exercising non-illusory Free Will to maximize meaning in my behavior. Meaning, in this case, is bringing lunch into line with a long-term goal and program.
Conclusion.
It seems clear to me that predictability and meaningful choice are separate descriptive axes and the Free Will operates along the meaning axis. Predictability, or the lack thereof, is more or less irrelevant. In all three case, we’re dealing with embodied Free Will, i.e. ‘determined’ physical systems of particles interacting in chains of cause and effect. I don’t see a way to make that component illusory without ridiculous descriptive contortions.